Deep Learning Approaches for Detecting Freezing of Gait in Parkinson’s Disease Patients through On-Body Acceleration Sensors
Open accessAbstract
Freezing of gait (FOG) is one of the most incapacitating motor symptoms in Parkinson’s disease (PD). The occurrence of FOG reduces the patients’ quality of live and leads to falls. FOG assessment has usually been made through questionnaires, however, this method can be subjective and could not provide an accurate representation of the severity of this symptom. The use of sensor-based systems can provide accurate and objective information to track the symptoms’ evolution to optimize PD management and treatments. Several authors have proposed specific methods based on wearables and the analysis of inertial signals to detect FOG in laboratory conditions, however, its performance is usually lower when being used at patients’ homes. This study presents a new approach based on a recurrent neural network (RNN) and a single waist-worn triaxial accelerometer to enhance the FOG detection performance to be used in real home-environments. Also, several machine and deep learning approaches for FOG detection are evaluated using a leave-one-subject-out (LOSO) cross-validation. Results show that modeling spectral information of adjacent windows through an RNN can bring a significant improvement in the performance of FOG detection without increasing the length of the analysis window (required to using it as a cue-system).
AI snapshot
Written from the paper’s full text.
TL;DR
This study evaluates machine and deep learning classifiers for detecting freezing of gait (FOG) in Parkinson’s disease using a single waist-worn triaxial accelerometer in home environments. A CNN-LSTM model using FFT spectral features from the current and three previous 3.2-s windows achieved the best reported LOSO AUC of 0.939, improving on a Mazilu-feature Random Forest baseline (AUC 0.904).
Key findings
- The best model was a CNN-LSTM using 64-bin FFT features with the current and three previous spectral windows and 75% overlap, evaluated with LOSO: AUC 0.939, EER 12.9%, and sensitivity/specificity 87.1% (Section 5.5 also reports AUC 0.936 and EER 12.7% for this comparison).
- The reproduced baseline using Mazilu features and Random Forest under LOSO achieved sensitivity 0.826, specificity 0.825, GM 82.5%, and AUC 0.904; the CNN-LSTM improved AUC and reduced EER.
- Supervised methods outperformed novelty detection; CNN-LSTM was best, while Random Forest and CNN-MLP had similar AUCs (0.921 and 0.920), and SVM was weakest (AUC 0.876).
- Data came from 21 PD patients (3 women, 18 men; mean age 69) at home, with a FOG-to-non-FOG ratio of about 10:1 (10.5% FOG windows).
Methods
- Used the Rodríguez-Martín et al. dataset: single waist-worn IMU, home ADL recordings, 200 Hz signals resampled to 40 Hz, filtered, and windowed into 128-sample (3.2 s) windows with 75% overlap; a window was labeled FOG only if >50% of its samples were FOG.
- Compared data representations including Mazilu features, MFCCs adapted to inertial signals, 64-bin FFT, and contextual FFT stacking up to three previous windows; classifiers included OC-SVM, denoiser autoencoder, Random Forest, AdaBoost, SVM, CNN, and CNN-LSTM.
- Main evaluation used leave-one-subject-out (LOSO) cross-validation repeated six times over 21 patients, with sensitivity, specificity, AUC, GM, and EER as metrics.
Limitations
- Stated: deep recurrent models require large computational resources and large accurately labeled datasets; using more than three previous windows can be impractical for real-time applications; comparison with related work is difficult due to diverse approaches and validation methods; imbalanced FOG data present an additional challenge.
- Stated: some patients (2, 4, and 20) showed different FOG patterns that reduced LOSO performance, and patient 1 had sensitivity/specificity below 0.804.
- Evident from the text: the study relies on a single 21-patient home-monitoring dataset without external validation, and the reported best AUC/EER differ slightly across sections (0.939/12.9% vs. 0.936/12.7%).
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
0 of 40 cited sources checked · 0 supported · 0 flagged
- Findingown data
Modeling spectral information of adjacent windows through an RNN can bring a significant improvement in the performance of FOG detection without increasing the length of the analysis window.
“Results show that modeling spectral information of adjacent windows through an RNN can bring a significant improvement in the performance of FOG detection without increasing the length of the analysis window (required to using it as a cue-system).”
Quote in the paperSubstantial claim - Findingnone given
The best result achieved an improvement from 0.904 to 0.939 of the area under the curve (AUC) with a LOSO evaluation.
“The best result achieved shows an improvement from 0.904 to 0.939 of the area under the curve (AUC) with a LOSO evaluation.”
Quote in the paperSubstantial claim - Findingown data
The shallow models present lower performances than the deep learning classifiers.
“In this experiment, the shallow models present lower performances than the deep learning classifiers.”
Quote in the paperSubstantial claim - Findingown data
When using previous windows through the CNN-LSTM model, there is a progressive increase in performance when adding more previous spectral windows.
“According to the results shown in Table 7, when using previous windows through the CNN-LSTM model, there is a progressive increase in the performance when adding more previous spectral windows.”
Quote in the paperSubstantial claim - Findingown data
AUC increased from 0.923 without previous windows to 0.939 with three previous windows.
“The results are improved with a reduction in the EER from 15.1% (with no previous windows) to 12.99% with tree previous windows, while the AUC is increased from 0.923 (without previous windows) to 0.939 with three previous windows.”
Quote in the paperSubstantial claim
- Findingown data
The performance increases when using overlapped windows in the three data representations and the two evaluation methodologies.
“According to Table 3 and Table 4, the performance increases when using overlapped windows in the three data representations and the two evaluation methodologies.”
Quote in the paperSubstantial claim - Findingown data
The average result presents a high AUC performance for a generic model by using LOSO evaluation.
“However, the average result presents a high AUC performance for a generic model by using LOSO evaluation.”
Quote in the paperSubstantial claim - Findingown data
AdaBoost with 100 estimators and SVM classifiers presented the weakest performance among supervised methods in the experiments.
“AdaBoost with 100 estimators and SVM classifiers present the weakest performance in the experiments for supervised methods.”
Quote in the paperNot a substantial claim - Findingown data
The SVM classifier presents the lower results (AUC 0.876) using a radial basis function kernel.
“SVM classifier presents the lower results (AUC 0.876) using a radial basis function kernel.”
Quote in the paperNot a substantial claim - Findingown data
The addition of previous spectral windows gradually improves the discriminative power of the deep network.
“According to Table 7, the addition of previous spectral windows gradually improves the discriminative power of the deep network.”
Quote in the paperSubstantial claim - Findingown data
Among supervised methods, the CNN-LSTM deep network presented the best performance, with sensitivity and specificity of 0.849.
“In regard to supervised methods, Table 6 shows that the CNN-LSTM deep network presents the best performance with 0.849 in sensitivity and specificity.”
Quote in the paperSubstantial claim - Findingreasoning
The results obtained with LOSO evaluation are lower than the results obtained through R10fold.
“According to Table 4, the results obtained with LOSO evaluation are lower than the results obtained through R10fold (Table 3).”
Quote in the paperNot a substantial claim - Findingown data
Random forest and CNN-MLP showed similar AUC results, 0.921 and 0.920 respectively.
“Random forest and CNN-MLP present a slightly lower performance to that of CNN-LSTM, but they show similar AUC results (0.921 and 0.920 respectively).”
Quote in the paperNot a substantial claim - Findingown data
Supervised methods presented better performance than novelty detection.
“According to Table 5 and Table 6, the supervised methods present a better performance than novelty detection.”
- Also stated in: § 5.5. Comparison of the Base Line and the CNN-LSTM · boosted
“In this study, the supervised methods have overperformed novelty detection.”
Quote in the paperSubstantial claim - Also stated in: § 5.5. Comparison of the Base Line and the CNN-LSTM · boosted
- Findingown data
EER was reduced from 15.1% with no previous windows to 12.99% with three previous windows.
“The results are improved with a reduction in the EER from 15.1% (with no previous windows) to 12.99% with tree previous windows, while the AUC is increased from 0.923 (without previous windows) to 0.939 with three previous windows.”
Quote in the paperSubstantial claim - Findingown data
The best results in R10fold evaluation were achieved by using Mazilu representation with a 75% overlap.
“The best results were achieved by using Mazilu representation with a 75% overlap.”
Quote in the paperNot a substantial claim - Findingown data
The three data representations (Mazilu, MFCC, FFT) present similar performances when using the same overlap in R10fold evaluation.
“The three-data representation present similar performances when using the same overlap.”
Quote in the paperNot a substantial claim - Findingown data
The best results in this study with LOSO evaluation were obtained with the CNN-LSTM model when using the current and three previous spectral windows.
“The best results in this study with LOSO evaluation were obtained with the CNN-LSTM model when using the current and three previous spectral windows.”
Quote in the paperSubstantial claim - Findingreasoning
Sensitivity, specificity, and AUC increase progressively as the percentage of overlap increases in all data representations tested in R10fold evaluation.
“According to Table 3, sensitivity, specificity, and the AUC increase progressively as the percentage of overlap increases in all data representations tested.”
Quote in the paperNot a substantial claim - Findingown data
The MLP denoiser auto encoder presents a better performance over the OC-SVM in novelty detection.
“According to Table 5, the MLP denoiser auto encoder presents a better performance over the OC-SVM.”
Quote in the paperNot a substantial claim- The authors state this as an assumption: In prediction, patterns with abnormal movement (with significant statistical differences from normal patterns) are FOG events.
- Findingown data
The results show a high performance in FOG detection in the majority of the patients.
“The results show a high performance in FOG detection in the majority of the patients.”
Quote in the paperSubstantial claim - Findingnone given
Most results improve when using overlapped windows in LOSO evaluation.
“Most results improve when using overlapped windows.”
Quote in the paperNot a substantial claim - Findingown data
The mean sensitivity and specificity is 0.871 by using an equal error threshold for the CNN-LSTM with three previous windows.
“According to Table 8, the mean sensitivity and specificity is 0.871 by using an equal error threshold instead of maximizing the GM between sensitivity and specificity.”
Quote in the paperNot a substantial claim - Findingown data
Random forest and CNN-MLP presented slightly lower performance than CNN-LSTM.
“Random forest and CNN-MLP present a slightly lower performance to that of CNN-LSTM, but they show similar AUC results (0.921 and 0.920 respectively).”
Quote in the paperNot a substantial claim - Findingown data
The performance of the novelty detection methods is lower than the results achieved with the supervised methods.
“Nevertheless, the performance of the novelty detection method is lower than the results achieved with the supervised methods shown in Table 4.”
Quote in the paperNot a substantial claim- The authors state this as an assumption: In prediction, patterns with abnormal movement (with significant statistical differences from normal patterns) are FOG events.
- Findingown data
The best results for a generalized model (LOSO) were achieved by using a 64-bin FFT.
“The best results for a generalized model were achieved by using a 64-bin FFT.”
Quote in the paperNot a substantial claim - Findingcitation
The MLP Denoiser Autoencoder achieved higher sensitivity and specificity than the OC-SVM in the novelty detection comparison.
“An improvement in the sensitivity and specificity is achieved using MLP Denoiser Autoencoder over the OC-SVM.”
Quote in the paperSubstantial claim - Findingown data
The CONV-LSTM deep network presents the best performance compared to other supervised methods.
“According to Table 6, the CONV-LSTM deep network presents the best performance compared to other supervised methods.”
Quote in the paperSubstantial claim - Findingown data
In both supervised and novelty detection methods, deep learning approaches presented improvement over shallow machine learning approaches.
“In both supervised and novelty detection methods, the deep learning approaches present improvement over the shallow machine learning approaches.”
- Also stated in: § 5.5. Comparison of the Base Line and the CNN-LSTM · plain
“DL approaches present a better performance than shallow algorithms in all of the present study’s experiments.”
Quote in the paperSubstantial claim - Also stated in: § 5.5. Comparison of the Base Line and the CNN-LSTM · plain
- Findingown data
Spectral data representations, such as MFFCs and FFT, show high performance in generalized models evaluated through LOSO, over hand-made features.
“Spectral data representations (e.g., MFFCs and FFT) show a high performance in generalized models evaluated through LOSO, over the hand-made features.”
Quote in the paperSubstantial claim - Findingown data
CNN-LSTM with 64-bin FFT and three previous windows increased sensitivity from 0.826 to 0.871 compared to the baseline.
“apart from sensitivity and a specificity increase from 0.826 to 0.871 and 0.825 to 0871, respectively.”
Quote in the paperSubstantial claim - Findingown data
Per patient analysis results in Table 8 show AUC values exceeding 0.915, except for patients 2, 4, and 20.
