Classification and Analysis of Human Body Movement Characteristics Associated with Acrophobia Induced by Virtual Reality Scenes of Heights
Open accessThe DOI matches its Crossref record; 39 of the 40 references it cites were found in a registry or an index; 1 of the 16 sources its claims cite were checked for support.
Abstract
Acrophobia (fear of heights), a prevalent psychological disorder, elicits profound fear and evokes a range of adverse physiological responses in individuals when exposed to heights, which will lead to a very dangerous state for people in actual heights. In this paper, we explore the behavioral influences in terms of movements in people confronted with virtual reality scenes of extreme heights and develop an acrophobia classification model based on human movement characteristics. To this end, we used wireless miniaturized inertial navigation sensors (WMINS) network to obtain the information of limb movements in the virtual environment. Based on these data, we constructed a series of data feature processing processes, proposed a system model for the classification of acrophobia and non-acrophobia based on human motion feature analysis, and realized the classification recognition of acrophobia and non-acrophobia through the designed integrated learning model. The final accuracy of acrophobia dichotomous classification based on limb motion information reached 94.64%, which has higher accuracy and efficiency compared with other existing research models. Overall, our study demonstrates a strong correlation between people’s mental state during fear of heights and their limb movements at that time.
Overview
CoChat Summary
## TL;DR
The paper builds a virtual-reality height exposure task plus a wearable-sensor pipeline to classify acrophobia from limb motion. Using 43 usable samples and a voting ensemble, it reports up to 94.64% classification accuracy.
## Key findings
- From 31 recruited subjects, the authors obtained 43 usable samples after data cleaning, with 16 labeled acrophobia and 27 non-acrophobia.
- Using 88 motion features and leave-one-out cross validation, the designed voting ensemble achieved 88.37% accuracy on the original feature samples; after K-means SMOTE oversampling, accuracy improved to 91%.
- With feature selection and correlation filtering, the paper reports a best accuracy of 94.63% for the top 80 feature combination, and also states a final acrophobia dichotomous classification accuracy of 94.64% in the abstract.
- The best voting ensemble performance reported in Table 3 was 96.2% precision, 92.8% recall, and 94.5% F-score; for class-specific results, non-fear-of-heights accuracy was 96.3% and fear-of-heights accuracy was 93.1%.
## Methods
- The study used a VR high-altitude scenario with 11 tasks grouped into three stages: rest adaptation, board walking, and jumping; participants walked on a rotating plank, retrieved tennis balls, and completed a jump task.
- Four wireless miniaturized inertial navigation sensors (WMINS) were attached to the left/right wrists and left/right lower tibiae; data were collected at 30 Hz for 100 s.
- The analysis extracted 29 features per data dimension from time and frequency domains, reduced features using p-values and correlation thresholds, handled imbalance with K-means SMOTE, and evaluated models with leave-one-out cross validation using a hard-voting ensemble of KNN, extra trees, random forest, SVM, and logistic regression.
## Limitations
- The paper states the sample was small relative to the feature space: 43 samples versus 3480 extracted feature types, requiring feature selection to reduce overfitting.
- The paper notes class imbalance between acrophobia and non-acrophobia samples, which is why K-means SMOTE was needed.
- Evident from the text: the sample was limited to adults aged 23–30 and to one VR-induced laboratory paradigm, so the reported model’s generalizability outside this setting is not established.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures











Verification
Claims the paper makes
1 Valid · 0 Not valid · 9 Not checked · 54 Own results and interpretations
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Among them, AdaBoost was the worst-performing classification model with an accuracy of only 55.8%.”
