Temporal convolutional networks and data rebalancing for clinical length of stay and mortality prediction
Open accessAbstract
It is critical for hospitals to accurately predict patient length of stay (LOS) and mortality in real-time. We evaluate temporal convolutional networks (TCNs) and data rebalancing methods to predict LOS and mortality. This is a retrospective cohort study utilizing the MIMIC-III database. The MIMIC-Extract pipeline processes 24 hour time-series clinical objective data for 23,944 unique patient records. TCN performance is compared to both baseline and state-of-the-art machine learning models including logistic regression, random forest, gated recurrent unit with decay (GRU-D). Models are evaluated for binary classification tasks (LOS > 3 days, LOS > 7 days, mortality in-hospital, and mortality in-ICU) with and without data rebalancing and analyzed for clinical runtime feasibility. Data is split temporally, and evaluations utilize tenfold cross-validation (stratified splits) followed by simulated prospective hold-out validation. In mortality tasks, TCN outperforms baselines in 6 of 8 metrics (area under receiver operating characteristic, area under precision-recall curve (AUPRC), and F-1 measure for in-hospital mortality; AUPRC, accuracy, and F-1 for in-ICU mortality). In LOS tasks, TCN performs competitively to the GRU-D (best in 6 of 8) and the random forest model (best in 2 of 8). Rebalancing improves predictive power across multiple methods and outcome ratios. The TCN offers strong performance in mortality classification and offers improved computational efficiency on GPU-enabled systems over popular RNN architectures. Dataset rebalancing can improve model predictive power in imbalanced learning. We conclude that temporal convolutional networks should be included in model searches for critical care outcome prediction systems.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
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?
TCNs have demonstrated efficacy in capturing temporal dependencies across diverse applications.
“TCNs have demonstrated efficacy in capturing temporal dependencies across diverse applications. For example, Lea et al. [51] utilized a TCN for action segmentation and detection, Zhang et al. [52] applied it for text classification, and Bednarski et al. [53] employed it in health informatics for predicting clinical length of stay and mortality.”
From Reducing Label Dependency in Human Activity Recognition with Wearables: From Supervised Learning to Novel Weakly Self-Supervised Approaches (2025) · cites this paper as [53]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
Hochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318 · see citing works
SMOTE synthetic minority over-sampling technique
Chawla, N. V., Bowyer, K. W., Hall, L. O., & Kegelmeyer, W. P. (2002) · Journal of Artificial Intelligence Research · doi:10.1613/jair.953 · Cited by 32,865 · see citing works
An introduction to ROC analysis
Fawcett, T. (2006) · Pattern Recognition Letters · doi:10.1016/j.patrec.2005.10.010 · Cited by 21,879 · see citing works
Goldberger, A. L., Amaral, L. A. N., Glass, L., Hausdorff, J. M., Ivanov, P. C., Mark, R. G., Mietus, J. E., Moody, G. B., Peng, C.-K., & Stanley, H. E. (2000) · Circulation · doi:10.1161/01.cir.101.23.e215 · Cited by 14,809 · see citing works
Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling
Chung, J., Gulcehre, C., Cho, K., & Bengio, Y. (2014) · arXiv · doi:10.48550/arxiv.1412.3555 · Cited by 11,008 · cited 1× · see citing works
“The GRU-D model is a recurrent neural network (RNN) architecture (similar to the long short-term memory network [LSTM]) 49 .”
MethodMIMIC-III, a freely accessible critical care database
Johnson, A. E. W., Pollard, T. J., Shen, L., Lehman, L.-W. H., Feng, M., Ghassemi, M., Moody, B., Szolovits, P., Anthony Celi, L., & Mark, R. G. (2016) · Scientific Data · doi:10.1038/sdata.2016.35 · Cited by 8,141 · see citing works
17. A Value for n-Person Games
Shapley, L. S. (1953) · Contributions to the Theory of Games (AM-28), Volume II · doi:10.1515/9781400881970-018 · Cited by 6,950 · see citing works
The relationship between Precision-Recall and ROC curves
Davis, J., & Goadrich, M. (2006) · Proceedings of the 23rd international conference on Machine learning - ICML '06 · doi:10.1145/1143844.1143874 · Cited by 6,559 · cited 1× · see citing works
“AUPRC is considered as better evaluation metric for imbalanced datasets compared to AUROC, as it directly includes false-positive (FP) and false-negative (FN) predictions in its evaluation 52 .”
MethodRehospitalizations among Patients in the Medicare Fee-for-Service Program
Jencks, S. F., Williams, M. V., & Coleman, E. A. (2009) · New England Journal of Medicine · doi:10.1056/nejmsa0803563 · Cited by 5,164 · see citing works
ADASYN: Adaptive synthetic sampling approach for imbalanced learning
He, H., Bai, Y., Garcia, E. A., & Shutao Li, A. (2008) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn.2008.4633969 · Cited by 4,945 · cited 1× · see citing works
“There are numerous synthetic oversampling techniques presented in the literature, including the Synthetic Minority Oversampling Technique (SMOTE) 29,30 and the adaptive synthetic (ADASYN) 34 classes of solutions.”