“With regard to the per patient analysis, the results in Table 8 show AUC values that exceed 0.915, except for the patients 2, 4, and 20.”
Quote in the paperNot a substantial claim - Findingown data
The method is capable of working with signals from a single sensor.
“Also, the results show that this method is capable of working with signals from a single sensor.”
Quote in the paperSubstantial claim - Findingown data
CNN-LSTM with 64-bin FFT and three previous windows improved AUC from 0.904 to 0.936 compared to the baseline of Mazilu features with Random Forest under LOSO.
“one may observe an improvement in AUC from 0.904 to 0.936”
Quote in the paperSubstantial claim - Findingown data
CNN-LSTM with 64-bin FFT and three previous windows increased specificity from 0.825 to 0.871 compared to the baseline.
“apart from sensitivity and a specificity increase from 0.826 to 0.871 and 0.825 to 0871, respectively.”
Quote in the paperSubstantial claim - Findingown data
Using LOSO evaluation, the CONV-LSTM model with three FFT previous windows achieved an AUC value of 0.939 and a sensitivity and specificity of 87.1%.
“Using LOSO evaluation, the CONV-LSTM model with three FFT previous windows achieved an AUC value of 0.939 and a sensitivity and specificity of 87.1%, the highest values of AUC in FOG detection using a single triaxial accelerometer, to the authors’ knowledge.”
Quote in the paperSubstantial claim - Findingown data
Only patient 1 presented a sensitivity and specificity below 0.804.
“Despite decreased performance in these patients, only patient 1 presents a sensitivity and specificity bellow 0.804.”
Quote in the paperNot a substantial claim - Study design and conductown data
The results in the FOG-Q were a mean index of 15.2 (standard deviation: 4.47) with a range from 6 to 23.
“The results in the FOG-Q were a mean index of 15.2 (standard deviation: 4.47) with a range from 6 to 23.”
Quote in the paperNot a substantial claim - Interpretationreasoning
The dataset can be considered as a binary unbalanced classification problem.
“According to this ratio, the dataset can be considered as a binary unbalanced classification problem.”
Quote in the paperSubstantial claim - Premisecitation
A sampling rate of 40 Hz is enough to analyze human movements.
“For this study, the raw accelerometer signals have been resampled at 40 Hz since this frequency has been shown to be enough to analyze human movements according to Zhou [73].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [73] Zhou, 2008, Human motion tracking for rehabilitation-A survey, Biomed. Signal. Process. Control. source
Not checked in the free run.
- Premisecitation
The data are composed of acceleration signals from 21 patients (3 women and 18 men) with FOG symptoms and a mean age of 69 years.
“The data are composed of acceleration signals from 21 patients (3 women and 18 men) with FOG symptoms and a mean age of 69 years.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [59] Rodríguez-Martín, D., Samà, A., Pérez-López, C., Català, A., Moreno Arostegui, J.M., Cabestany, J., Bayés, À., Alcaine, S., Mestre, B., and Prats, A. (2017). Home detection of freezing of gait using support vector machines through a single waist-worn triaxial accelerometer. PLoS ONE, 12. source
Not checked in the free run.
- Premisecitation
Some common characteristics associated with FOG events in inertial signals and spectral information can be used for automatic detection.
“However, in terms of inertial signals and its spectral information, some common characteristics associated with FOG events can be used for automatic detection [64].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [64] Alcaine, 2018, Determining the optimal features in freezing of gait detection through a single waist accelerometer in home environments, Pattern Recog. Lett. source
Not checked in the free run.
- Premisecitation
Normal gait steps have a main frequency from 0.5 to 3 Hz.
“While normal gait steps have a main frequency from 0.5 to 3 Hz (determined with sensors placed on the ankles), FOG events present a main frequency in the range from 6 to 8 Hz [14,15,16].”
Quote in the paperSubstantial claim0 of 3 sources checked- cites [14] Moore, 2008, Ambulatory monitoring of freezing of gait in Parkinson’s disease, J. Neurosci. Methods source
Not checked in the free run.
- cites [15] Ahlrichs, 2016, Detecting freezing of gait with a tri-axial accelerometer in Parkinson’s disease patients, Med. Biol. Eng. Comput source
Not checked in the free run.
- cites [16] Kim, H., Lee, H.J., Lee, W., Kwon, S., Kim, S.K., Jeon, H.S., Park, H., Shin, C.W., Yi, W.J., and Jeon, B.S. (2015, January 25–29). Unconstrained detection of freezing of Gait in Parkinson’s disease patients using smartphone. Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBS), Milano, Italy.
Not checked in the free run.
- Premisecitation
FOG events present a main frequency in the range from 6 to 8 Hz.
“While normal gait steps have a main frequency from 0.5 to 3 Hz (determined with sensors placed on the ankles), FOG events present a main frequency in the range from 6 to 8 Hz [14,15,16].”
Quote in the paperNot a substantial claim0 of 3 sources checked- cites [14] Moore, 2008, Ambulatory monitoring of freezing of gait in Parkinson’s disease, J. Neurosci. Methods source
Not checked in the free run.
- cites [15] Ahlrichs, 2016, Detecting freezing of gait with a tri-axial accelerometer in Parkinson’s disease patients, Med. Biol. Eng. Comput source
Not checked in the free run.
- cites [16] Kim, H., Lee, H.J., Lee, W., Kwon, S., Kim, S.K., Jeon, H.S., Park, H., Shin, C.W., Yi, W.J., and Jeon, B.S. (2015, January 25–29). Unconstrained detection of freezing of Gait in Parkinson’s disease patients using smartphone. Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBS), Milano, Italy.
Not checked in the free run.
- Premisecitation
A sampling rate of 40 Hz is enough to analyze the frequency spectrum (below 20 Hz) during FOG episodes.
“Furthermore, the use of 40 Hz of sampling rate is enough to analyze the frequency spectrum (below 20 Hz) during FOG episodes according to Moore [14].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [14] Moore, 2008, Ambulatory monitoring of freezing of gait in Parkinson’s disease, J. Neurosci. Methods source
Not checked in the free run.
- Premisecitation
When using a short window size, there is an important decrease in the sensitivity at short episodes, which are normally the most frequent in patients with FOG.
“Whereas, when using a short window size, it has been reported an important decrease in the sensitivity at short episodes, which are normally the most frequent in patients with FOG [61,68].”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [61] Moore, S., Yungher, D., Morris, T., Dilda, V., Macdougall, H., Shine, J., Naismith, S., and Lewis, S. (2013). Autonomous identification of freezing of gait in Parkinson’s disease from lower-body segmental accelerometry. J. Neuroeng Rehabil, 10. source
Not checked in the free run.
- cites [68] Plotnik, 2010, Wearable Assistant for Parkinson’s Disease Patients With the Freezing of Gait Symptom, IEEE Trans. Inf. Technol. Biomed. source
Not checked in the free run.
- Premisereasoning
For making a usable FOG detection system, event detection must be fast without adding delays to the detection process.
“Therefore, for making a usable FOG detection system it is necessary that event detection must be fast without adding delays to the detection process [67,74].”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [67] San-Segundo, R., Torres-Sánchez, R., Hodgins, J., and Fernando De La, T. (2019). Increasing Robustness in the Detection of Freezing of Gait in Parkinson’s Disease. Electronics, 8. source
Not checked in the free run.
- cites [74] Sweeney, D., Quinlan, L.R., Browne, P., Richardson, M., Meskell, P., and Ólaighin, G. (2019). A technological review of wearable cueing devices addressing freezing of gait in Parkinson’s disease. Sensors, 19. source
Not checked in the free run.
- Interpretationown data
Using spectral features from previous windows and a classifier based on CNN and LSTM layers it is possible to enhance the performance of the FOG detection system in terms of accuracy while maintaining a low-latency time.
“Results show that using spectral features from previous windows and a classifier based on CNN and LSTM layers it is possible to enhance the performance of the FOG detection system in terms of accuracy while maintaining a low-latency time.”
Quote in the paperSubstantial claim - Interpretationreasoning
LOSO methodology allows the development of generalized detection systems for using on new patients with data not included in the study.
“LOSO methodology allows the development of generalized detection systems for using on new patients with data not included in the study.”
Quote in the paperSubstantial claim - Interpretationown data
Using a higher amount of data in training can improve the generalization capacity of the network, increasing performance in classification of data not used previously in training.
“According to the results, the use of a higher amount of data in training can improve the generalization capacity of the network, increasing the performance in the classification of data that has not been used previously in training.”
- Also stated in: § 5.5. Comparison of the Base Line and the CNN-LSTM · plain
“A higher amount of data obtained from a high overlap setting have benefited the DL approaches, this amount of data has allowed the models to have a better generalization capacity to classify data from unseen patients.”
Quote in the paperSubstantial claim - Also stated in: § 5.5. Comparison of the Base Line and the CNN-LSTM · plain
- Premisecitation
OC-SVM achieved lower performance in high-dimensional datasets.
“This behavior was analyzed in [87], where OC-SVM achieved a lower performance in high-dimensional datasets.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [87] Erfani, 2016, High-dimensional and large-scale anomaly detection using a linear one-class SVM with deep learning, Pattern Recognit source
Not checked in the free run.
- Interpretationreasoning
The reduction of training data in novelty detection, by ignoring FOG events, affects performance by reducing the generalization capacity of the algorithm.
“This data reduction affects performance, reducing the generalization capacity of the algorithm.”
Quote in the paperSubstantial claim - Premisecitation
FOG events present temporal phases that appear at the beginning of the FOG event.
“considering that FOG events present temporal phases that appear at the beginning of the FOG event [82,83]”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [82] Nieuwboer, A., Dom, R., De Weerdt, W., Desloovere, K., Fieuws, S., and Broens-Kaucsik, E. (2001). Abnormalities of the spatiotemporal characteristics of gait at the onset of freezing in Parkinson’s disease. Mov Disord. source
Not checked in the free run.
- cites [83] Yungher, D.A., Morris, T.R., Dilda, V., Shine, J.M., Naismith, S.L., Lewis, S.J.G., and Moore, S.T. (2014). Temporal Characteristics of High-Frequency Lower-Limb Oscillation during Freezing of Gait in Parkinson’s Disease. Parkinsons Dis., 606427. source
Not checked in the free run.
- Interpretationown data
Wrapped CNN layers are able to identify different FOG patterns with high accuracy, even with a reduced number of samples.
“The result shows that wrapped CNN layers are able to identify different FOG patterns with high accuracy, even with a reduced number of samples.”
Quote in the paperSubstantial claim - Interpretationreasoning
The decrease in performance of novelty detection is due to the fact that data corresponding to FOG events are not considered in training.
“The decrease in performance is due to fact that the data corresponding to FOG events are not considered in training and it is assumed that FOG events occur (but may not necessarily) as a deviation from a normal pattern.”
Quote in the paperSubstantial claim- The authors state this as an assumption: FOG events occur, but may not necessarily, as a deviation from a normal pattern.
- Interpretationown data
The CONV-LSTM model with three FFT previous windows achieved the highest values of AUC in FOG detection using a single triaxial accelerometer, to the authors' knowledge.
“Using LOSO evaluation, the CONV-LSTM model with three FFT previous windows achieved an AUC value of 0.939 and a sensitivity and specificity of 87.1%, the highest values of AUC in FOG detection using a single triaxial accelerometer, to the authors’ knowledge.”
Quote in the paperSubstantial claim - Interpretationreasoning
The use of FFT spectral representation shows high generalization capability to detect spectral features related to gait patterns that can be beneficial to train generalized models.
“In this study, the use of FFT spectral representation shows high generalization-capability to detect spectral features related to gait patterns that can be beneficial to train generalized models.”
Quote in the paperSubstantial claim - Interpretationown data
Different FOG patterns between subjects present a challenge for making a system that maintains high performance.
“The different FOG patterns between subjects present a challenge for making a system that maintains high performances.”
Quote in the paperSubstantial claim - Interpretationown data
Using temporal information (previous windows) modeled through LSTM layers shows further improvement in FOG detection without increasing the length of the window.
“The results suggest that using temporal information (previous windows) modeled through LSTM layers shows a further improvement in FOG detection without increasing the length of the window.”
Quote in the paperSubstantial claim - Interpretationown data
The use of spectral representations with previous windows shows an accurate trade between performance on several patients.
“In this study, the use of spectral representations with previous windows shows an accurate trade between performance on several patients.”
