Attention-Enhanced CNN-LSTM Model for Exercise Oxygen Consumption Prediction with Multi-Source Temporal Features
Dynamic oxygen uptake (VO<sub>2</sub>) 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—the limited fusion of heterogeneous sensor data and inadequate...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-06-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/25/13/4062 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849701815389519872 |
|---|---|
| author | Zhen Wang Yingzhe Song Lei Pang Shanjun Li Gang Sun |
| author_facet | Zhen Wang Yingzhe Song Lei Pang Shanjun Li Gang Sun |
| author_sort | Zhen Wang |
| collection | DOAJ |
| description | Dynamic oxygen uptake (VO<sub>2</sub>) 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—the limited fusion of heterogeneous sensor data and inadequate modeling of long-range temporal patterns—by integrating wearable accelerometer and heart-rate streams with a convolutional neural network–LSTM (CNN-LSTM) architecture and optional attention modules. Physiological signals and VO<sub>2</sub> 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 <i>R</i><sup>2</sup> = 0.946, outperforming a plain LSTM (<i>R</i><sup>2</sup> = 0.926) thanks to stronger local spatio-temporal feature extraction. Introducing a spatial attention mechanism raised accuracy further (<i>R</i><sup>2</sup> = 0.962), whereas temporal attention reduced it (<i>R</i><sup>2</sup> = 0.930), indicating that attention success depends on how well the attended features align with exercise dynamics. Stacking both attentions (spatio-temporal) yielded <i>R</i><sup>2</sup> = 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. |
| format | Article |
| id | doaj-art-4f2bd56c8fd14b87a32fb41cf1469b02 |
| institution | DOAJ |
| issn | 1424-8220 |
| language | English |
| publishDate | 2025-06-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj-art-4f2bd56c8fd14b87a32fb41cf1469b022025-08-20T03:17:51ZengMDPI AGSensors1424-82202025-06-012513406210.3390/s25134062Attention-Enhanced CNN-LSTM Model for Exercise Oxygen Consumption Prediction with Multi-Source Temporal FeaturesZhen Wang0Yingzhe Song1Lei Pang2Shanjun Li3Gang Sun4Institute of Artificial Intelligence in Sports, Capital University of Physical Education and Sports, Beijing 100191, ChinaInstitute of Artificial Intelligence in Sports, Capital University of Physical Education and Sports, Beijing 100191, ChinaInstitute of Artificial Intelligence in Sports, Capital University of Physical Education and Sports, Beijing 100191, ChinaInstitute of Artificial Intelligence in Sports, Capital University of Physical Education and Sports, Beijing 100191, ChinaInstitute of Artificial Intelligence in Sports, Capital University of Physical Education and Sports, Beijing 100191, ChinaDynamic oxygen uptake (VO<sub>2</sub>) 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—the limited fusion of heterogeneous sensor data and inadequate modeling of long-range temporal patterns—by integrating wearable accelerometer and heart-rate streams with a convolutional neural network–LSTM (CNN-LSTM) architecture and optional attention modules. Physiological signals and VO<sub>2</sub> 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 <i>R</i><sup>2</sup> = 0.946, outperforming a plain LSTM (<i>R</i><sup>2</sup> = 0.926) thanks to stronger local spatio-temporal feature extraction. Introducing a spatial attention mechanism raised accuracy further (<i>R</i><sup>2</sup> = 0.962), whereas temporal attention reduced it (<i>R</i><sup>2</sup> = 0.930), indicating that attention success depends on how well the attended features align with exercise dynamics. Stacking both attentions (spatio-temporal) yielded <i>R</i><sup>2</sup> = 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.https://www.mdpi.com/1424-8220/25/13/4062oxygen uptakedeep learningneural networkattention mechanism |
| spellingShingle | Zhen Wang Yingzhe Song Lei Pang Shanjun Li Gang Sun Attention-Enhanced CNN-LSTM Model for Exercise Oxygen Consumption Prediction with Multi-Source Temporal Features Sensors oxygen uptake deep learning neural network attention mechanism |
| title | Attention-Enhanced CNN-LSTM Model for Exercise Oxygen Consumption Prediction with Multi-Source Temporal Features |
| title_full | Attention-Enhanced CNN-LSTM Model for Exercise Oxygen Consumption Prediction with Multi-Source Temporal Features |
| title_fullStr | Attention-Enhanced CNN-LSTM Model for Exercise Oxygen Consumption Prediction with Multi-Source Temporal Features |
| title_full_unstemmed | Attention-Enhanced CNN-LSTM Model for Exercise Oxygen Consumption Prediction with Multi-Source Temporal Features |
| title_short | Attention-Enhanced CNN-LSTM Model for Exercise Oxygen Consumption Prediction with Multi-Source Temporal Features |
| title_sort | attention enhanced cnn lstm model for exercise oxygen consumption prediction with multi source temporal features |
| topic | oxygen uptake deep learning neural network attention mechanism |
| url | https://www.mdpi.com/1424-8220/25/13/4062 |
| work_keys_str_mv | AT zhenwang attentionenhancedcnnlstmmodelforexerciseoxygenconsumptionpredictionwithmultisourcetemporalfeatures AT yingzhesong attentionenhancedcnnlstmmodelforexerciseoxygenconsumptionpredictionwithmultisourcetemporalfeatures AT leipang attentionenhancedcnnlstmmodelforexerciseoxygenconsumptionpredictionwithmultisourcetemporalfeatures AT shanjunli attentionenhancedcnnlstmmodelforexerciseoxygenconsumptionpredictionwithmultisourcetemporalfeatures AT gangsun attentionenhancedcnnlstmmodelforexerciseoxygenconsumptionpredictionwithmultisourcetemporalfeatures |