MethodAn Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling
Bai, S., Kolter, J. Z., & Koltun, V. (2018) · arXiv · doi:10.48550/arxiv.1803.01271 · Cited by 4,655 · cited 5× · see citing works
“However, the hierarchical architecture of TCNs can capture spatio-temporal information simultaneously with a high degree of parallelism, making them favorable in the applications of graphics processing unit (GPU) to AI applications 19–24 .”
BackgroundMethodInfluentialBorderline-SMOTE : A New Over-Sampling Method in Imbalanced Data Sets Learning
Han, H., Wang, W.-Y., & Mao, B.-H. (2005) · Lecture notes in computer science · doi:10.1007/11538059_91 · Cited by 4,331 · cited 1× · see citing works
“…• Random (majority) under-sampling (no synthetic data) 53 • Random (minority) over-sampling (duplicate data) 54 • Synthetic Minority Oversampling Technique (SMOTE) (synthetic data) 30,54 • Borderline (BL) SMOTE (synthetic data) 54,55 • Support vector machine (SVM) SMOTE (synthetic data) 54,56”
Recurrent Neural Networks for Multivariate Time Series with Missing Values
Che, Z., Purushotham, S., Cho, K., Sontag, D., & Liu, Y. (2018) · Scientific Reports · doi:10.1038/s41598-018-24271-9 · Cited by 2,370 · cited 1× · see citing works
“GRU-D was selected over LSTM because it was demonstrated as a state-of-the-art for this dataset in 41 and to outperform LSTM for MIMIC data 50 .”
MethodSMOTE for Learning from Imbalanced Data: Progress and Challenges, Marking the 15-year Anniversary
Fernández, A., García, S., Herrera, F., & Chawla, N. V. (2018) · Journal of Artificial Intelligence Research · doi:10.1613/jair.1.11192 · Cited by 2,234 · cited 6× · see citing works
“SMOTEBoost and RUSBoot are examples of ensemble rebalancing methods 30,37 .”
BackgroundMethodInfluentialA Novel Connectionist System for Unconstrained Handwriting Recognition
Graves, A., Liwicki, M., Fernandez, S., Bertolami, R., Horst Bunke, A., & Schmidhuber, J. (2009) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2008.137 · Cited by 1,984 · see citing works
RUSBoost: A Hybrid Approach to Alleviating Class Imbalance
Seiffert, C., Khoshgoftaar, T. M., Hulse, J. V., & Napolitano, A. (2009) · IEEE Transactions on Systems Man and Cybernetics - Part A Systems and Humans · doi:10.1109/tsmca.2009.2029559 · Cited by 1,872 · cited 2× · see citing works
“SMOTEBoost and RUSBoot are examples of ensemble rebalancing methods 30,37 .”
MethodSMOTEBoost: Improving Prediction of the Minority Class in Boosting
Chawla, N. V., Lazarević, A. M., Hall, L., & Bowyer, K. W. (2003) · Lecture notes in computer science · doi:10.1007/978-3-540-39804-2_12 · Cited by 1,741 · cited 1× · see citing works
“Our rebalancing analysis demonstrates that a variety of methods and ratios can lead to significant model prediction improvement in terms of F-1 measure 29,32–37 , 59– 61 .”
MethodMultimodal Compact Bilinear Pooling for Visual Question Answering and Visual Grounding
Fukui, A., Park, D. H., Yang, D., Rohrbach, A., Darrell, T., & Rohrbach, M. (2016) · Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d16-1044 · Cited by 1,582 · cited 1× · see citing works
“However, Fukuia et al. 57 and Deng et al. 21 point out that these methods are likely suboptimal as they do not leverage the interaction between weights and features at each network layer.”
BackgroundA governance model for the application of AI in health care
Reddy, S., Allan, S., Coghlan, S., & Cooper, P. (2020) · Journal of the American Medical Informatics Association · doi:10.1093/jamia/ocz192 · Cited by 979 · cited 1× · see citing works
“This understanding will allow for in-depth clinical analyses of model predictions and the reduction of unnecessary input parameters without compromising performance 62 .”
BackgroundInterpretable 3D Human Action Analysis with Temporal Convolutional Networks
Kim, T. S., & Reiter, A. (2017) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2017.207 · Cited by 696 · cited 1× · see citing works
“Additionally, a residual block connection has been added between every other layer to prevent overfitting 22 .”