§ Classification and Analysis of Human Body Movement Characteristics Associated with Acrophobia Induced by Virtual Reality Scenes of Heights > 3. Results > 3.6. Results and Analysis
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“and the jumping movements appeared discontinuous”
§ Classification and Analysis of Human Body Movement Characteristics Associated with Acrophobia Induced by Virtual Reality Scenes of Heights > 3. Results > 3.1. Feature Analysis We conducted a detailed feature analysis of the preprocessed data. During the experiment, subjects with acrophobia label generally took longer to perform the tasks compared to non-fearful subjects, particularly in T4, where fearful subjects had more hesitation time during the final jumping moments, and the jumping movements appeared discontinuous. In T0, the subjects were in a closed space elevator where the floor number increased to represent the elevator going up. For fearful subjects, some of them may have had fewer limb movements due to their inner fear. In T2, when the elevator door opened, the sudden appearance of a high-altitude environment could give subjects a visual shock, increasing the sense of reality and stimulating the subjects’ belief in the high-altitude environment. Fearful subjects would spend more time building their confidence after the elevator door opened, and most of them would be in a state of external calmness, rather than a natural movement state as observed in non-fearful subjects who could adapt quickly to the environment. In T3, fearful subjects walked more carefully on the wooden board and their walking steps became smaller and slower, resulting in poorer continuity of their walking movements. In addition, their balance will become worse due to the movement of the plank. During the actual experiment, they keep the center of mass and the plank on one side at all times by a slight outward extension of the arms and an up-and-down movement, while the subject will bend slightly to lower the center of mass and increase stability as they move. In contrast, non-fearful subjects had better balance and control over their body movements. Based on the observed behavioral characteristics, we conducted feature analysis on preprocessed data in terms of three-axis acceleration and three-axis angular velocity. Feature extraction was performed in both time and frequency domains to obtain a comprehensive set of features. In the time domain [29], we extracted typical features such as mean, variance, maximum, and minimum values for each data dimension under different task types. In addition, we also obtained specific features such as impulse factor, skewness factor, and peak-to-peak. In total, we extracted 16 features in the time domain. In the frequency domain [30], we extracted 13 different types of features, including frequency center, RMS frequency, and DC component. Ultimately, each data dimension for different task types had a total of 29 features in both time and frequency domains. Table 2 provides detailed information on these 29 features. Finally, we split and combined the feature data according to the joint positions specified by WMINS, resulting in 4 joint positions with 5 × 29 × 6 feature matrices. After feature data segmentation and solving, we obtained 3480 different feature types. However, the number of samples was only 43, much less than the number of feature types. To prevent overfitting in the subsequent classification model, we performed feature selection using p-values, a statistical method [31]. p-value is a probability value obtained from statistical analysis, which represents the probability of observing the results or more extreme results under the null hypothesis. When the p-value is less than the set significance level (e.g., 0.05), the null hypothesis is rejected, and the observed results are considered significant, indicating that the differences observed in the sample may not be due to chance. When the p-value is greater than the set significance level, the null hypothesis cannot be rejected. Here, we have labeled the acrophobia group “0” and the non-acrophobia group “1”. Similarly, the features solved for the acrophobia group will be labelled “0” and the features solved for the non-acrophobia group will be labelled “1”, and finally, the p-values for each class of features will be solved directly for that class of features based on these feature values and the corresponding labels. We sorted the WMINS under each joint in ascending order according to their p-values and selected the features whose p-values were less than 0.02. We finally obtained 88 data features and replaced the original motion data with these features to represent the motion state of each subject. These 88 features were used as the original feature data set for building the classification model.
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“During the actual experiment, they keep the center of mass and the plank on one side at all times by a slight outward extension of the arms and an up-and-down movement, while the subject will bend slightly to lower the center of mass and increase stability as they move.”
§ Classification and Analysis of Human Body Movement Characteristics Associated with Acrophobia Induced by Virtual Reality Scenes of Heights > 3. Results > 3.1. Feature Analysis We conducted a detailed feature analysis of the preprocessed data. During the experiment, subjects with acrophobia label generally took longer to perform the tasks compared to non-fearful subjects, particularly in T4, where fearful subjects had more hesitation time during the final jumping moments, and the jumping movements appeared discontinuous. In T0, the subjects were in a closed space elevator where the floor number increased to represent the elevator going up. For fearful subjects, some of them may have had fewer limb movements due to their inner fear. In T2, when the elevator door opened, the sudden appearance of a high-altitude environment could give subjects a visual shock, increasing the sense of reality and stimulating the subjects’ belief in the high-altitude environment. Fearful subjects would spend more time building their confidence after the elevator door opened, and most of them would be in a state of external calmness, rather than a natural movement state as observed in non-fearful subjects who could adapt quickly to the environment. In T3, fearful subjects walked more carefully on the wooden board and their walking steps became smaller and slower, resulting in poorer continuity of their walking movements. In addition, their balance will become worse due to the movement of the plank. During the actual experiment, they keep the center of mass and the plank on one side at all times by a slight outward extension of the arms and an up-and-down movement, while the subject will bend slightly to lower the center of mass and increase stability as they move. In contrast, non-fearful subjects had better balance and control over their body movements. Based on the observed behavioral characteristics, we conducted feature analysis on preprocessed data in terms of three-axis acceleration and three-axis angular velocity. Feature extraction was performed