{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,4]],"date-time":"2026-09-04T18:32:02Z","timestamp":1788546722215,"version":"build-2803163510"},"reference-count":37,"publisher":"MDPI AG","issue":"13","license":[{"start":{"date-parts":[[2025,6,29]],"date-time":"2025-06-29T00:00:00Z","timestamp":1751155200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100012401","name":"Beijing Science and Technology Plan Project","doi-asserted-by":"publisher","award":["Z221100005222031"],"award-info":[{"award-number":["Z221100005222031"]}],"id":[{"id":"10.13039\/501100012401","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Sensors"],"abstract":"<jats:p>Dynamic oxygen uptake (VO2) reflects moment-to-moment changes in oxygen consumption during exercise and underpins training design, performance enhancement, and clinical decision-making. We tackled two key obstacles\u2014the limited fusion of heterogeneous sensor data and inadequate modeling of long-range temporal patterns\u2014by integrating wearable accelerometer and heart-rate streams with a convolutional neural network\u2013LSTM (CNN-LSTM) architecture and optional attention modules. Physiological signals and VO2 were recorded from 21 adults through resting assessment and cardiopulmonary exercise testing. The results showed that pairing accelerometer with heart-rate inputs improves prediction compared with considering the heart rate alone. The baseline CNN-LSTM reached R2 = 0.946, outperforming a plain LSTM (R2 = 0.926) thanks to stronger local spatio-temporal feature extraction. Introducing a spatial attention mechanism raised accuracy further (R2 = 0.962), whereas temporal attention reduced it (R2 = 0.930), indicating that attention success depends on how well the attended features align with exercise dynamics. Stacking both attentions (spatio-temporal) yielded R2 = 0.960, slightly below the value for spatial attention alone, implying that added complexity does not guarantee better performance. Across all models, prediction errors grew during high-intensity bouts, highlighting a bottleneck in capturing non-linear physiological responses under heavy load. These findings inform architecture selection for wearable metabolic monitoring and clarify when attention mechanisms add value.<\/jats:p>","DOI":"10.3390\/s25134062","type":"journal-article","created":{"date-parts":[[2025,6,30]],"date-time":"2025-06-30T12:10:31Z","timestamp":1751285431000},"page":"4062","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":9,"title":["Attention-Enhanced CNN-LSTM Model for Exercise Oxygen Consumption Prediction with Multi-Source Temporal Features"],"prefix":"10.3390","volume":"25","author":[{"ORCID":"https:\/\/orcid.org\/0009-0000-3452-6041","authenticated-orcid":false,"given":"Zhen","family":"Wang","sequence":"first","affiliation":[{"name":"Institute of Artificial Intelligence in Sports, Capital University of Physical Education and Sports, Beijing 100191, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yingzhe","family":"Song","sequence":"additional","affiliation":[{"name":"Institute of Artificial Intelligence in Sports, Capital University of Physical Education and Sports, Beijing 100191, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lei","family":"Pang","sequence":"additional","affiliation":[{"name":"Institute of Artificial Intelligence in Sports, Capital University of Physical Education and Sports, Beijing 100191, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Shanjun","family":"Li","sequence":"additional","affiliation":[{"name":"Institute of Artificial Intelligence in Sports, Capital University of Physical Education and Sports, Beijing 100191, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Gang","family":"Sun","sequence":"additional","affiliation":[{"name":"Institute of Artificial Intelligence in Sports, Capital University of Physical Education and Sports, Beijing 100191, China"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2025,6,29]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"1054","DOI":"10.1016\/j.mayocp.2022.02.029","article-title":"Objectively Assessed Cardiorespiratory Fitness and All-Cause Mortality Risk","volume":"97","author":"Laukkanen","year":"2022","journal-title":"Mayo Clin. Proc."