Quote in the paperSubstantial claim - Interpretationown data
To the authors' knowledge, these are the highest values achieved for AUC, sensitivity, and specificity when using only a triaxial accelerometer located in the waist through LOSO evaluation (generic model) in this data set, which considers a real environment with ADL instead of laboratory conditions.
“To the authors’ knowledge, these are the highest values achieved for AUC, sensitivity, and specificity when using only a triaxial accelerometer located in the waist through the LOSO evaluation (generic model) in this data set, which considers a real environment with ADL instead of laboratory conditions.”
Quote in the paperSubstantial claim - Interpretationreasoning
The proposed CONV-LSTM architecture is a suitable method for FOG detection, considering previous temporal windows that provide temporal information about the initial phases before FOG events.
“The proposed CONV-LSTM architecture shows a suitable method for FOG detection, considering previous temporal windows that provide temporal information about the initial phases before FOG events.”
Quote in the paperSubstantial claim - Interpretationcitation
Novelty detection is a feasible method for fast detection of abnormal gait patterns in related applications.
“Despite these results, the use of novelty detection presents a feasible method for the fast detection of abnormal gait patterns in related applications [66].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [66] Mohammadian Rad, N., van Laarhoven, T., Furlanello, C., and Marchiori, E. (2018). Novelty Detection using Deep Normative Modeling for IMU-Based Abnormal Movement Monitoring in Parkinson’s Disease and Autism Spectrum Disorders. Sensors, 18. source
Not checked in the free run.
- The authors state this as an assumption: FOG events occur, but may not necessarily, as a deviation from a normal pattern.
- May combine two claims. 26
- The quote only partly states this claim. 1
- Classified as probably not a substantial claim. 22
May combine two claims.
Findingown dataThe ratio between FOG and normal ADL movement is 10 to 1, where 89.5% corresponds to the absence of FOG symptoms and 10.5% correspond to FOG events.
“The ratio between FOG and normal ADL movement is 10 to 1, where the 89.5% corresponds to the absence of FOG symptoms and the remaining 10.5% correspond to FOG events.”
Quote in the paperSubstantial claimMay combine two claims. The quote only partly states this claim.
Findingown dataThe AUC reaches a maximum value of 0.939 while the equal error rate is reduced to 12.9% for the CNN-LSTM with three previous windows.
“The AUC reaches a maximum value of 0.939 while the equal error rate is reduced to 12.9%.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataThe best data representation for a personalized model (R10fold) was Mazilu, however, MFCCS and FFT representations present similar performances for sensitivity and specificity.
“The three data representations tested in the first experiment (Section 4.2) show that the best data representation for a personalized model (R10fold evaluation) was Mazilu [44], however, the MFCCS and FFT representations present similar performances for sensitivity and specificity.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [44] Mazilu, S., Hardegger, M., Zhu, Z., Roggen, D., Tröster, G., Plotnik, M., and Hausdorff, J. (2012, January 21–24). Online Detection of Freezing of Gait with Smartphones and Machine Learning Techniques. Proceedings of the 6th International Conference on Pervasive Computing Technologies for Healthcare, San Diego, CA, USA. source
Not checked in the free run.
May combine two claims.
FindingreasoningProgressive improvement is presented in the LOSO evaluation, while in R10fold a significant improvement is presented when using higher overlap values, because of the random data sharing between the train and test subsets.
“Progressive improvement is presented in the LOSO evaluation, while in R10fold a significant improvement is presented when using higher overlap values, because of the random data sharing between the train and test subsets.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataFor a generic model with LOSO evaluation, the best results were achieved with the 64-bin FFT data representation extracted from 128 samples window, while MFCCs features show a significant improvement for generic models over the Mazilu features when using overlapped windows.
“For a generic model with LOSO evaluation, the best results were achieved with the 64-bin FFT data representation extracted from 128 samples window, while MFCCs features show a significant improvement for generic models over the Mazuli features when using overlapped windows.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataThe best model proposed in this study achieved an AUC of 0.939 and an ERR of 12.9%.
“In the best model proposed in this study, an AUC of 0.939 and an ERR of 12.9% were achieved.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataIn most patients, the AUC value reaches 0.915 except for patients 2 and 20, while the EER remains below 15% except in patients 2, 4, and 20.
“In most patients, the AUC value reaches 0.915 except for the patients 2 and 20, while the EER remains below 15% except in patients 2, 4, and 20.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataRandom forest presented an improvement in sensitivity and specificity, as well as a lower EER.
“Notwithstanding, the Random forest presents an improvement in sensitivity and specificity, as well as a lower EER.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataCNN-LSTM with 64-bin FFT and three previous windows decreased EER by 4.8%, from 17.5% to 12.7%, compared to the baseline.
“a 4.8% decrease in EER (from 17.5% to 12.7%)”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Study design and conductcitationPatients were included if they scored greater than 2 in the OFF state according to the Hoehn and Yahr scale.
“Criteria for the inclusion of patients were: a score higher than 6 on the FOG Questionnaire (FoG-Q) [23], a score greater than 2 in the OFF state according to the Hoehn and Yahr scale [72].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [72] Hoehn, 1967, Parkinsonism: Onset, Progression and Mortality, Neurology source
Not checked in the free run.
Classified as probably not a substantial claim.
PremisecitationAll data collection was carried out at the patients' homes.
“All data collection was carried out at the patients’ homes.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [59] Rodríguez-Martín, D., Samà, A., Pérez-López, C., Català, A., Moreno Arostegui, J.M., Cabestany, J., Bayés, À., Alcaine, S., Mestre, B., and Prats, A. (2017). Home detection of freezing of gait using support vector machines through a single waist-worn triaxial accelerometer. PLoS ONE, 12. source
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationWith this approach of free-living, an increase in the time and frequency of the FOG episodes is expected.
“With this approach of free-living, an increase in the time and frequency of the FOG episodes is expected [25,71].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [25] Nieuwboer, 1998, A frequency and correlation analysis of motor deficits in Parkinson patients, Disabil. Rehabil. source
Not checked in the free run.
- cites [71] Cabestany, 1992, Assessing Motor Fluctuations in Parkinson’s Disease Patients Based on a Single Inertial Sensor; Accuracy of clinical diagnosis of idiopathic Parkinson’s disease: A clinico-pathological study of 100 cases, J. Neurol. Neurosurg Psychiatry
Not checked in the free run.
Classified as probably not a substantial claim.
Study design and conductcitationPatients were included if they scored higher than 6 on the FOG Questionnaire (FoG-Q).
“Criteria for the inclusion of patients were: a score higher than 6 on the FOG Questionnaire (FoG-Q) [23], a score greater than 2 in the OFF state according to the Hoehn and Yahr scale [72].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [23] Giladi, 2009, Validation of the freezing of gait questionnaire in patients with Parkinson’s disease, Mov. Disord. source
Not checked in the free run.
May combine two claims.
Interpretationown dataThe best method is able to outperform the state-of-the-art on FOG detection by using a single triaxial accelerometer and without adding delay or increasing the size of the temporal analysis window.
“The results show that the best method is able to outperform the state-of-the-art on FOG detection by using a single triaxial accelerometer and without adding delay or increasing the size of the temporal analysis window.”
Quote in the paperSubstantial claimMay combine two claims.
PremisecitationThe data were collected through a wearable IMU located at the waist, in both ON and OFF states.
“The data were collected through a wearable IMU located at the waist, in both ON and OFF states.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [59] Rodríguez-Martín, D., Samà, A., Pérez-López, C., Català, A., Moreno Arostegui, J.M., Cabestany, J., Bayés, À., Alcaine, S., Mestre, B., and Prats, A. (2017). Home detection of freezing of gait using support vector machines through a single waist-worn triaxial accelerometer. PLoS ONE, 12. source
Not checked in the free run.
Classified as probably not a substantial claim.
Study design and conductnone givenThe data were labeled by an experienced clinician using video recordings of the tests.
“The data were labeled by an experienced clinician using video recordings of the tests.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductown dataThe mean Hoehn and Yahr scoring in OFF state was 3.07 (standard deviation: 0.426).
“While the mean Hoehn and Yahr scoring in OFF state was 3.07 (standard deviation: 0.426).”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationThe data used for testing the proposed methods in FOG detection were collected for the article [59].
“The data used for testing the proposed methods in FOG detection were collected for the article [59].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [59] Rodríguez-Martín, D., Samà, A., Pérez-López, C., Català, A., Moreno Arostegui, J.M., Cabestany, J., Bayés, À., Alcaine, S., Mestre, B., and Prats, A. (2017). Home detection of freezing of gait using support vector machines through a single waist-worn triaxial accelerometer. PLoS ONE, 12. source
Not checked in the free run.
Classified as probably not a substantial claim.
Study design and conductreasoningWindows containing FOG samples with less than 50% were discarded.
“Windows containing FOG samples with less than 50% were discarded.”
Quote in the paperSubstantial claimMay combine two claims.
InterpretationreasoningThe LOSO methodology shows more rigorous and accurate results when working with overlapped windows, avoiding data sharing between the train and test subsets.
“The results achieved with the LOSO evaluation show a decrease in the performance over the R10Fold, however, the LOSO methodology shows more rigorous and accurate results when working with overlapped windows, avoiding data sharing between the train and test subsets.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S2 (checkable).
PremisereasoningA more precise evaluation can be made by using a LOSO cross-validation.
“In order to improve the evaluation of R10fold evaluation with overlapped windows, a more precise evaluation can be made by using a LOSO cross-validation.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
Interpretationnone givenThe addition of new data to train personalized models can be difficult, because of the inherent time and resource consumption to collect and label new data.
“The addition of new data to train personalized models can be difficult, because of the inherent time and resource consumption to collect and label new data.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
InterpretationcitationKeeping a reduced batch size can help to achieve better performance on generic models.
“Therefore, keeping a reduced batch size can help to achieve better performance on generic models.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductcitationMost of the experiments in this study were carried out with a LOSO cross-validation.
“Most of the experiments in this study were carried out with a LOSO cross-validation.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductnone givenAll the experiments made in this subsection were evaluated through LOSO evaluation.
“All the experiments made in this subsection were evaluated through LOSO evaluation.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim. May combine two claims.
Study design and conductnone givenThe LOSO evaluation was made for each of the 21 patients, then the entire process has been repeated six times to verify the variability.
“For this work, the LOSO evaluation was made for each of the 21 patients, then the entire process has been repeated six times to verify the variability.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Study design and conductcitationA window was labeled as a FOG only if more than 50% of its samples were labeled as FOG.
“To address the reduced amount of FOG events in the data and to achieve a windowing strategy similar to that reported in the literature [65,67] a window was labeled as a FOG only if more than 50% of its samples were labeled as FOG.”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [65] Camps, 2018, Deep learning for freezing of gait detection in Parkinson’s disease patients in their homes using a waist-worn inertial measurement unit, Knowl. Based Syst. source
Not checked in the free run.
- cites [67] San-Segundo, R., Torres-Sánchez, R., Hodgins, J., and Fernando De La, T. (2019). Increasing Robustness in the Detection of Freezing of Gait in Parkinson’s Disease. Electronics, 8. source
Not checked in the free run.
May combine two claims.
PremisecitationR10Fold evaluation could show overly optimistic performance when using overlapped windows, because there is a high probability that the training and test subsets share some of the same data, because of the random splitting process.
“Despite being a widely used methodology, the R10Fold evaluation could show overly optimistic performance when using overlapped windows, because there is a high probability that the training and test subsets share some of the same data, because of the random splitting process [67].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [67] San-Segundo, R., Torres-Sánchez, R., Hodgins, J., and Fernando De La, T. (2019). Increasing Robustness in the Detection of Freezing of Gait in Parkinson’s Disease. Electronics, 8. source
Not checked in the free run.
Classified as probably not a substantial claim.
Study design and conductnone givenTo consider a window as a non-FOG window, all samples must be labeled as not FOG.
“Whereas to considered it as a non-FOG window, all samples must be labeled as not FOG.”
Quote in the paperSubstantial claimMay combine two claims.
InterpretationcitationNovelty detection is a reliable method for FOG detection and abnormal gait detection with a low computational burden and without the need for labeled data.
“Despite these results, the use of novelty detection presents a reliable method for FOG detection and abnormal gait detection with a low computational burden, without the need for labeled data [66].”
Quote in the paperSubstantial claim0 of 1 sources checked- cites [66] Mohammadian Rad, N., van Laarhoven, T., Furlanello, C., and Marchiori, E. (2018). Novelty Detection using Deep Normative Modeling for IMU-Based Abnormal Movement Monitoring in Parkinson’s Disease and Autism Spectrum Disorders. Sensors, 18. source
Not checked in the free run.