in both time and frequency domains to obtain a comprehensive set of features. In the time domain [29], we extracted typical features such as mean, variance, maximum, and minimum values for each data dimension under different task types. In addition, we also obtained specific features such as impulse factor, skewness factor, and peak-to-peak. In total, we extracted 16 features in the time domain. In the frequency domain [30], we extracted 13 different types of features, including frequency center, RMS frequency, and DC component. Ultimately, each data dimension for different task types had a total of 29 features in both time and frequency domains. Table 2 provides detailed information on these 29 features. Finally, we split and combined the feature data according to the joint positions specified by WMINS, resulting in 4 joint positions with 5 × 29 × 6 feature matrices. After feature data segmentation and solving, we obtained 3480 different feature types. However, the number of samples was only 43, much less than the number of feature types. To prevent overfitting in the subsequent classification model, we performed feature selection using p-values, a statistical method [31]. p-value is a probability value obtained from statistical analysis, which represents the probability of observing the results or more extreme results under the null hypothesis. When the p-value is less than the set significance level (e.g., 0.05), the null hypothesis is rejected, and the observed results are considered significant, indicating that the differences observed in the sample may not be due to chance. When the p-value is greater than the set significance level, the null hypothesis cannot be rejected. Here, we have labeled the acrophobia group “0” and the non-acrophobia group “1”. Similarly, the features solved for the acrophobia group will be labelled “0” and the features solved for the non-acrophobia group will be labelled “1”, and finally, the p-values for each class of features will be solved directly for that class of features based on these feature values and the corresponding labels. We sorted the WMINS under each joint in ascending order according to their p-values and selected the features whose p-values were less than 0.02. We finally obtained 88 data features and replaced the original motion data with these features to represent the motion state of each subject. These 88 features were used as the original feature data set for building the classification model.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“During the experiment, subjects with acrophobia label generally took longer to perform the tasks compared to non-fearful subjects”
§ Classification and Analysis of Human Body Movement Characteristics Associated with Acrophobia Induced by Virtual Reality Scenes of Heights > 3. Results > 3.1. Feature Analysis We conducted a detailed feature analysis of the preprocessed data. During the experiment, subjects with acrophobia label generally took longer to perform the tasks compared to non-fearful subjects, particularly in T4, where fearful subjects had more hesitation time during the final jumping moments, and the jumping movements appeared discontinuous. In T0, the subjects were in a closed space elevator where the floor number increased to represent the elevator going up. For fearful subjects, some of them may have had fewer limb movements due to their inner fear. In T2, when the elevator door opened, the sudden appearance of a high-altitude environment could give subjects a visual shock, increasing the sense of reality and stimulating the subjects’ belief in the high-altitude environment. Fearful subjects would spend more time building their confidence after the elevator door opened, and most of them would be in a state of external calmness, rather than a natural movement state as observed in non-fearful subjects who could adapt quickly to the environment. In T3, fearful subjects walked more carefully on the wooden board and their walking steps became smaller and slower, resulting in poorer continuity of their walking movements. In addition, their balance will become worse due to the movement of the plank. During the actual experiment, they keep the center of mass and the plank on one side at all times by a slight outward extension of the arms and an up-and-down movement, while the subject will bend slightly to lower the center of mass and increase stability as they move. In contrast, non-fearful subjects had better balance and control over their body movements. Based on the observed behavioral characteristics, we conducted feature analysis on preprocessed data in terms of three-axis acceleration and three-axis angular velocity. Feature extraction was performed in both time and frequency domains to obtain a comprehensive set of features. In the time domain [29], we extracted typical features such as mean, variance, maximum, and minimum values for each data dimension under different task types. In addition, we also obtained specific features such as impulse factor, skewness factor, and peak-to-peak. In total, we extracted 16 features in the time domain. In the frequency domain [30], we extracted 13 different types of features, including frequency center, RMS frequency, and DC component. Ultimately, each data dimension for different task types had a total of 29 features in both time and frequency domains. Table 2 provides detailed information on these 29 features. Finally, we split and combined the feature data according to the joint positions specified by WMINS, resulting in 4 joint positions with 5 × 29 × 6 feature matrices. After feature data segmentation and solving, we obtained 3480 different feature types. However, the number of samples was only 43, much less than the number of feature types. To prevent overfitting in the subsequent classification model, we performed feature selection using p-values, a statistical method [31]. p-value is a probability value obtained from statistical analysis, which represents the probability of observing the results or more extreme results under the null hypothesis. When the p-value is less than the set significance level (e.g., 0.05), the null hypothesis is rejected, and the observed results are considered significant, indicating that the differences observed in the sample may not be due to chance. When the p-value is greater than the set significance level, the null hypothesis cannot be rejected. Here, we have labeled the acrophobia group “0” and the non-acrophobia group “1”. Similarly, the features solved for the acrophobia group will be labelled “0” and the features solved for the non-acrophobia group will be labelled “1”, and finally, the p-values for each class of features will be solved directly for that class of features based on these feature values and the corresponding labels. We sorted the WMINS under each joint in ascending order according to their p-values and selected the features whose p-values were less than 0.02. We finally obtained 88 data features and replaced the original motion data with these features to represent the motion state of each subject. These 88 features were used as the original feature data set for building the classification model.