BackgroundBorderline over-sampling for imbalanced data classification
Nguyen, H. M., Cooper, E. W., & Kamei, K. (2011) · International Journal of Knowledge Engineering and Soft Data Paradigms · doi:10.1504/ijkesdp.2011.039875 · Cited by 595 · cited 1× · see citing works
“…• Random (majority) under-sampling (no synthetic data) 53 • Random (minority) over-sampling (duplicate data) 54 • Synthetic Minority Oversampling Technique (SMOTE) (synthetic data) 30,54 • Borderline (BL) SMOTE (synthetic data) 54,55 • Support vector machine (SVM) SMOTE (synthetic data) 54,56”
Hou, N., Li, M., He, L., Xie, B., Wang, L., Zhang, R., Yu, Y., Sun, X., Pan, Z., & Wang, K. (2020) · Journal of Translational Medicine · doi:10.1186/s12967-020-02620-5 · Cited by 580 · cited 1× · see citing works
“However, multiple studies have already demonstrated superiority of modern machine learning algorithms to these models 16,17 .”
BackgroundA Multi-stream Bi-directional Recurrent Neural Network for Fine-Grained Action Detection
Singh, B., Marks, T. K., Jones, M., Tuzel, O., & Shao, M. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.216 · Cited by 461 · cited 1× · see citing works
“Conversely, RNN-based architectures (like the GRU-D) must be sequentially evaluated and demonstrate poor compute efficiency per parameter 66–68 .”
MethodLearning and making decisions when costs and probabilities are both unknown
Zadrozny, B., & Elkan, C. (2001) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/502512.502540 · Cited by 456 · cited 1× · see citing works
“Cost-sensitive learning methods penalize more for the misclassifications of the minority class in the loss function 35,36 .”
Optimal number of features as a function of sample size for various classification rules
Hua, J., Xiong, Z., Lowey, J., Suh, E. B., & Dougherty, E. R. (2004) · Computer applications in the biosciences · doi:10.1093/bioinformatics/bti171 · Cited by 440 · see citing works
Diagnostic Errors—The Next Frontier for Patient Safety
Newman‐Toker, D. E. (2009) · JAMA · doi:10.1001/jama.2009.249 · Cited by 434 · cited 1× · see citing works
“Binary classification tasks (4) ordered by rows; evaluation metrics (4) ordered by columns.”
BackgroundPredictors of patient satisfaction with hospital health care
Quintana, J. M., González, N., Bilbao, A., Aizpuru, F., Escobar, A., Esteban, C., San-Sebastián, J. A., de la Sierra, E., & Thompson, A. B. (2006) · BMC Health Services Research · doi:10.1186/1472-6963-6-102 · Cited by 385 · see citing works
Oneto, L., & Chiappa, S. (2020) · Studies in computational intelligence · doi:10.1007/978-3-030-43883-8_7 · Cited by 315 · see citing works
Mitigating bias in machine learning for medicine
Vokinger, K. N., Feuerriegel, S., & Kesselheim, A. S. (2021) · Communications Medicine · doi:10.1038/s43856-021-00028-w · Cited by 269 · cited 2× · see citing works
“Unlike random forest ensemble models (like the popular XGBoost algorithm 47 ), which have built-in interpretability, deep learning models are not equipped with feature importance scores by default.”
BackgroundRecurrent Neural Networks for Multivariate Time Series with Missing ValuesPreprint
Che, Z., Purushotham, S., Cho, K., et al. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1606.01865 · Cited by 230 · cited 1× · see citing works
“GRU-D was selected over LSTM because it was demonstrated as a state-of-the-art for this dataset in 41 and to outperform LSTM for MIMIC data 50 .”
MethodHoyer, E. H., Friedman, M. A., Lavezza, A., Wagner‐Kosmakos, K., Lewis‐Cherry, R., Skolnik, J. L., Byers, S. P., Atanelov, L., Colantuoni, E., Brotman, D. J., et al. (2016) · Journal of Hospital Medicine · doi:10.1002/jhm.2546 · Cited by 229 · see citing works
Leger, S., Zwanenburg, A., Pilz, K., Lohaus, F., Linge, A., Zöphel, K., Kotzerke, J., Schreiber, A., Tinhofer, I., Budach, V., et al. (2017) · Scientific Reports · doi:10.1038/s41598-017-13448-3 · Cited by 217 · see citing works
Knowledge-Driven Stock Trend Prediction and Explanation via Temporal Convolutional Network
Deng, S., Zhang, N., Zhang, W., Chen, J., Pan, J. Z., & Chen, H. (2019) · International World Wide Web Conference (WWW) · doi:10.1145/3308560.3317701 · Cited by 203 · cited 1× · see citing works
“However, Fukuia et al. 57 and Deng et al. 21 point out that these methods are likely suboptimal as they do not leverage the interaction between weights and features at each network layer.”
BackgroundSelf-citationPredictors of in-hospital length of stay among cardiac patients: A machine learning approach
Daghistani, T., Elshawi, R., Sakr, S., Ahmed, A. M., Al-Thwayee, A., & Al-Mallah, M. H. (2019) · International Journal of Cardiology · doi:10.1016/j.ijcard.2019.01.046 · Cited by 186 · see citing works
Is patient length of stay related to quality of care?