},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"373","DOI":"10.2165\/00007256-200029060-00001","article-title":"The Effect of Endurance Training on Parameters of Aerobic Fitness","volume":"29","author":"Jones","year":"2000","journal-title":"Sports Med."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"207","DOI":"10.2114\/ahs1983.11.207","article-title":"Gas Exchange Dynamics and the Tolerance to Muscular Exercise: Effects of Fitness and Training","volume":"11","author":"Whipp","year":"1992","journal-title":"Ann. Physiol. Anthropol."},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"2261","DOI":"10.1161\/CIR.0b013e31826fb946","article-title":"Clinical Recommendations for Cardiopulmonary Exercise Testing Data Assessment in Specific Patient Populations","volume":"126","author":"Guazzi","year":"2012","journal-title":"Circulation"},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"139","DOI":"10.1007\/s00421-006-0255-0","article-title":"Accuracy and Reliability of the ParvoMedics TrueOne 2400 and MedGraphics VO2000 Metabolic Systems","volume":"98","author":"Crouter","year":"2006","journal-title":"Eur. J. Appl. Physiol."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"e14490","DOI":"10.1111\/sms.14490","article-title":"Accuracy of Respiratory Gas Variables, Substrate, and Energy Use from 15 CPET Systems During Simulated and Human Exercise","volume":"34","author":"Souren","year":"2024","journal-title":"Scand. J. Med. Sci. Sports"},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"2005","DOI":"10.1249\/MSS.0b013e318217276e","article-title":"HR Index\u2014A Simple Method for the Prediction of Oxygen Uptake","volume":"43","author":"Wicks","year":"2011","journal-title":"Med. Sci. Sports Exerc."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"289","DOI":"10.1080\/02640410470001730089","article-title":"Prediction of Energy Expenditure from Heart Rate Monitoring During Submaximal Exercise","volume":"23","author":"Keytel","year":"2005","journal-title":"J. Sports Sci."},{"key":"ref_9","doi-asserted-by":"crossref","unstructured":"Davidson, P., Trinh, H., Vekki, S., and M\u00fcller, P. (2023). Surrogate Modelling for Oxygen Uptake Prediction Using LSTM Neural Network. Sensors, 23.","DOI":"10.3390\/s23042249"},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Li, F., Chang, C.-H., Chung, Y.-C., Wu, H.-J., Kan, N.-W., ChangChien, W.-S., Ho, C.-S., and Huang, C.-C. (2021). Development and Validation of 3 Min Incremental Step-In-Place Test for Predicting Maximal Oxygen Uptake in Home Settings: A Submaximal Exercise Study to Assess Cardiorespiratory Fitness. Int. J. Environ. Res. Public Health, 18.","DOI":"10.3390\/ijerph182010750"},{"key":"ref_11","doi-asserted-by":"crossref","unstructured":"DiPietro, R., and Hager, G.D. (2020). Deep learning: RNNs and LSTM. Handbook of Medical Image Computing and Computer Assisted Intervention, Elsevier.","DOI":"10.1016\/B978-0-12-816176-0.00026-0"},{"key":"ref_12","doi-asserted-by":"crossref","first-page":"1596","DOI":"10.1249\/01.MSS.0000084593.56786.DA","article-title":"A Treadmill Ramp Protocol Using Simultaneous Changes in Speed and Grade","volume":"35","author":"Porszasz","year":"2003","journal-title":"Med. Sci. Sports Exerc."