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In contrast, non-fearful subjects had better balance and control over their body movements.”
§ Classification and Analysis of Human Body Movement Characteristics Associated with Acrophobia Induced by Virtual Reality Scenes of Heights > 3. Results > 3.1. Feature Analysis We conducted a detailed feature analysis of the preprocessed data. During the experiment, subjects with acrophobia label generally took longer to perform the tasks compared to non-fearful subjects, particularly in T4, where fearful subjects had more hesitation time during the final jumping moments, and the jumping movements appeared discontinuous. In T0, the subjects were in a closed space elevator where the floor number increased to represent the elevator going up. For fearful subjects, some of them may have had fewer limb movements due to their inner fear. In T2, when the elevator door opened, the sudden appearance of a high-altitude environment could give subjects a visual shock, increasing the sense of reality and stimulating the subjects’ belief in the high-altitude environment. Fearful subjects would spend more time building their confidence after the elevator door opened, and most of them would be in a state of external calmness, rather than a natural movement state as observed in non-fearful subjects who could adapt quickly to the environment. In T3, fearful subjects walked more carefully on the wooden board and their walking steps became smaller and slower, resulting in poorer continuity of their walking movements. In addition, their balance will become worse due to the movement of the plank. During the actual experiment, they keep the center of mass and the plank on one side at all times by a slight outward extension of the arms and an up-and-down movement, while the subject will bend slightly to lower the center of mass and increase stability as they move. In contrast, non-fearful subjects had better balance and control over their body movements. Based on the observed behavioral characteristics, we conducted feature analysis on preprocessed data in terms of three-axis acceleration and three-axis angular velocity. Feature extraction was performed in both time and frequency domains to obtain a comprehensive set of features. In the time domain [29], we extracted typical features such as mean, variance, maximum, and minimum values for each data dimension under different task types. In addition, we also obtained specific features such as impulse factor, skewness factor, and peak-to-peak. In total, we extracted 16 features in the time domain. In the frequency domain [30], we extracted 13 different types of features, including frequency center, RMS frequency, and DC component. Ultimately, each data dimension for different task types had a total of 29 features in both time and frequency domains. Table 2 provides detailed information on these 29 features. Finally, we split and combined the feature data according to the joint positions specified by WMINS, resulting in 4 joint positions with 5 × 29 × 6 feature matrices. After feature data segmentation and solving, we obtained 3480 different feature types. However, the number of samples was only 43, much less than the number of feature types. To prevent overfitting in the subsequent classification model, we performed feature selection using p-values, a statistical method [31]. p-value is a probability value obtained from statistical analysis, which represents the probability of observing the results or more extreme results under the null hypothesis. When the p-value is less than the set significance level (e.g., 0.05), the null hypothesis is rejected, and the observed results are considered significant, indicating that the differences observed in the sample may not be due to chance. When the p-value is greater than the set significance level, the null hypothesis cannot be rejected. Here, we have labeled the acrophobia group “0” and the non-acrophobia group “1”. Similarly, the features solved for the acrophobia group will be labelled “0” and the features solved for the non-acrophobia group will be labelled “1”, and finally, the p-values for each class of features will be solved directly for that class of features based on these feature values and the corresponding labels. We sorted the WMINS under each joint in ascending order according to their p-values and selected the features whose p-values were less than 0.02. We finally obtained 88 data features and replaced the original motion data with these features to represent the motion state of each subject. These 88 features were used as the original feature data set for building the classification model.
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
22 more claims
Create a free account to see all 64 claims and to verify the 15 references still open.
- It may state two separate claims as one. 29
- Probably not a substantial claim. 9
- The quote only partly states this claim. 2
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?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Acrophobia and claustrophobia are two common distressing anxiety disorders that can cause suffering and functional impairment, compromising a person’s quality of life [2].”
From SAFEvR MentalVeRse.app: Development of a Free Immersive Virtual Reality Exposure Therapy for Acrophobia and Claustrophobia (2024) · cites this paper as [2]
Experimental — accuracy not yet measured
Citations
Cites
Biau, G., & Scornet, E. (2016) · Test · doi:10.1007/s11749-016-0481-7 · Cited by 4,149 · cited 1×
“We used KNN [34], extra trees, random forest [35], SVM [36], and logistic regression [37] as the five classifiers for model ensemble.”