Thomas, J. W., Guire, K., & Horvat, G. (1997) · PubMed · Cited by 155 · see citing works
Factors associated with longer ED lengths of stay
Gardner, R. L., Sarkar, U., Maselli, J. H., & Gonzales, R. (2007) · The American Journal of Emergency Medicine · doi:10.1016/j.ajem.2006.11.037 · Cited by 150 · see citing works
Wang, S., McDermott, M. B. A., Chauhan, G., Ghassemi, M., Hughes, M. C., & Naumann, T. (2020) · ACM Conference on Health, Inference, and Learning (CHIL) · doi:10.1145/3368555.3384469 · Cited by 149 · see citing works
Elhassan, A. T., Aljourf, M., Al-Mohanna, F. A., & Shoukri, M. (2016) · Global Journal of Technology and Optimization · doi:10.4172/2229-8711.s1111 · Cited by 139 · cited 3× · see citing works
“This occurs because machine learning classifiers are often designed to minimize loss functions to maximize overall accuracy, which alone may not be satisfactory in application 32 .”
BackgroundMethodHierarchical Temporal Convolutional Networks for Dynamic Recommender Systems
You, J., Wang, Y., Pal, A., Eksombatchai, P., Rosenburg, C., & Leskovec, J. (2019) · International World Wide Web Conference (WWW) · doi:10.1145/3308558.3313747 · Cited by 126 · see citing works
Acute physiology and chronic health evaluation (APACHE II) and Medicare reimbursement.
Wagner, D. P., & Draper, E. A. (1984) · PubMed · Cited by 112 · see citing works
Tsai, P.-F. J., Chen, P.-C., Chen, Y.-Y., Hao-yuan, S., Lin, H.-M., Lin, F.-M., & Huang, Q.-P. (2016) · Journal of Healthcare Engineering · doi:10.1155/2016/7035463 · Cited by 105 · see citing works
Temporal pointwise convolutional networks for length of stay prediction in the intensive care unit
Rocheteau, E., Liò, P., & Hyland, S. L. (2020) · ACM Conference on Health, Inference, and Learning · doi:10.1145/3450439.3451860 · Cited by 87 · cited 4× · see citing works
“Rocheteau et. al . demonstrate that these methods are compatible with the TPC/TCN model architecture and are useful for clinical phenotyping and feature reduction before deployment 26 .”
BackgroundMethodInfluentialPoole, D., Rossi, C., Anghileri, A., Giardino, M., Latronico, N., Radrizzani, D., Langer, M., & Bertolini, G. (2009) · Intensive Care Medicine · doi:10.1007/s00134-009-1615-0 · Cited by 71 · cited 2× · see citing works
“Many electronic medical record integrated systems such as APACHE 14 and SAPS 15 historically utilize numerous static variables, so a direct comparison was not within scope.”
BackgroundTemporal Convolutional Networks for Action Segmentation and DetectionPreprint
Lea, C., Flynn, M. D., Vidal, R., Reiter, A., & Hager, G. D. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1611.05267 · Cited by 67 · cited 1× · see citing works
“TCNs were first proposed by Lea et. al. in 2016 20 and were largely popularized by their state-of-the-art performance in a wide range of applications (image classification, polyphonic music modeling, language modeling) as demonstrated by Bai et. al. in 2018 19 .”
MethodOsawa, I., Goto, T., Yamamoto, Y., & Tsugawa, Y. (2020) · npj Digital Medicine · doi:10.1038/s41746-020-00354-8 · Cited by 66 · cited 1× · see citing works
“Constraining decision support data to real-time applications precludes the use of ICD procedure and diagnosis codes, which become available to health practitioners days or weeks after discharge 42–45 .”
Is the length of stay in hospital correlated with patient satisfaction?
Borghans, I., Kleefstra, S. M., Kool, R. B., & Westert, G. P. (2012) · International Journal for Quality in Health Care · doi:10.1093/intqhc/mzs037 · Cited by 60 · see citing works
A Review of the Oversampling Techniques in Class Imbalance Problem
Sharma, S., Gosain, A., & Jain, S. (2021) · Advances in intelligent systems and computing · doi:10.1007/978-981-16-2594-7_38 · Cited by 60 · see citing works
Class Probability Estimation and Cost-Sensitive Classification Decisions
Margineantu, D. D. (2002) · Lecture notes in computer science · doi:10.1007/3-540-36755-1_23 · Cited by 59 · cited 2× · see citing works
“Cost-sensitive learning methods penalize more for the misclassifications of the minority class in the loss function 35,36 .”