- The authors state this as an assumption: FOG events occur, but may not necessarily, as a deviation from a normal pattern.
Probably not a substantial claim: may not pass S4 (based).
Interpretationnone givenThe CNN-LSTM model allows analyzing sequential data by using previous windows in order to take advantage of temporal behavior found in FOG events.
“The CNN-LSTM model allows analyzing sequential data by using previous windows in order to take advantage of temporal behavior found in FOG events.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
InterpretationreasoningThe partial results of patients 2 and 20 tend to reduce the global results and this may be explained by the fact that some patients present different freezing patterns when they are at their homes or in different scenarios.
“The partial results of the patients 2 and 20 tend to reduce the global results and this may be explained by the fact that some patients present different freezing patterns when they are at their homes or in different scenarios.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationUsing 3-second windows or window between 2 and 4 seconds may enable the achievement of similar performances to longer windows.
“Mazilu [44] and Zach [63] have reported that using 3-second windows or window between 2 and 4 seconds may enable the achievement of similar performances that to longer windows.”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [44] Mazilu, S., Hardegger, M., Zhu, Z., Roggen, D., Tröster, G., Plotnik, M., and Hausdorff, J. (2012, January 21–24). Online Detection of Freezing of Gait with Smartphones and Machine Learning Techniques. Proceedings of the 6th International Conference on Pervasive Computing Technologies for Healthcare, San Diego, CA, USA. source
Not checked in the free run.
- cites [63] Zach, 2015, Identifying freezing of gait in Parkinson’s disease during freezing provoking tasks using waist-mounted accelerometry, Parkinsonism Relat Disord source
Not checked in the free run.
May combine two claims.
PremisecitationThe use of longer temporal windows has shown an improvement performance in FOG detection, while increasing the latency time.
“In the literature, the use of longer temporal windows has shown an improvement performance in FOG detection, while increasing the latency time.”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [65] Camps, 2018, Deep learning for freezing of gait detection in Parkinson’s disease patients in their homes using a waist-worn inertial measurement unit, Knowl. Based Syst. source
Not checked in the free run.
- cites [67] San-Segundo, R., Torres-Sánchez, R., Hodgins, J., and Fernando De La, T. (2019). Increasing Robustness in the Detection of Freezing of Gait in Parkinson’s Disease. Electronics, 8. source
Not checked in the free run.
May combine two claims.
PremisecitationThe use of larger batch sizes has shown a reduction in the ability of the model to generalize solutions and require a higher amount of memory to process the data.
“In practice, the use of larger batch sizes has shown a reduction in the ability of the model to generalize solutions and require a higher amount of memory to process the data [80].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [80] Keskar, N.S., Mudigere, D., Nocedal, J., Smelyanskiy, M., and Tang, P.T.P. (2016). On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima. arXiv.
Not checked in the free run.
May combine two claims.
PremisecitationThe addition of data from previous windows has shown an improvement in FOG detection, without the need of increasing the window size and avoiding increasing latency.
“The addition of data from previous windows has shown an improvement in FOG detection, without the need of increasing the window size and avoiding increasing latency [65,67].”
Quote in the paperSubstantial claim0 of 2 sources checked- cites [65] Camps, 2018, Deep learning for freezing of gait detection in Parkinson’s disease patients in their homes using a waist-worn inertial measurement unit, Knowl. Based Syst. source
Not checked in the free run.
- cites [67] San-Segundo, R., Torres-Sánchez, R., Hodgins, J., and Fernando De La, T. (2019). Increasing Robustness in the Detection of Freezing of Gait in Parkinson’s Disease. Electronics, 8. source
Not checked in the free run.
May combine two claims.
Interpretationown dataSpectral representations can achieve high performances with generic models, while hand-crafted FOG specific features (e.g., Mazilu features) show an improvement in the performance of personalized models.
“These results suggest that spectral representations can achieve high performances with generic models, while the use of hand-crafted FOG specific features (e.g., Mazilu features) show an improvement in the performance of personalized models.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationDeep neural networks are hard to train because of a large number of parameters that need to be adjusted in each layer.
“Deep neural networks are hard to train because of a large number of parameters that need to be adjusted in each layer [79].”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites [79] Glorot, 2010, Understanding the difficulty of training deep feedforward neural networks, J. Mach. Learn. Res.
Not checked in the free run.
May combine two claims.
Interpretationown dataPatients 2, 4, and 20 show a different pattern, making it difficult to identify FOG events through LOSO.
“These patients show a different pattern, making it difficult to identify FOG events though LOSO.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
Interpretationnone givenImbalanced data in FOG detection present an additional challenge to make accurate and reliable systems.
“Also, imbalanced data in FOG detection present an additional challenge to make accurate and reliable systems.”
Quote in the paperNot a substantial claimMay combine two claims.
InterpretationreasoningIn the proposed model, CNN layers work as a feature extractor, while LSTM increases classification capabilities by modeling the temporal phases of FOG events.
“In the proposed model the CNN layers work as a feature extractor, while the LSTM increases the classification capabilities by modeling the temporal phases of the FOG events.”
Quote in the paperSubstantial claimMay combine two claims.
InterpretationreasoningUsing more than three previous windows can be impractical for real-time applications due to computational burden, and the improvement tends to be more reduced at each additional time step.
“However, the use of more than three previous windows can be impractical for real-time applications, to avoid the computational burden, also the improvement tends to be more reduced at each additional time step.”
Quote in the paperSubstantial claimMay combine two claims.
Interpretationown dataThe use of previous windows shows a progressive improvement in FOG detection without increasing latency for use as an external cue.
“The use of previous windows shows a progressive improvement in FOG detection without increasing the latency in order to use the system as an external cue.”
Quote in the paperSubstantial claimMay combine two claims.
PremisecitationThe best results reported in similar studies using a single sensor in a free-living environment with LOSO evaluation are 79.03% and 74.67% [59] and 84.49% sensitivity and 85.83 specificity [64].
“The best results reported in similar studies using a single sensor on a free-living environment with LOSO evaluation are 79.03% and 74.67% [59] and 84.49% on sensitivity and 85.83 on specificity [64].”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites [59] Rodríguez-Martín, D., Samà, A., Pérez-López, C., Català, A., Moreno Arostegui, J.M., Cabestany, J., Bayés, À., Alcaine, S., Mestre, B., and Prats, A. (2017). Home detection of freezing of gait using support vector machines through a single waist-worn triaxial accelerometer. PLoS ONE, 12. source
Not checked in the free run.
- cites [64] Alcaine, 2018, Determining the optimal features in freezing of gait detection through a single waist accelerometer in home environments, Pattern Recog. Lett. source
Not checked in the free run.
Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).
PremisereasoningThe comparison of the results with most related literature work is difficult because of the diversity of approaches and validation methodologies.
“The comparison of the results with most of the related literature work is difficult, because of the diversity of approaches and validation methodologies.”
Quote in the paperNot a substantial claimMay combine two claims.
InterpretationreasoningThe best results with deep learning are achieved only when using a great amount of accurately labelled data, which is a time-consuming process.
“Moreover, the best results with DL are achieved only when using a great amount of data (accurately labelled) which is a time-consuming process.”
Quote in the paperNot a substantial claim
The paper relies on this without support.
FOG assessment through questionnaires can be subjective and may not provide an accurate representation of the severity of the symptom.
“FOG assessment has usually been made through questionnaires, however, this method can be subjective and could not provide an accurate representation of the severity of this symptom.”
The paper relies on this without support.
Sensor-based systems can provide accurate and objective information to track the symptoms’ evolution to optimize PD management and treatments.
“The use of sensor-based systems can provide accurate and objective information to track the symptoms’ evolution to optimize PD management and treatments.”
The paper relies on this without support.
The performance of existing wearable methods for FOG detection is usually lower when used at patients’ homes than in laboratory conditions.
“Several authors have proposed specific methods based on wearables and the analysis of inertial signals to detect FOG in laboratory conditions, however, its performance is usually lower when being used at patients’ homes.”
The authors state this as an assumption.
In prediction, patterns with abnormal movement (with significant statistical differences from normal patterns) are FOG events.
“In prediction, it is assumed that patterns with abnormal movement (with significant statistical differences from normal patterns) are FOG events.”
The authors state this as an assumption.
FOG events occur, but may not necessarily, as a deviation from a normal pattern.
“it is assumed that FOG events occur (but may not necessarily) as a deviation from a normal pattern.”
The paper relies on this without support.
Training deep models with recurrent layers requires a large number of computational resources.
“Despite the high performance of the DL approaches in FOG detection, it is important to consider that training deep models with recurrent layers requires a large number of computational resources.”
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Sigcha et al. uses a denoising autoencoder to detect freezing of gait (FOG) in Parkinson's disease patients.
“Sigcha et al. in [149] uses a denoising autoencoder to detect freezing of gait (FOG) in Parkinson’s disease patients.”
From Deep Learning in Human Activity Recognition with Wearable Sensors: A Review on Advances (2022) · cites this paper as [149]
Autoencoders are also commonly used to detect abnormal muscle movements, such as Parkinson's Disease and Autism Spectrum Disorder (ASD).
“Autoencoders are also commonly used to detect abnormal muscle movements, such as Parkinson’s Disease and Autism Spectrum Disorder (ASD).”
From Deep Learning in Human Activity Recognition with Wearable Sensors: A Review on Advances (2022) · cites this paper as [149]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Denoiser optimizer is reported as both RMSprop and Adam
MajorThe denoiser autoencoder's architecture section specifies RMSprop, while the general training section says the DL models were trained with Adam.
“The selected loss function was the binary cross entropy and the optimizer was the root-mean-square propagation (RMSprop) [76].”
“After fine tuning the hyperparameters, the DL models were trained via backpropagation using the following configuration: the optimization method was set to Adam, the learning rate was set to 1 × 10−4, a gradient clipping was implemented with clip value set to 1.”
Best model's AUC changes from 0.939 to 0.936
MinorThe baseline comparison gives the best CNN-LSTM model an AUC of 0.936, whereas the results and conclusion give it 0.939.
“an improvement in AUC from 0.904 to 0.936”
“Using LOSO evaluation, the CONV-LSTM model with three FFT previous windows achieved an AUC value of 0.939”
Reported FOG-Q range includes an excluded score
MinorThe stated inclusion criterion requires a score higher than 6, but the reported range starts at 6.
“a score higher than 6 on the FOG Questionnaire (FoG-Q) [23]”
“The results in the FOG-Q were a mean index of 15.2 (standard deviation: 4.47) with a range from 6 to 23.”
Patient 4 is inconsistently listed as an AUC exception
MinorOne per-patient summary lists only patients 2 and 20 as exceptions to the 0.915 AUC threshold; another also lists patient 4.
“In most patients, the AUC value reaches 0.915 except for the patients 2 and 20”
“the results in Table 8 show AUC values that exceed 0.915, except for the patients 2, 4, and 20.”
Reported class ratio does not match the percentages
MinorThe percentages imply an absence-of-FOG to FOG ratio of about 8.52 to 1, not 10 to 1. Read in the stated FOG-to-normal order, the ratio is reversed as well.
Reported 10 · recomputed 8.52381
“The ratio between FOG and normal ADL movement is 10 to 1, where the 89.5% corresponds to the absence of FOG symptoms and the remaining 10.5% correspond to FOG events.”
Cites
Breiman, L. (2001) · Machine Learning · doi:10.1023/a:1010933404324 · Cited by 133,538 · see citing works
Hochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318 · see citing works
LeCun, Y., Bengio, Y., & Hinton, G. (2015) · Nature · doi:10.1038/nature14539 · Cited by 84,126 · see citing works
Cortes, C., & Vapnik, V. (1995) · Machine Learning · doi:10.1007/bf00994018 · Cited by 59,421 · see citing works
Cortes, C., & Vapnik, V. N. (1995) · Machine Learning · doi:10.1023/a:1022627411411 · Cited by 34,565 · see citing works
A decision-theoretic generalization of on-line learning and an application to boosting
Freund, Y., & Schapire, R. E. (1997) · Journal of Computer and System Sciences · doi:10.1006/jcss.1997.1504 · Cited by 20,466 · see citing works
(2010) · Cited by 14,134 · see citing works
Representation Learning: A Review and New Perspectives
Bengio, Y., Courville, A., & Vincent, P. (2013) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2013.50 · Cited by 13,001 · see citing works
Parkinsonism: onset, progression, and mortality
Hoehn, M. M., & Yahr, M. D. (1967) · Neurology · doi:10.1212/wnl.17.5.427 · Cited by 12,107 · cited 1× · see citing works
“With three acceleration signals, a vector of 36 (12 × 3) features is obtained for each The first set of features used contain seven features proposed by Mazilu et al. [44]. was composed of the symmetric part (64-bin) of a 128-point fast Fourier transform When using three acceleration signals, a 21…”
MethodOverdependence on degraded gist memory in Alzheimer's disease.