Improving imbalanced learning through a heuristic oversampling method based on k-means and SMOTE
Douzas, G., Bacao, F., & Last, F. (2018) · Information Sciences · doi:10.1016/j.ins.2018.06.056 · Cited by 1,277 · cited 1×
“As there was an imbalance in the sample data between non-acrophobia and acrophobia samples, which could lead to a decrease in model generalization and prediction accuracy, K-means smote was used to oversample the sample data [32].”
Andrade, C. (2019) · Indian Journal of Psychological Medicine · doi:10.4103/ijpsym.ijpsym_193_19 · Cited by 478 · cited 1×
“To prevent overfitting in the subsequent classification model, we performed feature selection using p -values, a statistical method [31]. p -value is a probability value obtained from statistical analysis, which represents the probability of observing the results or more extreme results under the…”
Artificial neural network based epileptic detection using time-domain and frequency-domain features
Srinivasan, V., Eswaran, C., & Sriraam, A. N. (2005) · Journal of Medical Systems · doi:10.1007/s10916-005-6133-1 · Cited by 417
Freeman, D., Haselton, P., Freeman, J., Spanlang, B., Kishore, S., Albery, E., Denne, M., Brown, P., Slater, M., & Nickless, A. (2018) · The Lancet Psychiatry · doi:10.1016/s2215-0366(18)30226-8 · Cited by 377 · cited 1×
“…axes, and calculated three-axis combined acceleration and angular velocity. before and after the experiment to label the subjects, those with HIQ scores greater than 29 were classi fi ed as the acrophobia group and those with scores less than 29 were classi fi ed as the non-acrophobia group [28].”
A multibranch CNN-BiLSTM model for human activity recognition using wearable sensor data
Challa, S. K., Kumar, A., & Semwal, V. B. (2022) · The Visual Computer · doi:10.1007/s00371-021-02283-3 · Cited by 338 · cited 1×
“…direct analysis is performed and a classification model [3] is built to enable activity recognition [4,5] for everyday activities such as walking, running, and walking up and down stairs [6], as well as more in-depth gait analysis [7], fall detection [8], and biomedical information detection [9,10].”
Redefining therapeutic success with virtual reality exposure therapy
Garcia-Palacios, A., Hoffman, H. G., Kwong See, S., Tsai, A., & Botella, C. (2001) · CyberPsychology & Behavior · doi:10.1089/109493101300210231 · Cited by 212 · cited 1×
“Patients immerse themselves in a computer-generated virtual environment that can provide various scenarios that are diffi-cult to create in the real world [18].”
AMES, S. L., WOLFFSOHN, J. S., & MCBRIEN, N. A. (2005) · Optometry and Vision Science · doi:10.1097/01.opx.0000156307.95086.6 · Cited by 188
Fault diagnosis of rolling element bearing using time-domain features and neural networks
Sreejith, B., Verma, A. K., & Srividya, A. (2008) · IEEE International Conference on Industrial and Information Systems (ICIIS) · doi:10.1109/iciinfs.2008.4798444 · Cited by 142 · cited 1×
“In the time doma [29], we extracted typical features such as mean, va ance, maximu , and minimum valu s for each data dimension under differe t task types.”
Anxiety level recognition for virtual reality therapy system using physiological signals
Šalkevicius, J., Damaševičius, R., Maskeliunas, R., & Laukienė, I. (2019) · Electronics · doi:10.3390/electronics8091039 · Cited by 135 · cited 1×
“[42] GSR, BVP, skin temperature SVM 4 86.”
The Mechanism of Physiological Height Vertigo: I. Theoretical Approach and Psychophysics
Brandt, T., Arnold, F., Bles, W., & Kapteyn, T. S. (1980) · Acta Oto-Laryngologica · doi:10.3109/00016488009127169 · Cited by 118 · cited 1×
“Specifically, acrophobia patients tend to exhibit more cautious, defensive postures such as hunching and neck retraction in high places [12,13], while individuals without acrophobia stand or walk more relaxed, confident, and naturally.”
The Detection of Counterfeit Banknotes Using Ensemble Learning Techniques of AdaBoost and Voting
Babylon, D. G. O. E. I., Khairy, R. S., Hussein, A., Babylon, D. G. O. E. I., & ALRikabi, H. T. S. (2020) · International journal of intelligent engineering and systems · doi:10.22266/ijies2021.0228.31 · Cited by 90 · cited 1×
“Voting is an ensemble learning-based classification algo-rithm [33] that uses a majority vote to determine the final classification.”