BackgroundTemporal convolutional networks allow early prediction of events in critical care
Catling, F. J. R., & Wolff, A. (2019) · Journal of the American Medical Informatics Association · doi:10.1093/jamia/ocz205 · Cited by 49 · cited 5× · see citing works
“Catling and colleagues used TCNs to develop risk-prediction models which either perform comparably or outperform long short-term memory (LSTM) recurrent neural networks (RNN) in prediction of clinical events when provided one hour of temporal data 25 .”
BackgroundMethodInfluentialLewis, M., Elad, G., Beladev, M., Maor, G., Radinsky, K., Hermann, D., Litani, Y., Geller, T. P., Pines, J. M., Shapiro, N. L., et al. (2021) · Scientific Reports · doi:10.1038/s41598-020-80856-3 · Cited by 32 · cited 1× · see citing works
“Constraining decision support data to real-time applications precludes the use of ICD procedure and diagnosis codes, which become available to health practitioners days or weeks after discharge 42–45 .”
Bednarski, B. P., Singh, A. D., & Jones, W. M. (2020) · Journal of the American Medical Informatics Association · doi:10.1093/jamia/ocaa324 · Cited by 31 · cited 1× · see citing works
“time-series by nature 18,19 .”
Self-citationRFCL: A new under-sampling method of reducing the degree of imbalance and overlap
Zhang, R., Zhang, Z., & Wang, D. (2020) · Pattern Analysis and Applications · doi:10.1007/s10044-020-00929-x · Cited by 29 · cited 1× · see citing works
“Undersampling techniques remove random samples from the majority class, leaving all minority samples in place to achieve a desired outcome ratio 32,33 .”
BackgroundImprovement of APACHE II score system for disease severity based on XGBoost algorithm
Luo, Y., Wang, Z., & Wang, C. (2021) · BMC Medical Informatics and Decision Making · doi:10.1186/s12911-021-01591-x · Cited by 27 · cited 2× · see citing works
“However, multiple studies have already demonstrated superiority of modern machine learning algorithms to these models 16,17 .”
BackgroundWOTBoost: Weighted Oversampling Technique in Boosting for imbalanced learning
Zhang, W., Ramezani, R., & Naeim, A. (2019) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata47090.2019.9006091 · Cited by 26 · cited 1× · see citing works
“Irrespective of the model, predictive performance of classifiers can be unsatisfactory with imbalanced data-sets for which classes are not equally represented 28,29 .”
BackgroundSelf-citationMIMIC-III Clinical Database Demo
Johnson, A., Pollard, T., & Mark, R. (2019) · PhysioNet · doi:10.13026/c2hm2q · Cited by 24 · see citing works
Hierarchical Temporal Convolutional Networks for Dynamic Recommender SystemsPreprint
You, J., Wang, Y., Pal, A., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1904.04381 · Cited by 22 · see citing works
Designing a Successful Primary Care Physician Capitation Model
Emanuel, E., Mostashari, F., & Navathe, A. S. (2021) · JAMA · doi:10.1001/jama.2021.5133 · Cited by 19 · see citing works
GenSample: A Genetic Algorithm for Oversampling in Imbalanced DatasetsPreprint
Karia, V., Zhang, W., Naeim, A., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.10806 · Cited by 17 · cited 1× · see citing works
“Our rebalancing analysis demonstrates that a variety of methods and ratios can lead to significant model prediction improvement in terms of F-1 measure 29,32–37 , 59– 61 .”
MethodSelf-citationPredicting 30-Day Risk and Cost of "All-Cause" Hospital Readmissions
Sushmita, S., Khulbe, G., Hasan, A., Newman, S., Ravindra, P., Roy, S. B., Cock, M. D., & Teredesai, A. M. (2016) · National Conference on Artificial Intelligence · Cited by 16 · see citing works
Machine Learning as a Catalyst for Value-Based Health Care
Crowson, M. G., & Chan, T. C. Y. (2020) · Journal of Medical Systems · doi:10.1007/s10916-020-01607-5 · Cited by 12 · see citing works
Huang, W., Chen, Y., Wang, P., Liu, X., & Liu, S. (2021) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm52615.2021.9669653 · Cited by 2 · cited 2× · see citing works
“As the applications for AI in medicine expand to diverse tasks 25–27 , system architects are increasingly responsible for comprehensive model architecture searches to identify optimal methods.”
BackgroundMethodInterpretable 3D Human Action Analysis with Temporal Convolutional NetworksPreprint
Kim, T. S., & Reiter, A. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1704.04516 · cited 1×
“Additionally, a residual block connection has been added between every other layer to prevent overfitting 22 .”
Background
Flagged first, then by citations
Research works printed but not linked: 4
- [24]Martinez, B., Ma, P., Petridis, S. & Pantic, M. Lipreading using Temporal Convolutional Networks (Accessed 24 October 2021). http://arxiv.org/abs/2001.08702 [cs, eess] (2020). · arXiv
- [31]Provost, F. Machine Learning from Imbalanced Data Sets 101. Invited paper for the AAAI, Workshop on Imbalanced Data Sets, Menlo Park, CA (2000).