Gallo, D. A., Shahid, K. R., Olson, M. A., Solomon, T. M., Schacter, D. L., & Budson, A. E. (2006) · Neuropsychology · doi:10.1037/0894-4105.20.6.625 · Cited by 4,756 · see citing works
Convolutional networks for images, speech, and time series
LeCun, Y., & Bengio, Y. (1998) · HAL · doi:10.5555/303568.303704 · Cited by 4,637 · see citing works
(2010) · Journal of Machine Learning Research · doi:10.5555/1756006.1953039 · Cited by 4,051 · see citing works
Adams, J., Dinesh, K., Xiong, M., Tarolli, C. G., Sharma, S., Sheth, N., Aranyosi, A. J., Zhu, W. Y., Goldenthal, S. B., Biglan, K., et al. (2017) · Digital Biomarkers · doi:10.1159/000479018 · Cited by 3,190 · see citing works
Deep Convolutional and LSTM Recurrent Neural Networks for Multimodal Wearable Activity Recognition
Ordóñez, F., & Roggen, D. (2016) · Sensors · doi:10.3390/s16010115 · Cited by 2,761 · see citing works
A Survey on Human Activity Recognition using Wearable Sensors
Lara, O. D., & Labrador, M. A. (2013) · IEEE Communications Surveys & Tutorials · doi:10.1109/surv.2012.110112.00192 · Cited by 2,662 · see citing works
Support Vector Method for Novelty Detection
Schölkopf, B., Williamson, R. C., Smola, A. J., Shawe-Taylor, J. S., & Platt, J. C. (1999) · Cited by 2,038 · see citing works
Deep learning for sensor-based activity recognition: A survey
Wang, J., Chen, Y., Hao, S., Peng, X., & Hu, L. (2019) · Pattern Recognition Letters · doi:10.1016/j.patrec.2018.02.010 · Cited by 2,014 · cited 1× · see citing works
“The CNN-LSTM model was also used for the analysis of data representation with previous temporal windows.”
MethodBergstra, J., Yamins, D. L. K., & Cox, D. (2013) · Digital Access to Scholarship at Harvard (DASH) (Harvard University) · Cited by 1,817 · see citing works
A State-of-the-Art Survey on Deep Learning Theory and Architectures
Alom, M. Z., Taha, T. M., Yakopcic, C., Westberg, S. K. E., Sidike, P., Nasrin, M. S., Hasan, M., Essen, B. C. V., Awwal, A. A. S., & Asari, V. K. (2019) · Electronics · doi:10.3390/electronics8030292 · Cited by 1,643 · see citing works
The clinical symptoms of Parkinson's disease.
Sveinbjornsdottir, S. (2016) · Journal of Neurochemistry · doi:10.1111/jnc.13691 · Cited by 1,410 · cited 2× · see citing works
“The third is a spectral data representation based fast Fourier transform similar to that used in Camps et al. [65].”
MethodLSTM Fully Convolutional Networks for Time Series Classification
Karim, F., Majumdar, S., Darabi, H., & Chen, S. (2017) · IEEE Access · doi:10.1109/access.2017.2779939 · Cited by 1,358 · see citing works
Freezing of gait: moving forward on a mysterious clinical phenomenon
Nutt, J. G., Bloem, B. R., Giladi, N., Hallett, M., Horak, F. B., & Nieuwboer, A. (2011) · The Lancet Neurology · doi:10.1016/s1474-4422(11)70143-0 · Cited by 1,308 · see citing works
Epidemiology and etiology of Parkinson's disease: a review of the evidence
Wirdefeldt, K., Adami, H.-O., Cole, P., Trichopoulos, D., & Mandel, J. (2011) · European Journal of Epidemiology · doi:10.1007/s10654-011-9581-6 · Cited by 1,199 · see citing works
High-dimensional and large-scale anomaly detection using a linear one-class SVM with deep learning
Erfani, S. M., Rajasegarar, S., Karunasekera, S., & Leckie, C. (2016) · Pattern Recognition · doi:10.1016/j.patcog.2016.03.028 · Cited by 1,162 · see citing works
(2015) · Cited by 989 · see citing works
Ageing and Parkinson's disease: Why is advancing age the biggest risk factor?
Reeve, A. K., Simcox, E. M., & Turnbull, D. M. (2014) · Ageing Research Reviews · doi:10.1016/j.arr.2014.01.004 · Cited by 961 · see citing works
Nweke, H. F., Teh, Y. W., Al-garadi, M. A., & Alo, U. R. (2018) · Expert Systems with Applications · doi:10.1016/j.eswa.2018.03.056 · Cited by 930 · see citing works
Human motion tracking for rehabilitation-A survey
Zhou, H., & Hu, H. (2008) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2007.09.001 · Cited by 813 · see citing works
Construction of freezing of gait questionnaire for patients with Parkinsonism
Giladi, N., Shabtai, H., Simon, E. S., Biran, S., Tal, J., & Korczyn, A. D. (2000) · Parkinsonism & Related Disorders · doi:10.1016/s1353-8020(99)00062-0 · Cited by 789 · see citing works
Schaafsma, J. D., Balash, Y., Gurevich, T., Bartels, A. L., Hausdorff, J. M., & Giladi, N. (2003) · European Journal of Neurology · doi:10.1046/j.1468-1331.2003.00611.x · Cited by 727 · see citing works
Wearable Assistant for Parkinson’s Disease Patients With the Freezing of Gait Symptom
Bachlin, M., Plotnik, M., Roggen, D., Maidan, I., Hausdorff, J. M., Giladi, N., & Troster, G. (2010) · IEEE Transactions on Information Technology in Biomedicine · doi:10.1109/titb.2009.2036165 · Cited by 693 · cited 7× · see citing works
“[66] Novelty detection with CNN denoising autoencoders 9 accelerometer signals from Daphnet [68] 10 PD (8 with FOG) Validation of a method to detect abnormal movement without the need for labeled data for training.”
BackgroundMethodInfluentialUnderstanding heterogeneity in clinical cohorts using normative models: Beyond case-control studies
Marquand, A. F., Rezek, I., Buitelaar, J., & Beckmann, C. F. (2016) · Biological Psychiatry · doi:10.1016/j.biopsych.2015.12.023 · Cited by 683 · see citing works
Nieuwboer, A., Rochester, L., Herman, T., Vandenberghe, W., Emil, G. E., Thomaes, T., & Giladi, N. (2009) · Gait & Posture · doi:10.1016/j.gaitpost.2009.07.108 · Cited by 637 · see citing works
Digital Signal Processing: A Practical Guide for Engineers and Scientists
Smith, S. W. (2002) · doi:10.1016/b978-0-7506-7444-7.x5036-5 · Cited by 609 · see citing works
Understanding Dopaminergic Cell Death Pathways in Parkinson Disease
Michel, P. P., Hirsch, É. C., & Hunot, S. (2016) · Neuron · doi:10.1016/j.neuron.2016.03.038 · Cited by 594 · see citing works
Predictors of future falls in Parkinson disease
Kerr, G., Worringham, C. J., Cole, M. H., Lacherez, P. F., Wood, J. M., & Silburn, P. A. (2010) · Neurology · doi:10.1212/wnl.0b013e3181e7b688 · Cited by 550 · see citing works
Ambulatory monitoring of freezing of gait in Parkinson’s disease
Moore, S. T., MacDougall, H. G., & Ondo, W. G. (2008) · Journal of Neuroscience Methods · doi:10.1016/j.jneumeth.2007.08.023 · Cited by 526 · see citing works
Lim, I., van Wegen, E. E. H., de Goede, C. J. T., Deutekom, M. D. L. F., Nieuwboer, A., Willems, A., Jones, D. E., Rochester, L., & Kwakkel, G. (2005) · Clinical Rehabilitation · doi:10.1191/0269215505cr906oa · Cited by 494 · see citing works
The History of Parkinson's Disease: Early Clinical Descriptions and Neurological TherapiesPreprint
Goetz, C. G. (2011) · Cold Spring Harbor Perspectives in Medicine · doi:10.1101/cshperspect.a008862 · Cited by 472 · see citing works
How to Construct Deep Recurrent Neural NetworksPreprint
Pascanu, R., Gülçehre, Ç., Cho, K., & Bengio, Y. (2013) · arXiv (Cornell University) · doi:10.48550/arxiv.1312.6026 · Cited by 435 · see citing works
Validation of the freezing of gait questionnaire in patients with Parkinson’s disease
Giladi, N., Tal, J., Azulay, T., Rascol, O., Brooks, D. J., Melamed, E., Oertel, W., Poewe, W. H., Stocchi, F., & Tolosa, E. (2009) · Movement Disorders · doi:10.1002/mds.21745 · Cited by 431 · see citing works
Freezing of gait in patients with advanced Parkinson's disease
Giladi, N., Treves, T. A., Simon, E. S., Shabtai, H., Orlov, Y., Kandinov, B., Paleacu, D., & Korczyn, A. D. (2001) · Journal of Neural Transmission · doi:10.1007/s007020170096 · Cited by 404 · see citing works
Clinical and physiological assessments for elucidating falls risk in Parkinson's disease
Latt, M. D., Lord, S. R., Morris, J. G., & Fung, V. S. C. (2009) · Movement Disorders · doi:10.1002/mds.22561 · Cited by 402 · see citing works
Quantitative wearable sensors for objective assessment of Parkinson's disease
Maetzler, W., Domingos, J., Srulijes, K., Ferreira, J. J., & Bloem, B. R. (2013) · Movement Disorders · doi:10.1002/mds.25628 · Cited by 374 · see citing works
Moore, O., Peretz, C., & Giladi, N. (2007) · Movement Disorders · doi:10.1002/mds.21659 · Cited by 335 · see citing works
Nieuwboer, A., Dom, R., De Weerdt, W., Desloovere, K., Fieuws, S., & Broens‐Kaucsik, E. (2001) · Movement Disorders · doi:10.1002/mds.1206 · Cited by 320 · cited 3× · see citing works
“14 0.928 0.927 0.928 The results presented are the mean values of six iterations per patient with LOSO evaluation, in a similar manner to that of previous experiments.”
MethodResultInfluentialComparison of Feature Learning Methods for Human Activity Recognition Using Wearable Sensors
Li, F., Shirahama, K., Nisar, M., Köping, L., & Grzegorzek, M. (2018) · Sensors · doi:10.3390/s18020679 · Cited by 302 · see citing works
Steven Eyobu, O., & Han, D. S. (2018) · Sensors · doi:10.3390/s18092892 · Cited by 259 · see citing works
Rodríguez-Martín, D., Samà, A., Pérez-López, C., Català, A., Moreno Arostegui, J. M., Cabestany, J., Bayés, À., Alcaine, S., Mestre, B., Prats, A., et al. (2017) · PLOS ONE · doi:10.1371/journal.pone.0171764 · Cited by 237 · see citing works
Online Detection of Freezing of Gait with Smartphones and Machine Learning Techniques
Mazilu, S., Hardegger, M., Zhu, Z., Roggen, D., Troester, G., Plotnik, M., & Hausdorff, J. (2012) · Proceedings of the 6th International Conference on Pervasive Computing Technologies for Healthcare · doi:10.4108/icst.pervasivehealth.2012.248680 · Cited by 231 · see citing works
Camps, J., Samà, A., Martín, M., Rodríguez-Martín, D., Pérez-López, C., Moreno Arostegui, J. M., Cabestany, J., Català, A., Alcaine, S., Mestre, B., et al. (2018) · Knowledge-Based Systems · doi:10.1016/j.knosys.2017.10.017 · Cited by 224 · see citing works
Kubota, K. J., Chen, J. A., & Little, M. A. (2016) · Movement Disorders · doi:10.1002/mds.26693 · Cited by 210 · cited 1× · see citing works
“The CNN-LSTM model was also used for the analysis of data representation with previous temporal windows.”