Kalantari, S., & Shepley, M. M. (2020) · Housing Studies · doi:10.1080/02673037.2020.1752630 · Cited by 89 · cited 1×
“With the development of society, more and more skyscrapers have emerged [14], creating more opportunities for people to experience high altitude.”
Human proprioceptive adaptations during states of height-induced fear and anxiety
Davis, J. R., Horslen, B. C., Nishikawa, K., Fukushima, K., Chua, R., Inglis, J. T., & Carpenter, M. G. (2011) · Journal of Neurophysiology · doi:10.1152/jn.01030.2010 · Cited by 83 · cited 1×
“Due to the expanded visual range, high-altitude environments tend to cause vertigo and affect people’s control of their posture [20–22].”
Bălan, O., Moise, G., Moldoveanu, A., Leordeanu, M., & Moldoveanu, F. (2020) · Sensors · doi:10.3390/s20020496 · Cited by 82
Down on heights? One in three has visual height intolerance
Huppert, D., Grill, E., & Brandt, T. (2012) · Journal of Neurology · doi:10.1007/s00415-012-6685-1 · Cited by 82 · cited 1×
“Specifically, acrophobia patients tend to exhibit more cautious, defensive postures such as hunching and neck retraction in high places [12,13], while individuals without acrophobia stand or walk more relaxed, confident, and naturally.”
Data Transmission Efficiency in Bluetooth Low Energy Versions
Bulić, P., Kojek, G., & Biasizzo, A. (2019) · Sensors · doi:10.3390/s19173746 · Cited by 79 · cited 2×
“…technology for data transmission and reception [25]. comprises two major functional modules—a motion collection module for collecting the motion information of the a tt ached object and a data transmission module that uses low-power Bluetooth technology for data transmission and re-ception [25].”
Cognitive processing and acrophobia: validating the Heights Interpretation Questionnaire
Steinman, S. A., & Teachman, B. A. (2011) · Journal of Anxiety Disorders · doi:10.1016/j.janxdis.2011.05.001 · Cited by 69 · cited 1×
“We utilized two assessment scales, the virtual reality scene quality assessment (VRSQ) [26] and the height interpretation questionnaire (HIQ) [27], to evaluate the subjects’ response to specific VR environments and their level of acrophobia, respectively.”
Acrophobia impairs visual exploration and balance during standing and walking
Brandt, T., Kugler, G., Schniepp, R., Wuehr, M., & Huppert, D. (2015) · Annals of the New York Academy of Sciences · doi:10.1111/nyas.12692 · Cited by 60 · cited 1×
“Due to the expanded visual range, high-altitude environments tend to cause vertigo and affect people’s control of their posture [20–22].”
Wilson, R. B., Vangala, S., Elashoff, D., Safari, T., & Smith, B. A. (2021) · Sensors · doi:10.3390/s21020616 · Cited by 59 · cited 1×
“…motion acceleration data and angular velocity data to explore the relationship between motion complexity and developmental outcomes in infants at high familial risk for autism spectrum disorder (ASD) (HR infants) [1], and to develop a classification model between HR infants and normal infants.”
TransPose: Real-time 3D Human Translation and Pose Estimation with Six Inertial SensorsPreprint
Yi, X., Zhou, Y., & Xu, F. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2105.04605 · Cited by 49
Linear and nonlinear analyses of heart rate variability signals under mental load
Hao, T., Zheng, X., Wang, H., Xu, K., & Chen, S. (2022) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2022.103758 · Cited by 48 · cited 1×
“Specifically, we computed the cross-correlation coefficients [40] between different features and determined a series of threshold values to remove redundant features.”
Rodrigues, J., Liu, H., Folgado, D., Belo, D., Schultz, T., & Gamboa, H. F. S. (2022) · Biosensors · doi:10.3390/bios12121182 · Cited by 47 · cited 1×
“…direct analysis is performed and a classification model [3] is built to enable activity recognition [4,5] for everyday activities such as walking, running, and walking up and down stairs [6], as well as more in-depth gait analysis [7], fall detection [8], and biomedical information detection [9,10].”