- [58]Ali A Shamsuddin SM Ralescu AL Classification with class imbalance problem: A review Int. J. Adv. Soft Comput. Appl. 2013 5 3 1 30
- [65]Sundararajan, M., Taly, A. & Yan, Q. Axiomatic attribution for deep networks. In Proceedings of the 34th International Conference on Machine Learning, Vol. 70 3319–3328 (2017).
Cited by
Sorted by citations
Data-Driven Methods for the State of Charge Estimation of Lithium-Ion Batteries: An Overview
Eleftheriadis, P., Giazitzis, S., Leva, S., & Ogliari, E. (2023) · Forecasting · doi:10.3390/forecast5030032 · Cited by 59 · see citing works
Prognosis of COVID-19 severity using DERGA, a novel machine learning algorithm
Asteris, P. G., Gandomi, A. H., Armaghani, D. J., Kokoris, S. I., Papandreadi, A. T., Roumelioti, A., Papanikolaou, S., Tsoukalas, M. Z., Triantafyllidis, L., Koutras, E. I., et al. (2024) · European Journal of Internal Medicine · doi:10.1016/j.ejim.2024.02.037 · Cited by 35 · see citing works
Multi-modal learning for inpatient length of stay prediction
Chen, J., Wen, Y., Pokojovy, M., Tseng, T.-L., McCaffrey, P., Vo, A. H., Walser, E., & Moen, S. T. (2024) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2024.108121 · Cited by 26 · see citing works
Giesa, N., Sekutowicz, M., Rubarth, K., Spies, C., Balzer, F., Haufe, S., & Boie, S. D. (2024) · Communications Medicine · doi:10.1038/s43856-024-00681-x · Cited by 25 · see citing works
“Weacknowledgethataddi-tionalmodelarchitectures,suchasCNN-basedmodels 42,104 , could potentially be con fi gured for predictions on clinical time seriesdata.”
Ali, S., Wang, J., Leung, V. C. M., Bashir, F., Bhatti, U. A., Wadho, S. A., & Humayun, M. (2025) · Information Fusion · doi:10.1016/j.inffus.2025.103222 · Cited by 22 · see citing works
Artificial Intelligence and Machine Learning in Transfusion Practice: An Analytical Assessment
Li, N., Goel, R., Raza, S., Riazi, K., Pan, J., Nguyen, H. Q., Shih, A. W. Y., D’Souza, A. G., Dubey, R., Tobian, A. A. R., et al. (2025) · Transfusion Medicine Reviews · doi:10.1016/j.tmrv.2025.150926 · Cited by 17 · see citing works
Transformer-Based Multivariate Time Series Forecasting
Bharti, M. K., Wadhvani, R., Gyanchandani, M., & Gupta, M. (2024) · IEEE International Students' Conference on Electrical, Electronics and Computer Science (SCEECS) · doi:10.1109/sceecs61402.2024.10482217 · Cited by 15 · see citing works
“…Smoothing (ETS) [6] methods, Seasonal Decomposition of Time Series (STL) [7] and other models are not able to handle the long-term dependencies and the traditional model could not be able to predict better as compare to the new deep learning model like RNN [8], TCN [9] and Transformer model [10].”
MethodZarghami, M., Niknam, T., Aghaei, J., & Nezhad, A. H. (2024) · IET Renewable Power Generation · doi:10.1049/rpg2.13050 · Cited by 14 · see citing works
Huang, Y., Wang, M., Zheng, Z., Ma, M., Fei, X., Wei, L., & Chen, H. (2023) · Journal of Biomedical Informatics · doi:10.1016/j.jbi.2023.104427 · Cited by 8 · see citing works
Sheng, T., & Huber, M. (2025) · Sensors · doi:10.3390/s25134032 · Cited by 5 · see citing works
“For example, Lea et al. [51] utilized a TCN for action segmentation and detection, Zhang et al. [52] applied it for text classification, and Bednarski et al. [53] employed it in health informatics for predicting clinical length of stay and mortality.”
Meng, L., Li, J., Liu, X., Sun, Y., Li, Z., Cai, J., Parab, A. D., Lu, G., Budhkar, A., Kanakasabai, S., et al. (2025) · npj Digital Medicine · doi:10.1038/s41746-025-01863-0 · Cited by 4 · see citing works
“Existingtemporaldynamicmodels — suchaslong short-term memory networks 19 – 21 , temporal convolutional neural networks 22,23 , recurrent neural networks 24 , other neural network models 25,26 , and regression-based approaches like DYNAMIC-ICU 27 , and ensemble frameworks 34,35 — have advanced…”
Xie, Z.-A., Chow, C., Chuah, J. H., & Raymond, W. J. K. (2025) · Frontiers in Environmental Science · doi:10.3389/fenvs.2025.1623630 · Cited by 4 · see citing works
Pham, M.-K., Mai, T. T., Crane, M., Brennan, R., Ward, M. E., Geary, Ú., Byrne, D., O'Connell, B. C., Bergin, C. J., Creagh, D., et al. (2025) · BMC Medical Informatics and Decision Making · doi:10.1186/s12911-025-03214-1 · Cited by 4 · see citing works
“Tabular deep models like Convolutional Networks [34], TabNet [35], TabTransformer [36], ResNet [37], and TabPFN [38] have similarly demonstrated effectiveness across general tabular datasets, but not quite well studied in healthcare domain.”