MethodMoore, S. T., Yungher, D. A., Morris, T. R., Dilda, V., MacDougall, H. G., Shine, J. M., Naismith, S. L., & Lewis, S. J. (2013) · Journal of NeuroEngineering and Rehabilitation · doi:10.1186/1743-0003-10-19 · Cited by 204 · see citing works
Automatic detection of freezing of gait events in patients with Parkinson's disease
Tripoliti, E. E., Tzallas, A. T., Tsipouras, M. G., Rigas, G., Bougia, P., Leontiou, M., Konitsiotis, S., Chondrogiorgi, M., Tsouli, S. G., & Fotiadis, D. I. (2012) · Computer Methods and Programs in Biomedicine · doi:10.1016/j.cmpb.2012.10.016 · Cited by 196 · see citing works
Effects of external cues on gait parameters of Parkinson's disease patients: A systematic review
Rocha, P., Porfírio, G. M., Ferraz, H. B., & Trevisani, V. F. M. (2014) · Clinical Neurology and Neurosurgery · doi:10.1016/j.clineuro.2014.06.026 · Cited by 187 · see citing works
Anwary, A. R., Yu, H., & Vassallo, M. (2018) · IEEE Sensors Journal · doi:10.1109/jsen.2017.2786587 · Cited by 178 · see citing works
A frequency and correlation analysis of motor deficits in Parkinson patients
Nieuwboer, A., Weerdt, W. D., Dom, R., & Lesaffre, E. (1998) · Disability and Rehabilitation · doi:10.3109/09638289809166074 · Cited by 175 · see citing works
The freezing phenomenon in parkinsonism.
Fahn, S. (1995) · PubMed · Cited by 174 · see citing works
Wearable sensors : fundamentals, implementation and applications
Sazonov, E. E., & Neuman, M. R. (2014) · doi:10.1016/c2018-0-01225-6 · Cited by 169 · see citing works
Speech Quality of VoIP: Assessment and Prediction
Raake, A. (2006) · Cited by 163 · see citing works
A clinical view on the development of technology‐based tools in managing Parkinson's disease
Maetzler, W., Klucken, J., & Horne, M. (2016) · Movement Disorders · doi:10.1002/mds.26673 · Cited by 162 · see citing works
Arias, P., & Cudeiro, J. (2010) · PLoS ONE · doi:10.1371/journal.pone.0009675 · Cited by 161 · see citing works
New methods for the assessment of Parkinson's disease (2005 to 2015): A systematic review
Sánchez‐Ferro, Á., Elshehabi, M., Godinho, C., Salkovic, D., Hobert, M. A., Domingos, J., van Uem, J. M., Ferreira, J. J., & Maetzler, W. (2016) · Movement Disorders · doi:10.1002/mds.26723 · Cited by 145 · see citing works
Raake, A. (2006) · Wiley eBooks · doi:10.1002/9780470033005 · Cited by 144 · see citing works
Detecting freezing of gait with a tri-axial accelerometer in Parkinson’s disease patients
Ahlrichs, C., Samà, A., Lawo, M., Cabestany, J., Rodríguez-Martín, D., Pérez-López, C., Sweeney, D., Quinlan, L. R., Laighin, G. Ò., Counihan, T., et al. (2016) · Medical & Biological Engineering & Computing · doi:10.1007/s11517-015-1395-3 · Cited by 128 · see citing works
The clinical spectrum of freezing of gait in Parkinson's disease
Okuma, Y., & Yanagisawa, N. (2008) · Movement Disorders · doi:10.1002/mds.21934 · Cited by 119 · see citing works
Human Activity Recognition from Body Sensor Data using Deep Learning
Hassan, M. M., Huda, S., Uddin, M. Z., Almogren, A., & Alrubaian, M. (2018) · Journal of Medical Systems · doi:10.1007/s10916-018-0948-z · Cited by 113 · see citing works
A Technological Review of Wearable Cueing Devices Addressing Freezing of Gait in Parkinson’s Disease
Sweeney, D., Quinlan, L. R., Browne, P., Richardson, M., Meskell, P., & ÓLaighin, G. (2019) · Sensors · doi:10.3390/s19061277 · Cited by 103 · see citing works
A Forecast of the Adoption of Wearable Technology
Page, T. (2015) · International Journal of Technology Diffusion · doi:10.4018/ijtd.2015040102 · Cited by 102 · see citing works
Feature extraction from smartphone inertial signals for human activity segmentation
San-Segundo, R., Montero, J. M., Barra-Chicote, R., Fernández-Martínez, F., & Pardo, J. M. (2015) · Signal Processing · doi:10.1016/j.sigpro.2015.09.029 · Cited by 102 · see citing works
Freezing of Gait and Falls in Parkinson's Disease
Okuma, Y. (2014) · Journal of Parkinson s Disease · doi:10.3233/jpd-130282 · Cited by 100 · see citing works
Freezing of gait. Clinical overview.
Giladi, N. (2001) · PubMed · Cited by 100 · see citing works
Zach, H., Janssen, A. M., Snijders, A. H., Delval, A., Ferraye, M. U., Auff, E., Weerdesteyn, V., Bloem, B. R., & Nonnekes, J. (2015) · Parkinsonism & Related Disorders · doi:10.1016/j.parkreldis.2015.09.051 · Cited by 97 · see citing works
Increasing Robustness in the Detection of Freezing of Gait in Parkinson’s Disease
San-Segundo, R., Navarro-Hellín, H., Torres-Sánchez, R., Hodgins, J., & De la Torre, F. (2019) · Electronics · doi:10.3390/electronics8020119 · Cited by 86 · see citing works
Mobile phone applications in Parkinson's disease: a systematic review
Rey, M. L.-D., Vela, L., & la Cuerda, R. C.-D. (2018) · Neurología (English Edition) · doi:10.1016/j.nrleng.2018.12.002 · Cited by 83 · see citing works
Samà, A., Rodríguez-Martín, D., Pérez-López, C., Català, A., Alcaine, S., Mestre, B., Prats, A., Crespo, M. C., & Bayés, À. (2018) · Pattern Recognition Letters · doi:10.1016/j.patrec.2017.05.009 · Cited by 76 · see citing works
Mohammadian Rad, N., Van Laarhoven, T., Furlanello, C., & Marchiori, E. (2018) · Sensors · doi:10.3390/s18103533 · Cited by 76 · see citing works
Unconstrained detection of freezing of Gait in Parkinson's disease patients using smartphone
Kim, H., Lee, H. J., Lee, W., Kwon, S., Kim, S. K., Jeon, H. S., Park, H.-Y., Shin, C., Yi, W. J., Jeon, B. S., et al. (2015) · Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc.2015.7319209 · Cited by 75 · see citing works
Aplicaciones móviles en la enfermedad de Parkinson: una revisión sistemática
Rey, M. L.-D., Vela, L., & Cano‐de‐la‐Cuerda, R. (2017) · Neurología · doi:10.1016/j.nrl.2017.03.006 · Cited by 68 · see citing works
Online Detection of Freezing of Gait in Parkinson’s Disease Patients: A Performance Characterization
Bächlin, M., Roggen, D., Plotnik, M., Hausdorff, J. M., Giladi, N., & Tröster, G. (2009) · International Conference on Body Area Networks · doi:10.4108/icst.bodynets2009.5852 · Cited by 63 · see citing works
Wang, Z., Li, J., Wang, J., Zhao, H., Qiu, S., Yang, N., & Shi, X. (2018) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2018.2826198 · Cited by 61 · see citing works
The Epidemiology of Parkinson's Disease
Chen, S.-Y., & Tsai, S. (2010) · Tzu Chi Medical Journal · doi:10.1016/s1016-3190(10)60044-4 · Cited by 44 · see citing works
Towards Wearable-Inertial-Sensor-Based Gait Posture Evaluation for Subjects with Unbalanced Gaits
Qiu, S., Wang, H. H., Li, J., Zhao, H., Wang, Z., Wang, J., Wang, Q., Plettemeier, D., Bärhold, M., Bauer, T., et al. (2020) · Sensors · doi:10.3390/s20041193 · Cited by 41 · see citing works
A Perspective on Wearable Sensor Measurements and Data Science for Parkinson’s Disease
Matias, R., Paixão, V., Bouça‐Machado, R., & Ferreira, J. J. (2017) · Frontiers in Neurology · doi:10.3389/fneur.2017.00677 · Cited by 38 · see citing works
Yungher, D. A., Morris, T. R., Dilda, V., Shine, J. M., Naismith, S. L., Lewis, S. J. G., & Moore, S. T. (2014) · Parkinson's Disease · doi:10.1155/2014/606427 · Cited by 28 · see citing works
Flagged first, then by citations
2 of the 87 references couldn’t be matched to a known paper.
Research works printed but not linked: 3
- Cabestany, 1992, Assessing Motor Fluctuations in Parkinson’s Disease Patients Based on a Single Inertial Sensor; Accuracy of clinical diagnosis of idiopathic Parkinson’s disease: A clinico-pathological study of 100 cases, J. Neurol. Neurosurg Psychiatry, 16, 181
- Kingma, D.P., and Ba, L.J. (2014). Adam: A Method for Stochastic Optimization. arXiv.
- Keskar, N.S., Mudigere, D., Nocedal, J., Smelyanskiy, M., and Tang, P.T.P. (2016). On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima. arXiv.