},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"127934","DOI":"10.1016\/j.jhydrol.2022.127934","article-title":"Prediction Model of Drinking Water Source Quality with Potential Industrial-Agricultural Pollution Based on CNN-GRU-Attention","volume":"610","author":"Mei","year":"2022","journal-title":"J. Hydrol."},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"331","DOI":"10.1007\/s41095-022-0271-y","article-title":"Attention Mechanisms in Computer Vision: A Survey","volume":"8","author":"Guo","year":"2022","journal-title":"Comput. Vis. Media"},{"key":"ref_15","doi-asserted-by":"crossref","unstructured":"Hu, J., Shen, L., and Sun, G. (2018, January 18\u201322). Squeeze-and-Excitation Networks. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA.","DOI":"10.1109\/CVPR.2018.00745"},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"3","DOI":"10.1007\/978-3-030-01234-2_1","article-title":"CBAM: Convolutional Block Attention Module","volume":"Volume 11211","author":"Ferrari","year":"2018","journal-title":"Computer Vision\u2014ECCV 2018"},{"key":"ref_17","unstructured":"American College of Sports Medicine (2021). ACSM\u2019s Guidelines for Exercise Testing and Prescription, Wolters Kluwer. [11th ed.]."},{"key":"ref_18","doi-asserted-by":"crossref","unstructured":"Lu, Z., Yang, J., Tao, K., Li, X., Xu, H., and Qiu, J. (2024). Combined Impact of Heart Rate Sensor Placements with Respiratory Rate and Minute Ventilation on Oxygen Uptake Prediction. Sensors, 24.","DOI":"10.3390\/s24165412"},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Bangaru, S.S., Wang, C., Aghazadeh, F., Muley, S., and Willoughby, S. (2025). Oxygen Uptake Prediction for Timely Construction Worker Fatigue Monitoring Through Wearable Sensing Data Fusion. Sensors, 25.","DOI":"10.3390\/s25103204"},{"key":"ref_20","doi-asserted-by":"crossref","unstructured":"G\u00f3mez-Carmona, C.D., Bastida-Castillo, A., Ib\u00e1\u00f1ez, S.J., and Pino-Ortega, J. (2020). Accelerometry as a Method for External Workload Monitoring in Invasion Team Sports. A Systematic Review. PLoS ONE, 15.","DOI":"10.1371\/journal.pone.0236643"},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Sheridan, D., Jaspers, A., Viet Cuong, D., Op De Be\u00e9ck, T., Moyna, N.M., de Beukelaar, T.T., and Roantree, M. (2025). Estimating Oxygen Uptake in Simulated Team Sports Using Machine Learning Models and Wearable Sensor Data: A Pilot Study. PLoS ONE, 20.","DOI":"10.1371\/journal.pone.0319760"},{"key":"ref_22","doi-asserted-by":"crossref","first-page":"242","DOI":"10.1109\/JPROC.2015.2491979","article-title":"Multiscale Analysis of Intensive Longitudinal Biomedical Signals and Its Clinical Applications","volume":"104","author":"Nakamura","year":"2016","journal-title":"Proc. IEEE"},{"key":"ref_23","doi-asserted-by":"crossref","unstructured":"Ernst, G. (2017). Heart-Rate Variability\u2014More than Heart Beats?. Front. Public Health, 5.","DOI":"10.3389\/fpubh.2017.00240"},{"key":"ref_24","doi-asserted-by":"crossref","unstructured":"De Brabandere, A., Op De Be\u00e9ck, T., Sch\u00fctte, K.H., Meert, W., Vanwanseele, B., and Davis, J. (2018). Data Fusion of Body-Worn Accelerometers and Heart Rate to Predict VO2max during Submaximal Running. PLoS ONE, 13.","DOI":"10.1371\/journal.pone.0199509"},{"key":"ref_25","doi-asserted-by":"crossref","unstructured":"Lee, C.J., and Lee, J.K. (2024). IMU-Based Energy Expenditure Estimation for Various Walking Conditions Using a Hybrid CNN\u2013LSTM Model. Sensors, 24.","DOI":"10.3390\/s24020414"},{"key":"ref_26","doi-asserted-by":"crossref","unstructured":"Hossain, M.B., LaMunion, S.R., Crouter, S.E., Melanson, E.L., and Sazonov, E. (2024). A CNN Model for Physical Activity Recognition and Energy Expenditure Estimation from an Eyeglass-Mounted Wearable Sensor. Sensors, 24.","DOI":"10.3390\/s24103046"},{"key":"ref_27","doi-asserted-by":"crossref","first-page":"156","DOI":"10.1038\/s41746-021-00531-3","article-title":"Temporal Convolutional Networks Predict Dynamic Oxygen Uptake Response from Wearable Sensors Across Exercise Intensities","volume":"4","author":"Amelard","year":"2021","journal-title":"NPJ Digit. Med."