Zaribafzadeh, H., Howell, T. C., Webster, W. L., Vail, C. J., Kirk, A. D., Allen, P. J., Henao, R., & Buckland, D. M. (2025) · Annals of Surgery Open · doi:10.1097/as9.0000000000000547 · Cited by 3 · see citing works
“44 Predicting LOS before admission could help hospitals’ admission and bed management to monitor the quality of inpatient care, identify abnormalities or potential risks that require additional resources, and anticipate any bottleneck in bed availability before patients are admitted.”
Learning the Long-term Memory Effect of Power Amplifiers Using Temporal Convolutional Network
Akram, I., Ma, Y., He, Z., & Tong, F. (2023) · IEEE Vehicular Technology Conference · doi:10.1109/vtc2023-fall60731.2023.10333622 · Cited by 3 · see citing works
“More specifically, RVTDTCN is built upon the temporal convolutional network (TCN) architecture, known for its effectiveness in processing sequential data [13]–[15].”
BackgroundHanji, S. S., & Birje, M. N. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-19078-y · Cited by 3 · see citing works
“Bednarski et al. [24] presented a model to forecast the length of stay and mortality assess data rebalancing techniques using temporal convolutional networks (TCNs).”
Fan, G. F., Liu, A., & Zhang, C. (2025) · IEEE Access · doi:10.1109/access.2025.3540896 · Cited by 3 · see citing works
“For masked feature imputation, we randomly mask a portion of the input features and train the model to reconstruct these masked values, similar to the masked language modeling approach in BERT [36].”
InfluentialRavishan, D., Prasanna, R., Herath, P., & Doyle, E. E. H. (2026) · Scientific Reports · doi:10.1038/s41598-026-42568-y · Cited by 2 · see citing works
“RNNs suffer from slow, sequential processing and high memory consumption 47 , while the self-attention mechanism of Transformers scales quadratically with input length, demanding extensive computational resources and a large Fig.”
Early In-Hospital Mortality Prediction Based on xTimesNet and Time Series Interpretable Methods
Wang, X., Yuan, Z., Yao, Z., Yuan, Y., & Luo, L. (2025) · IEEE Access · doi:10.1109/access.2025.3571789 · Cited by 2 · see citing works
“Current methodologies predominantly employ temporal-domain architectures, including Long Short-Term Memory networks (LSTM) [5], Temporal Convolutional Networks (TCN) [6], and attention-based transformers [7].”
Johnson, H., & Tipirneni‐Sajja, A. (2024) · Metabolites · doi:10.3390/metabo14120666 · Cited by 1 · see citing works
Taskina, L. A., Abakumov, L., & Kazarkin, T. (2025) · Computer Optics · doi:10.18287/coj1849 · Cited by 1 · see citing works
“The trained hybrid model TCN on sequences plus ResNet–18 on attention images consistently outperforms single– branch baselines for both load level (multiclass L1, L2, L2–L3 and L3) and stress resistance (binary degradation under stress).”
InfluentialXin, X., xie, Z., Liu, S., Yu, G., & Cao, J. (2026) · Processes · doi:10.3390/pr14101558 · Cited by 1 · see citing works
Hmimdi, A. E. E., Palpanas, T., & Kapoula, Z. (2024) · Scientific Reports · doi:10.1038/s41598-024-68056-9 · Cited by 1 · see citing works
Rahimi, F., Sirocchi, C., Matschinske, J., Metzler, M., Zierk, J., & Blumenthal, D. B. (2025) · medRxiv · doi:10.64898/2025.12.12.25342142 · Cited by 1 · see citing works
Research on Chaotic Time Series Prediction Algorithms
韩, 宛. (2025) · Advances in Applied Mathematics · doi:10.12677/aam.2025.143124
Bhuiyan, M. S. M., Rafi, M. A., Rodrigues, G. N., Mir, M. N. H., Abir, M. G. R., Mridha, M. F., & Shin, J. (2024) · International Conference on Computer and Information Technology (ICCIT) · doi:10.1109/iccit64611.2024.11022412
“Bednarski et al. [11] employed Temporal Convolutional Networks (TCNs) for LOS and mortality prediction, outperforming GRU-D models in mortality tasks.”