Cited by
Sorted by citations
Deep Learning in Human Activity Recognition with Wearable Sensors: A Review on Advances
Zhang, S., Li, Y., Zhang, S., Shahabi, F., Xia, S., Deng, Y., & Alshurafa, N. (2022) · Sensors · doi:10.3390/s22041476 · Cited by 625 · see citing works
Khan, P., Kader, M. F., Islam, S. M. R., Rahman, A. B., Kamal, M. S., Toha, M. U., & Kwak, K. S. (2021) · IEEE Access · doi:10.1109/access.2021.3062484 · Cited by 294 · see citing works
Measuring freezing of gait during daily-life: an open-source, wearable sensors approach
Mancini, M., Shah, V. V., Stuart, S., Curtze, C., Horak, F. B., Safarpour, D., & Nutt, J. G. (2021) · Journal of NeuroEngineering and Rehabilitation · doi:10.1186/s12984-020-00774-3 · Cited by 180 · see citing works
Prediction of Freezing of Gait in Parkinson’s Disease Using Wearables and Machine Learning
Borzì, L., Mazzetta, I., Zampogna, A., Suppa, A., Olmo, G., & Irrera, F. (2021) · Sensors · doi:10.3390/s21020614 · Cited by 134 · see citing works
Sigcha, L., Borzì, L., Amato, F., Rechichi, I., Ramos-Romero, C. A., Cardenas, A. J. A., Gascó, L., & Olmo, G. (2023) · Expert Systems with Applications · doi:10.1016/j.eswa.2023.120541 · Cited by 139 · see citing works
Self-citationMulti-Modal Deep Learning Diagnosis of Parkinson’s Disease—A Systematic Review
Skaramagkas, V., Pentari, A., Kefalopoulou, Z. M., & Tsiknakis, M. (2023) · IEEE Transactions on Neural Systems and Rehabilitation Engineering · doi:10.1109/tnsre.2023.3277749 · Cited by 119 · see citing works
Chandrabhatla, A. S., Pomeraniec, I. J., & Ksendzovsky, A. (2022) · npj Digital Medicine · doi:10.1038/s41746-022-00568-y · Cited by 109 · see citing works
Din, S. D., Kirk, C., Yarnall, A. J., Rochester, L., & Hausdorff, J. M. (2021) · Journal of Parkinson s Disease · doi:10.3233/jpd-202471 · Cited by 108 · see citing works
Detection and assessment of Parkinson's disease based on gait analysis: A survey
Guo, Y., Yang, J., Liu, Y., Chen, X., & Yang, G. (2022) · Frontiers in Aging Neuroscience · doi:10.3389/fnagi.2022.916971 · Cited by 91 · see citing works
Shi, B., Tay, A. E., Au, W., Tan, D. M. L., Chia, N. S. Y., & Yen, S. (2022) · IEEE Transactions on Biomedical Engineering · doi:10.1109/tbme.2022.3140258 · Cited by 80 · see citing works
Borzì, L., Sigcha, L., Rodríguez-Martín, D., & Olmo, G. (2022) · Artificial Intelligence in Medicine · doi:10.1016/j.artmed.2022.102459 · Cited by 74 · see citing works
Self-citationAssessing inertial measurement unit locations for freezing of gait detection and patient preference
O’Day, J. J., Lee, M. R., Seagers, K. A., Hoffman, S. L., Jih-Schiff, A., Kidziński, Ł., Delp, S. L., & Bronte-Stewart, H. M. (2022) · Journal of NeuroEngineering and Rehabilitation · doi:10.1186/s12984-022-00992-x · Cited by 72 · see citing works
Reches, T., Dagan, M., Herman, T., Gazit, E., Gouskova, N., Giladi, N., Manor, B. D., & Hausdorff, J. M. (2020) · Sensors · doi:10.3390/s20164474 · Cited by 63 · see citing works
Hoai Thu, N. T., & Han, D. S. (2021) · IEEE Access · doi:10.1109/access.2021.3122298 · Cited by 54 · see citing works
Bačanin, N., Petrović, A., Jovanović, L., Zivkovic, M. Z., Živković, T., & Šarac, M. (2024) · International Conference on Mobile Computing and Sustainable Informatics (ICMCSI) · doi:10.1109/icmcsi61536.2024.00006 · Cited by 53 · see citing works
Chawla, P., Rana, S. B., Kaur, H., Singh, K., Yuvaraj, R., & Murugappan, M. (2022) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2022.104116 · Cited by 52 · see citing works
Sigcha, L., Borzì, L., Pavón, I., Costa, N., da Costa, S. P., Arezes, P., López, J. M., & de Arcas, G. (2022) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2022.105482 · Cited by 53 · see citing works
Self-citationSalomon, A., Gazit, E., Ginis, P., Urazalinov, B., Takoi, H., Yamaguchi, T., Goda, S., Lander, D., Lacombe, J., Sinha, A. K., et al. (2024) · Nature Communications · doi:10.1038/s41467-024-49027-0 · Cited by 54 · see citing works
Huang, T., Li, M., & Huang, J. (2023) · Frontiers in Aging Neuroscience · doi:10.3389/fnagi.2023.1119956 · Cited by 51 · see citing works
Parkinson’s disease diagnosis using deep learning: A bibliometric analysis and literature review
Abumalloh, R. A., Nilashi, M., Samad, S., Ahmadi, H., Alghamdi, A., Alrizq, M., & AlYami, S. S. (2024) · Ageing Research Reviews · doi:10.1016/j.arr.2024.102285 · Cited by 48 · see citing works
Parkinson’s disease diagnosis using neural networks: Survey and comprehensive evaluation
Tanveer, M., Rashid, A. H., Kumar, R., & Raman, B. (2022) · Information Processing & Management · doi:10.1016/j.ipm.2022.102909 · Cited by 46 · see citing works
Naghavi, N., & Wade, E. (2021) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2021.3103071 · Cited by 45 · see citing works
Filtjens, B., Ginis, P., Nieuwboer, A., Slaets, P., & Vanrumste, B. (2022) · Journal of NeuroEngineering and Rehabilitation · doi:10.1186/s12984-022-01025-3 · Cited by 40 · see citing works
Pratihar, R., & Sankar, R. T. (2024) · Computers · doi:10.3390/computers13110293 · Cited by 38 · see citing works
Recognition of freezing of gait in Parkinson’s disease based on combined wearable sensors
Ren, K., Chen, Z., Ling, Y., & Zhao, J. (2022) · BMC Neurology · doi:10.1186/s12883-022-02732-z · Cited by 39 · see citing works
IMU-Based Monitoring for Assistive Diagnosis and Management of IoHT: A Review
Bo, F., Yerebakan, M., Dai, Y., Wang, W., Li, J., Hu, B., & Gao, S. (2022) · Healthcare · doi:10.3390/healthcare10071210 · Cited by 35 · see citing works
Multi-Headed Conv-LSTM Network for Heart Rate Estimation during Daily Living Activities
Wilkosz, M., & Szczęsna, A. (2021) · Sensors · doi:10.3390/s21155212 · Cited by 32 · see citing works
Habib, Z., Mughal, M. A., Khan, M. A., & Shabaz, M. (2023) · Alexandria Engineering Journal · doi:10.1016/j.aej.2023.11.075 · Cited by 33 · see citing works
Sigcha, L., Domínguez, B., Borzì, L., Costa, N., da Costa, S. P., Arezes, P., López, J. M., de Arcas, G., & Pavón, I. (2022) · Electronics · doi:10.3390/electronics11233879 · Cited by 31 · see citing works
Self-citationAshfaque Mostafa, T., Soltaninejad, S., McIsaac, T. L., & Cheng, I. (2021) · Sensors · doi:10.3390/s21196446 · Cited by 32 · see citing works
Deep Learning for Daily Monitoring of Parkinson’s Disease Outside the Clinic Using Wearable Sensors
Atri, R., Urban, K., Marebwa, B., Simuni, T., Tanner, C. M., Siderowf, A. D., Frasier, M. A., Haas, M., & Lancashire, L. (2022) · Sensors · doi:10.3390/s22186831 · Cited by 29 · see citing works
Habib, Z., Mughal, M. A., Khan, M. A., Hamza, A., Alturki, N. M., & Jamel, L. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-18906-5 · Cited by 30 · see citing works
Dhivyaa, C. R., Nithya, K., & Anbukkarasi, S. (2024) · IEEE Access · doi:10.1109/access.2024.3487001 · Cited by 28 · see citing works
Detection and prediction of freezing of gait with wearable sensors in Parkinson’s disease
Zhang, W., Sun, H., Huang, D., Zhang, Z., Li, J., Wu, C., Sun, Y., Gong, M., Wang, Z., Sun, C., et al. (2023) · Neurological Sciences · doi:10.1007/s10072-023-07017-y · Cited by 31 · see citing works
Guo, C. C., Chiesa, P. A., de Moor, C., Fazeli, M. S., Schofield, T., Hofer, K., Belachew, S. M., & Scotland, A. (2022) · Journal of Medical Internet Research · doi:10.2196/37683 · Cited by 28 · see citing works
Filtjens, B., Ginis, P., Nieuwboer, A., Afzal, M. R., Spildooren, J., Vanrumste, B., & Slaets, P. (2021) · BMC Medical Informatics and Decision Making · doi:10.1186/s12911-021-01699-0 · Cited by 25 · see citing works
Borzì, L., Sigcha, L., & Olmo, G. (2023) · Sensors · doi:10.3390/s23094426 · Cited by 24 · see citing works
Self-citationSigcha, L., Borzì, L., & Olmo, G. (2024) · Expert Systems with Applications · doi:10.1016/j.eswa.2024.124522 · Cited by 24 · see citing works
Self-citationFreezing of Gait Detection for Parkinson’s Disease Patients using Accelerometer Data: Case Study
Tăuţan, A.-M., Andrei, A.-G., & Ionescu, B. E. (2020) · E-Health and Bioengineering Conference (EHB) · doi:10.1109/ehb50910.2020.9280223 · Cited by 24 · see citing works
Serena, R. S., Hinterwimmer, F., Burgkart, R., von Eisenhart‐Rothe, R., & Rueckert, D. (2024) · JMIR mhealth and uhealth · doi:10.2196/60521 · Cited by 26 · see citing works
Metaheuristics with Deep Learning-Enabled Parkinson’s Disease Diagnosis and Classification Model
Bahaddad, A. A. S., Ragab, M., Ashary, E. B., & Khalil, E. M. (2022) · Journal of Healthcare Engineering · doi:10.1155/2022/9276579 · Cited by 23 · see citing works
Yan, Y., Liu, Y.-S., Li, C., Wang, J., Ma, L., Xiong, J., Zhao, X., & Wang, L. (2022) · IEEE Sensors Journal · doi:10.1109/jsen.2022.3142750 · Cited by 22 · see citing works
Prediction of freezing of gait based on self-supervised pretraining via contrastive learning
Xia, Y., Sun, H. D., Zhang, B., Xu, Y., & Ye, Q. (2023) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2023.105765 · Cited by 22 · see citing works
Recognition of Freezing of Gait in Parkinson’s Disease Based on Machine Vision
Li, W., Chen, X., Zhang, J., Lu, J., Zhang, C., Bai, H., Liang, J., Wang, J., Du, H., Xue, G., et al. (2022) · Frontiers in Aging Neuroscience · doi:10.3389/fnagi.2022.921081 · Cited by 21 · see citing works
Elbatanouny, H., Kleanthous, N., Dahrouj, H., Alusi, S. H., Almajali, E. R., Mahmoud, S. A., & Hussain, A. J. (2024) · Sensors · doi:10.3390/s24123959 · Cited by 21 · see citing works
A novel automated Parkinson’s disease identification approach using deep learning and EEG
Obayya, M. I., Saeed, M. K., Maashi, M. S., Alotaibi, S. S., Salama, A. S., & Hamza, M. A. (2023) · PeerJ Computer Science · doi:10.7717/peerj-cs.1663 · Cited by 19 · see citing works
IoT Based Fall Detection System for Elderly Healthcare
Tahir, A., Taylor, W., Taha, A., Usman, M., Shah, S. A., Imran, M. A., & Abbasi, Q. H. (2022) · Studies in computational intelligence · doi:10.1007/978-981-16-8488-3_10 · Cited by 19 · see citing works
Abbasi, S., & Rezaee, K. (2024) · Brain and Behavior · doi:10.1002/brb3.70206 · Cited by 18 · see citing works
Wang, F., Liu, Y., Cao, H., Zhao, Y., Ge, Y., Song, X., Li, C., Yang, X., Jiang, X., & Gu, X. (2023) · Composites Science and Technology · doi:10.1016/j.compscitech.2023.110118 · Cited by 17 · see citing works
Technology for measuring freezing of gait: Current state of the art and recommendations
Mancini, M., McKay, J. L., Cockx, H. M., D’Cruz, N., Esper, C. D., Filtjens, B., Heimler, B., MacKinnon, C. D., Palmerini, L., Roerdink, M., et al. (2025) · Journal of Parkinson s Disease · doi:10.1177/1877718x241301065 · Cited by 18 · see citing works
Al-Adhaileh, M. H., Wadood, A., Aldhyani, T. H. H., Khan, S. H., Uddin, M. I., & Al-Nefaie, A. H. A. (2025) · Frontiers in Physiology · doi:10.3389/fphys.2025.1581699 · Cited by 17 · see citing works
Inertial Sensor-Based Instrumented Cane for Real-Time Walking Cane Kinematics Estimation
Fernandez, I. G., Ahmad, S. A., & Wada, C. (2020) · Sensors · doi:10.3390/s20174675 · Cited by 15 · see citing works
Diep, C., O’Day, J. J., Kehnemouyi, Y. M., Burnett, G., & Bronte-Stewart, H. M. (2021) · Sensors · doi:10.3390/s21082661 · Cited by 14 · see citing works
Dvorani, A., Waldheim, V., Jochner, M. C. E., Salchow-Hömmen, C., Meyer-Ohle, J., Kühn, A. A., Wenger, N., & Schauer, T. (2021) · Frontiers in Neurology · doi:10.3389/fneur.2021.720516 · Cited by 14 · see citing works
Klaver, E., Heijink, I. B., Silvestri, G., van Vugt, J. P. P., Janssen, S., Nonnekes, J. H., Wezel, R. J. A. V., & Tjepkema‐Cloostermans, M. C. (2023) · Frontiers in Neurology · doi:10.3389/fneur.2023.1306129 · Cited by 14 · see citing works