},{"key":"ref_28","doi-asserted-by":"crossref","unstructured":"Zhu, C., Liu, Q., Meng, W., Ai, Q., and Xie, S.Q. (2021, January 12\u201316). An Attention-Based CNN-LSTM Model with Limb Synergy for Joint Angles Prediction. Proceedings of the 2021 IEEE\/ASME International Conference on Advanced Intelligent Mechatronics (AIM), Delft, The Netherlands.","DOI":"10.1109\/AIM46487.2021.9517544"},{"key":"ref_29","unstructured":"Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, \u0141., and Polosukhin, I. (2017). Attention Is All You Need. Advances in Neural Information Processing Systems, MIT Press."},{"key":"ref_30","unstructured":"Cao, F., Yang, S., Chen, Z., Liu, Y., and Cui, L. (2024). Ister: Inverted Seasonal-Trend Decomposition Transformer for Explainable Multivariate Time Series Forecasting. arXiv."},{"key":"ref_31","doi-asserted-by":"crossref","first-page":"102655","DOI":"10.1016\/j.aei.2024.102655","article-title":"Multi-Fidelity Fusion for Soil Classification via LSTM and Multi-Head Self-Attention CNN Model","volume":"62","author":"Zhou","year":"2024","journal-title":"Adv. Eng. Inform."},{"key":"ref_32","doi-asserted-by":"crossref","unstructured":"Zheng, B., Luo, W., Zhang, M., and Jin, H. (2025). Arrhythmia Classification Based on Multi-Input Convolutional Neural Network with Attention Mechanism. PLoS ONE, 20.","DOI":"10.1371\/journal.pone.0326079"},{"key":"ref_33","doi-asserted-by":"crossref","unstructured":"Khan, M., and Hossni, Y. (2025). A Comparative Analysis of LSTM Models Aided with Attention and Squeeze and Excitation Blocks for Activity Recognition. Sci. Rep., 15.","DOI":"10.1038\/s41598-025-88378-6"},{"key":"ref_34","doi-asserted-by":"crossref","first-page":"100","DOI":"10.1007\/s00421-002-0690-5","article-title":"Oxygen Uptake and Heart Rate Kinetics During Heavy Exercise: A Comparison Between Arm Cranking and Leg Cycling","volume":"88","author":"Schneider","year":"2002","journal-title":"Eur. J. Appl. Physiol."},{"key":"ref_35","doi-asserted-by":"crossref","first-page":"97","DOI":"10.1016\/j.ins.2023.01.093","article-title":"Rethinking Attention Mechanism in Time Series Classification","volume":"627","author":"Zhao","year":"2023","journal-title":"Inf. Sci."},{"key":"ref_36","doi-asserted-by":"crossref","first-page":"1480","DOI":"10.1152\/japplphysiol.00570.2021","article-title":"Modeling Vo2 On-Kinetics Based on Intensity-Dependent Delayed Adjustment and Loss of Efficiency (DALE)","volume":"132","author":"Colosio","year":"2022","journal-title":"J. Appl. Physiol."},{"key":"ref_37","unstructured":"Yang, Y., Zha, K., Chen, Y., Wang, H., and Katabi, D. (2021, January 18\u201324). Delving into Deep Imbalanced Regression. Proceedings of the International Conference on Machine Learning, PMLR, Online."}],"container-title":["Sensors"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1424-8220\/25\/13\/4062\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,9]],"date-time":"2025-10-09T18:01:30Z","timestamp":1760032890000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1424-8220\/25\/13\/4062"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,6,29]]},"references-count":37,"journal-issue":{"issue":"13","published-online":{"date-parts":[[2025,7]]}},"alternative-id":["s25134062"],"URL":"https:\/\/doi.org\/10.3390\/s25134062","relation":{},"ISSN":["1424-8220"],"issn-type":[{"value":"1424-8220","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,6,29]]}}}