Mazziero, V. T., Veloso, I. I. K., & Cerri, M. O. (2025) · Research Square · doi:10.21203/rs.3.rs-7448392/v1
Gong, K., Xu, Y., & Xie, X. (2025) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2025.3639425
“88[37], [38],[39],[40],whileJiangetal.”
Techniques for Sequential Data in Healthcare
Pham, T. D., Holmes, S., Chatzopoulou, D., & Coulthard, P. (2026) · Springer eBooks · doi:10.1007/978-3-032-11531-7_4
Stock Price Prediction Based on Phase Space Reconstruction and TCN
Guo, W., Liu, K., & Zhao, J. (2025) · Journal of statistics and economics. · doi:10.62517/jse.202511413
Lin, W., & Grewal, S. S. (2026) · Sensors · doi:10.3390/s26041352
“…(b) causal dilated convolutions naturally respect temporal ordering, preventing information leakage from future time steps; and (c) prior work has demonstrated competitive or superior performance of TCNs relative to recurrent networks for biomedical time-series classification tasks [12–14].”
Celada-Bernal, S., Piñán-Roescher, A., Hernández-López, R., & Travieso, C. M. (2026) · Applied Sciences · doi:10.3390/app16052235
Chen, Y., Ou, Z., Deng, Y., Luo, A., Li, X., Yang, Y., Yang, Q., Wu, X., & Zhou, M. (2026) · Journal of Medical Internet Research · doi:10.2196/83821 · Cited by 3 · see citing works
“Besides, DL has high efficiency for image processing and excellent time effect capture ability when processing time series data [124,125], so it has the potential for judging stroke site and time.”
Zeleke, S. N., & Bochicchio, M. A. (2026) · Communications in computer and information science · doi:10.1007/978-3-032-21514-7_1
A physiology-guided temporal graph learning approach for intensive care mortality prediction
Joseph, J., & Judy, M. V. (2026) · Healthcare Analytics · doi:10.1016/j.health.2026.100467
Charulatha, R. T., & Aravind, T. (2026) · International Conference on Intelligent Computing (ICONIC) · doi:10.1109/iconic67661.2026.11517661
“Figure 2 shows the internal architecture of the MRD-TCN model, which consists of dilated causal convolutions, residual blocks, and multiscale convolutional operations to learn hierarchical features [15].”
Classifying Sleep Disorders Through Lifestyle-Driven Health Data Using Deep Learning
Arumugam, M., Manikandan, S., & Arun, G. (2026) · International Conference on Computational Robotics, Testing and Engineering Evaluation (ICCRTEE) · doi:10.1109/iccrtee68719.2026.11566172
“The authors of this paper claim that TCNs outperform RNNs in vital-sign monitoring, ECG, and EEG data, and are not subjected to the sequential bottleneck of RNNs as they can capture long-range temporal dependencies through dilated causal convolutions [2], [5], [6].”
HazardFlow: Enhancing Health Status Representations via Score-based Energy Modeling
Chen, Q., Li, X., Zhang, Y., & Wang, M. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770855.3817616
Residual Transformer–Decision Forest Hybrid Learning for Accurate Hospital Length-of-Stay Prediction
Neshat, M. (2026) · International Conference on Intelligent Systems, Metaheuristics & Swarm Intelligence (ISMSI) · doi:10.1145/3820820.3820830
Choi, K., Kim, S. J., & Kim, Y. (2026) · Communications in computer and information science · doi:10.1007/978-3-032-28463-1_19
Bhattacharya, A., Sharma, A., Mondal, S., & De, A. (2026) · Combustion Science and Technology · doi:10.1080/00102202.2026.2748443
- Authors
- Bednarski, B. P., Singh, A. D., Zhang, W., Jones, W. M., Naeim, A., & Ramezani, R.
- Affiliations
- University of California, Los Angeles (US)
- University of California, Irvine (US)
- Corresponding author
- Bednarski, Bryan P.
- Venue
- Scientific Reports
- Published
- 2022-12-08
- Volume
- 12
- Issue
- 1
- Pages
- 21247-21247
- ISSN
- 2045-2322 (ISSN-L)
- PMC
- PMC9732283
- OpenAlex
- W4311586200
- Cited by
- 41
- References
- 55
- Citations by year
- 2026: 14 · 2025: 15 · 2024: 9 · 2023: 3
- FWCI
- 2.22
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Locations
- Scientific Reports (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- eScholarship (California Digital Library) (submittedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning in Healthcare
- Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
- Health Sciences › Medicine › Epidemiology › Sepsis Diagnosis and Treatment
- Keywords
- temporal convolutional network, length of stay prediction, mortality prediction, in-hospital mortality, ICU mortality, data rebalancing, imbalanced learning, MIMIC-III
- MeSH terms
- Humans, Prospective Studies, Retrospective Studies
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.
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