Hüweler, R., Kandil, F. I., Alpers, G. W., & Gerlach, A. L. (2009) · Behaviour Research and Therapy · doi:10.1016/j.brat.2009.01.011 · Cited by 44
Bzdúšková, D., Marko, M., Hirjaková, Z., Kimijanová, J., Hlavačka, F., & Riečanský, I. (2022) · Frontiers in Human Neuroscience · doi:10.3389/fnhum.2021.773091 · Cited by 40
Effectiveness of Virtual Environment Desensitization in the Treatment of Agoraphobia
North, M. M., North, S. M., & Coble, J. R. (1995) · International Journal of Virtual Reality · doi:10.20870/ijvr.1995.1.2.2603 · Cited by 37
Hidden Markov Model and Its Application in Human Activity Recognition and Fall Detection: A Review
Xue, T., & Liu, H. (2022) · Lecture notes in electrical engineering · doi:10.1007/978-981-19-0390-8_108 · Cited by 37
Wright, B., Tindall, L., Scott, A. J., Lee, E. C., Cooper, C. L., Biggs, K., Bee, P., Wang, H., Gega, L., Hayward, E., et al. (2022) · Journal of Child Psychology and Psychiatry · doi:10.1111/jcpp.13665 · Cited by 37 · cited 1×
“Cognitive-behavioral therapy involves face-to-face communication between a psychotherapist and the patient for assessment and treatment [15,16].”
Liu, H., Gamboa, H. F. S., & Schultz, T. (2022) · Sensors · doi:10.3390/s23010125 · Cited by 37 · cited 1×
“…the human wrist and ankle, or from the accelerometer data that comes with the mobile phone, direct analysis is performed and a classification model [3] is built to enable activity recognition [4,5] for everyday activities such as walking, running, and walking up and down stairs [6], as well as…”
Muñoz-Organero, M., Powell, L., Heller, B., Harpin, V., & Parker, J. (2019) · Sensors · doi:10.3390/s19132935 · Cited by 36 · cited 1×
“Statistical analysis of motor acceleration signals at the wrist and ankle in children with normal development and children with attention deficit hyperactivity disorder (ADHD) [2] suggests that the two have different statistical properties of behavior.”
The direction of postural threat alters balance control when standing at virtual elevation
Raffegeau, T. E., Fawver, B. J., Young, W. R., Williams, A. M., Lohse, K. R., & Fino, P. C. (2020) · Experimental Brain Research · doi:10.1007/s00221-020-05917-5 · Cited by 33 · cited 1×
“In addition, the amplitude of body sway decreases with increasing virtual height [24].”
Liu, H., & Schultz, T. (2022) · International Conference on Health Informatics · doi:10.5220/0010896400003123 · Cited by 33 · cited 1×
“…direct analysis is performed and a classification model [3] is built to enable activity recognition [4,5] for everyday activities such as walking, running, and walking up and down stairs [6], as well as more in-depth gait analysis [7], fall detection [8], and biomedical information detection [9,10].”
On a Real Real-Time Wearable Human Activity Recognition System
Liu, H., Xue, T., & Schultz, T. (2023) · International Conference on Health Informatics · doi:10.5220/0011927700003414 · Cited by 33 · cited 1×
“…data that comes with the mobile phone, direct analysis is performed and a classification model [3] is built to enable activity recognition [4,5] for everyday activities such as walking, running, and walking up and down stairs [6], as well as more in-depth gait analysis [7], fall detection…”
TSSEARCH: Time Series Subsequence Search Library
Folgado, D., Barandas, M., Antunes, M., Nunes, M. L., Liu, H., Hartmann, Y., Schultz, T., & Gamboa, H. F. S. (2022) · SoftwareX · doi:10.1016/j.softx.2022.101049 · Cited by 32 · cited 1×
“…direct analysis is performed and a classification model [3] is built to enable activity recognition [4,5] for everyday activities such as walking, running, and walking up and down stairs [6], as well as more in-depth gait analysis [7], fall detection [8], and biomedical information detection [9,10].”
Interactive and Interpretable Online Human Activity Recognition
Hartmann, Y., Liu, H., & Schultz, T. (2022) · IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops) · doi:10.1109/percomworkshops53856.2022.9767207 · Cited by 28 · cited 1×
“With the help of motion data from inertial guidance sensors at the human wrist and ankle, or from the accelerometer data that comes with the mobile phone, direct analysis is performed and a classification model [3] is built to enable activity recognition [4,5] for everyday activities such as walking, running, and walk…”
Research on the characteristics of acrophobia in virtual altitude environment
Hu, F., Wang, H., Chen, J., & Gong, J. (2018) · IEEE International Conference on Intelligence and Safety for Robotics (ISR) · doi:10.1109/iisr.2018.8535774 · Cited by 20
Multi-modal physiological signals based fear of heights analysis in virtual reality scenes
Zheng, R., Wang, T., Cao, J., Vidal, P., & Wang, D. (2021) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2021.102988 · Cited by 18
Recognition of public speaking anxiety on the recurrence quantification analysis of GSR signals
Zhang, X., Wen, W., Liu, G., & Hu, H. (2016) · International Conference on Information Science and Technology · doi:10.1109/icist.2016.7483472 · Cited by 16
A Systematic Literature Review: Acrophobia Treatment with Virtual Reality
Giraldy, D. J., & Novaldo, W. (2022) · Engineering MAthematics and Computer Science (EMACS) Journal · doi:10.21512/emacsjournal.v4i1.8077 · Cited by 8 · cited 1×
“The second method is virtual reality exposure therapy (VRET) [17], a behavioral therapy used for anxiety disorders, including phobias.”