Qi, W., Shen, S., Dong, C., Zhao, M., Zang, S., Zhu, X., Li, J., Wang, B., Shi, Y., Dong, Y., et al. (2025) · Journal of Medical Internet Research · doi:10.2196/71560 · Cited by 16 · see citing works
Phokaewvarangkul, O., Vateekul, P., Wichakam, I., Anan, C., & Bhidayasiri, R. (2021) · Frontiers in Aging Neuroscience · doi:10.3389/fnagi.2021.727654 · Cited by 11 · see citing works
Enhancing Human Activity Recognition Through Sensor Fusion And Hybrid Deep Learning Model
Tarekegn, A. N., Ullah, M., Cheikh, F. A., & Sajjad, M. (2023) · IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW) · doi:10.1109/icasspw59220.2023.10193698 · Cited by 11 · see citing works
Dadashzadeh, A., Whone, A. L., Rolinski, M., & Mirmehdi, M. (2021) · International Conference on Pattern Recognition Applications and Methods · doi:10.5220/0010309200890097 · Cited by 10 · see citing works
Efficient Human Gait Activity Recognition Based on Sensor Fusion and Intelligent Stacking Framework
Tarekegn, A. N., Sajjad, M., Cheikh, F. A., Ullah, M., & Muhammad, K. (2023) · IEEE Sensors Journal · doi:10.1109/jsen.2023.3319353 · Cited by 10 · see citing works
A Study on the Application of LSTM to Judge Bike Accidents for Inflating Wearable Airbags
Jo, S.-H., Woo, J., Byun, G., Kwon, B.-S., & Jeong, J. H. (2021) · Sensors · doi:10.3390/s21196541 · Cited by 10 · see citing works
Al-Nefaie, A. H. A., Aldhyani, T. H. H., Farhah, N. S., & Koundal, D. (2024) · Frontiers in Medicine · doi:10.3389/fmed.2024.1418684 · Cited by 9 · see citing works
Lin, O., Tian, Y., Hou, Y., Zhu, Y., & Liu, X. (2023) · International IEEE/EMBS Conference on Neural Engineering · doi:10.1109/ner52421.2023.10123828 · Cited by 9 · see citing works
Real-Time Detection of Parkinson’s States Based on Wearable Sensor With Embedded Machine Learning
Su, F., Chen, M., Zhang, M., Wang, X., Wu, L., Liu, C., Wang, J., Chang, S., & Xin, T. (2024) · IEEE Sensors Journal · doi:10.1109/jsen.2024.3472069 · Cited by 8 · see citing works
A Survey of Embedded Machine Learning for Smart and Sustainable Healthcare Applications
An, S., Tuncel, Y., Başaklar, T., & Ogras, U. Y. (2023) · Springer eBooks · doi:10.1007/978-3-031-40677-5_6 · Cited by 8 · see citing works
IMU-Based Monitoring for Assistive Disease Diagnosis and Management of IoHT: A ReviewPreprint
Bo, F., Yerebakan, M. O., Dai, Y., Wang, W., Li, J., Gao, S., & Boyi, H. (2022) · doi:10.36227/techrxiv.19200482 · Cited by 10 · see citing works
Stefanou, E. E., Bitilis, P., Bouchouras, G., & Kotis, K. (2025) · Applied Sciences · doi:10.3390/app151910541 · Cited by 7 · see citing works
A Survey on Freezing of Gait Detection and Prediction in Parkinson’s Disease
Martínez-Villaseñor, L., Ponce-Espinosa, H., & Miralles‐Pechuán, L. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-60884-2_13 · Cited by 6 · see citing works
FPGA-Based Realtime Detection of Freezing of Gait of Parkinson Patients
Langer, P., Esfahani, A. H., Dyka, Z., & Langendörfer, P. (2022) · Lecture notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering · doi:10.1007/978-3-030-95593-9_9 · Cited by 7 · see citing works
Edge-based freezing of gait recognition in Parkinson’s disease
Borzì, L., Sigcha, L., Firouzi, F., Olmo, G., & Demrozi, F. (2025) · Computers & Electrical Engineering · doi:10.1016/j.compeleceng.2025.110530 · Cited by 5 · see citing works
Self-citationThimmapuram, M., Akepogu, A. R., & Raju, P. R. (2025) · IEEE Transactions on Neural Systems and Rehabilitation Engineering · doi:10.1109/tnsre.2025.3605204 · Cited by 5 · see citing works
Integrative Federated Learning Framework for Multimodal Parkinson’s Disease Biomarker Fusion
Pratihar, R., & Sankar, R. T. (2025) · Computers · doi:10.3390/computers14090388 · Cited by 5 · see citing works
Bibbo, D., Marchis, C. D., Schmid, M., & Ranaldi, S. (2023) · Physiological Measurement · doi:10.1088/1361-6579/ad133b · Cited by 5 · see citing works
Park, H., Shin, S., Youm, C., & Cheon, S. (2024) · Scientific Reports · doi:10.1038/s41598-024-75445-7 · Cited by 5 · see citing works
Murtaza, G., Hammoud, M., & Somov, A. (2025) · IEEE Access · doi:10.1109/access.2025.3585099 · Cited by 7 · see citing works
Khalil, R. M., Shulman, L. M., Gruber‐Baldini, A. L., Shakya, S., Fenderson, R., Hoven, M. V., Hausdorff, J. M., von Coelln, R., & Cummings, M. P. (2024) · Sensors · doi:10.3390/s24154983 · Cited by 5 · see citing works
Gait Phase Recognition using Textile-based Sensor
Pazar, A., Khalilbayli, F., Özlem, K., Yilmaz, A. F., Atalay, A. T., Atalay, Ö., & İnce, G. (2022) · doi:10.1109/ubmk55850.2022.9919491 · Cited by 5 · see citing works
Li, S., Chen, S., Yu, X., Shang, H., Tu, T.-T., & Quan, M. (2026) · Journal of Medical Internet Research · doi:10.2196/85596 · Cited by 4 · see citing works
Shaban, M. (2024) · Machine Learning with Applications · doi:10.1016/j.mlwa.2024.100553 · Cited by 4 · see citing works
Care for the Mind amid Chronic Diseases: An Interpretable AI Approach Using IoT
Xie, J., Zhao, X., Liu, X., & Fang, X. (2025) · Management Science · doi:10.1287/mnsc.2023.04183 · Cited by 4 · see citing works
Multimodel Sensor Based Fog Prediction in Parkinson's Disease: A Deep Learning Approach
Preethi, B. B., Jayanthi, V. E., & Indhumathy, M. (2023) · Innovations in Power and Advanced Computing Technologies (i-PACT) · doi:10.1109/i-pact58649.2023.10434791 · Cited by 4 · see citing works
Parkinson’s Disease Detection Through Deep Learning Model
Bhakar, S., & Verma, S. S. (2022) · Lecture notes in networks and systems · doi:10.1007/978-981-19-5221-0_10 · Cited by 3 · see citing works
Lax-net: Freezing of Gait Detection in Parkinson's Disease Using LSTM with Attention and XGBoost
Ayoub, B., Aicha, K., Taleb, N., & Bouthaina, M. (2023) · International Conference on Decision Aid Sciences and Applications (DASA) · doi:10.1109/dasa59624.2023.10286631 · Cited by 3 · see citing works
Dual-Level Freezing of Gait Recognition Algorithm
Chen, Q., Chen, Z., Zhang, F., Chen, S., Ren, K., & Lu, N. (2025) · IEEE Sensors Journal · doi:10.1109/jsen.2025.3557407 · Cited by 3 · see citing works
IMU-Based Monitoring for Assistive Disease Diagnosis and Management of IoHT: A ReviewPreprint
Bo, F., Yerebakan, M. O., Dai, Y., Wang, W., Li, J., Gao, S., & Boyi, H. (2022) · doi:10.36227/techrxiv.19200482.v1 · Cited by 3 · see citing works
Freezing of Gait detection in Parkinson's Disease: Comparison of Deep Learning Frameworks
Andrei, A.-G., Tăuţan, A.-M., & Ionescu, B. E. (2024) · IEEE International Symposium on Medical Measurements and Applications (MeMeA) · doi:10.1109/memea60663.2024.10596827 · Cited by 2 · see citing works
Attention based 1-D CNN system for detection of Freezing of gait (FOG) in Parkinsons Patients
Kumar, D., Sowmya, B. J., Vincent, A., & Pandiyan, A. (2024) · International Conference on Computing, Communication and Networking Technologies (ICCCNT) · doi:10.1109/icccnt61001.2024.10724148 · Cited by 2 · see citing works
Deep Learning in Medical Robotics for Parkinson's disease Symptom Assessment
Sharma, K., Gori, Y., Deepak, A., Mayuri, K., Mehta, A., & Srivastava, A. (2023) · IEEE Uttar Pradesh Section International Conference on Electrical, Electronics and Computer Engineering (UPCON) · doi:10.1109/upcon59197.2023.10434813 · Cited by 2 · see citing works
Borzì, L., Demrozi, F., Bacchin, R. A., Turetta, C., Tebaldi, M., Sigcha, L., Zolfaghari, S., Rinaldi, D., Fazzina, G., Balestro, G., et al. (2026) · Scientific Data · doi:10.1038/s41597-026-06645-1 · Cited by 2 · see citing works
Self-citationPrediction of Freezing of Gait for Patients with Parkinson’s Disease Using Deep Learning
Shaban, M. (2024) · International Conference on Artificial Intelligence, Blockchain, and Internet of Things (AIBThings) · doi:10.1109/aibthings63359.2024.10863112 · Cited by 1 · see citing works
Multimodal dynamic-static combined graph convolutional network for detecting freezing of gait
Xia, Y., & Cao, A. (2025) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2025.108368 · Cited by 1 · see citing works
Shallow multi-view network for sensor-based recognition of Parkinson’s freezing of gait
Chen, Q., Zhang, F., Chen, S., Ren, K., Chen, S., & Zhao, J. (2025) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2025.108473 · Cited by 1 · see citing works
Birdal, R. G. (2025) · Journal of Mathematical Sciences and Modelling · doi:10.33187/jmsm.1705745 · Cited by 1 · see citing works
A GRU-LSTM Network Bank for Efficient Detection of Freezing of Gait for Parkinson's Disease Patients
Shaban, M. (2025) · IEEE International Conference on Computing and Machine Intelligence (ICMI) · doi:10.1109/icmi65310.2025.11141325 · Cited by 1 · see citing works
Shastry, K. A., & Shastry, A. (2025) · SN Computer Science · doi:10.1007/s42979-025-04500-5 · Cited by 1 · see citing works
Sayem, F. R., Ahmed, M. U., Khalil, H., Pedersen, S., Ali, M. S. M., Hasan, A., Murugappan, M., & Chowdhury, M. E. H. (2026) · International Journal of Data Science and Analytics · doi:10.1007/s41060-026-01031-x · Cited by 1 · see citing works
Bayat, N., Tran, A., Kim, J.-H., & Park, J. (2026) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2026.3657490 · Cited by 1 · see citing works
Freezing of Gait in Parkinson’s Disease: A Scoping Review on the Path Towards Real-Time Therapies
Singhal, M., Grannie, C., Burnette, P., Hernandez, M. E., & Hegazy, S. A. (2026) · Sensors · doi:10.3390/s26072042 · Cited by 1 · see citing works
Pardoel, S. (2021) · UWSpace (University of Waterloo) · Cited by 1 · see citing works
Filtjens, B., Ginis, P., Nieuwboer, A., Slaets, P., & Vanrumste, B. (2021) · arXiv (Cornell University)
- Authors
- Sigcha, L., Costa, N., Pavón, I., Costa, S., Arezes, P., López, J. M., & De Arcas, G.
- Affiliations
- Universidad Politécnica de Madrid (ES)
- University of Minho (PT)
- Grupo de Investigación en Instrumentación y Acústica Aplicada (I2A2), ETSI Industriales, Universidad Politécnica de Madrid, Campus Sur UPM, Ctra. Valencia, Km 7., 28031 Madrid, Spain
- Corresponding author
- Pavón, Ignacio
- ORCID
- Venue
- Sensors
- Publisher
- MDPI AG
- Published
- 2020-03-29
- Volume
- 20
- Issue
- 7
- Pages
- 1895
- ISSN
- 1424-8220 (ISSN-L)
- PMC
- PMC7181252
- OpenAlex
- W3013838736
- Cited by
- 126
- References
- 87
- Citations by year
- 2026: 13 · 2025: 26 · 2024: 20 · 2023: 19 · 2022: 28 · 2021: 16 · 2020: 4
- FWCI
- 30.44
- Type
- journal article
- Language
- en
- Open access
- gold
- dx.doi.org
- Licence
- Locations
- Sensors (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT) (submittedVersion, cc-by)
- Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT) (submittedVersion, other-oa)
- Funders
- Centro Internacional sobre el envejecimiento, CENIE (0348_CIE_6_E)
- Secretaría de Educación Superior, Ciencia, Tecnología e Innovación (Becas internacionales de posgrado 2019)
- Topics
- Health Sciences › Health Professions › Physical Therapy, Sports Therapy and Rehabilitation › Balance, Gait, and Falls Prevention
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Context-Aware Activity Recognition Systems
- Physical Sciences › Engineering › Biomedical Engineering › Gait Recognition and Analysis
- Keywords
- freezing of gait, Parkinson's disease, fog detection, recurrent neural networks, wearable sensors, home monitoring, spectral analysis, gait analysis, leave-one-subject-out cross-validation
- MeSH terms
- Wearable Electronic Devices, Deep Learning, Aged, Aged, 80 and over, Female, Gait, Humans, Male, Monitoring, Physiologic, Parkinson Disease, Signal Processing, Computer-Assisted, Biosensing Techniques, Accelerometry
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-07.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-07.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