Research works printed but not linked: 4
- Heimberg, R.G., and Juster, H.R. (1995). Social Phobia: Diagnosis, Assessment, and Treatment, The Guilford Press.
- Guo, G., Wang, H., Bell, D., Bi, Y., and Greer, K. (2003). On The Move to Meaningful Internet Systems 2003: CoopIS, DOA, and ODBASE: OTM Confederated International Conferences, CoopIS, DOA, and ODBASE 2003, Catania, Sicily, Italy, November 3–7, 2003. Proceedings, Springer.
- Jakkula, 2006, Tutorial on support vector machine (svm), Sch. EECS Wash. State Univ., 37, 3
- Powers, D.M. (2020). Evaluation: From Precision, Recall and F-Measure to ROC, Informedness, Markedness and Correlation. arXiv.
Cited by
Human Pose Estimation and Event Recognition via Feature Extraction and Neuro-Fuzzy Classifier
Hanzla, M., Alshammari, N. S., Alharbi, S. S., Wahid, W., Almujally, N. A., Jalal, A., & Liu, H. (2025) · IEEE Access · doi:10.1109/access.2025.3538691 · Cited by 22
“Cheng et al. [12] proposed a novel HAR system integrating virtual reality (VR) data with traditional sensing methods to classify movement characteristics, particularly in phobia scenarios like acrophobia.”
Gaina, M.-A., Sbarcea, S.-V., Popa, B.-S., Stefanescu, B.-V., Gaina, A.-M., Szalontay, A.-S., Bolos, A., & Stefanescu, C. (2024) · Brain Sciences · doi:10.3390/brainsci14070651 · Cited by 9
“Acrophobia and claustrophobia are two common distressing anxiety disorders that can cause suffering and functional impairment, compromising a person’s quality of life [2].”
BackgroundArtificial intelligence (AI) for virtual reality exposure therapy (VRET): A systematic review
Bergsnev, K., & Laws, A. L. S. (2026) · Translational Psychiatry · doi:10.1038/s41398-026-03936-4 · Cited by 5
The role of different sensory stimuli in an at height simulation
Shah, D., & Lawson, G. (2025) · Applied Ergonomics · doi:10.1016/j.apergo.2025.104559 · Cited by 4
VR Journey: Assessing Affective State Through Kinesics
Zhukova, M., Brueva, A., Demarev, A., Zaladina, A., Petrova, I. E., & Демарева, В. (2024) · Studies in computational intelligence · doi:10.1007/978-3-031-76516-2_46
Metadata
- Authors
- Cheng, X., Bao, B., Cui, W., Liu, S., Zhong, J., Cai, L., & Yang, H.
- Affiliations
- University of Science and Technology of China (CN)
- Chinese Academy of Sciences (CN)
- Corresponding author
- Yang, Hongbo
- Venue
- Sensors
- Publisher
- MDPI AG
- Published
- 2023-06-10
- Volume
- 23
- Issue
- 12
- Pages
- 5482
- ISSN
- 1424-8220 (ISSN-L)
- PMC
- PMC10305026
- OpenAlex
- W4380263719
- Cited by
- 7
- References
- 40
- Citations by year
- 2026: 2 · 2025: 3 · 2024: 2
- FWCI
- 1.21
- Type
- journal article
- Language
- en
- Open access
- gold
- 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)
- Funders
- National Key R&D Program of China (2022YFC2405603, BE2022064-2)
- Key R& D Program of Jiangsu Province (2022YFC2405603, BE2022064-2)
- Topics
- Physical Sciences › Computer Science › Human-Computer Interaction › Virtual Reality Applications and Impacts
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Context-Aware Activity Recognition Systems
- Physical Sciences › Computer Science › Artificial Intelligence › Emotion and Mood Recognition
- Keywords
- acrophobia, fear of heights, virtual reality, limb motion, inertial navigation sensors, psychological state, fear response
- MeSH terms
- Virtual Reality, Fear, Humans, Phobic Disorders, Human Body
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.
A language model drew the claims from the paper on 2026-10-11, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-11 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
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
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.