G-CTRNN: A Trainable Low-Power Continuous-Time Neural Network for Human Activity Recognition in Healthcare Applications
Continuous-time Recurrent Neural Networks (CTRNNs) are well-suited for modeling temporal dynamics in low-power neuromorphic and analog computing systems, making them promising candidates for edge-based human activity recognition (HAR) in healthcare. However, training CTRNNs remains challenging due t...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-07-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/15/13/7508 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850118923647713280 |
|---|---|
| author | Abdallah Alzubi David Lin Johan Reimann Fadi Alsaleem |
| author_facet | Abdallah Alzubi David Lin Johan Reimann Fadi Alsaleem |
| author_sort | Abdallah Alzubi |
| collection | DOAJ |
| description | Continuous-time Recurrent Neural Networks (CTRNNs) are well-suited for modeling temporal dynamics in low-power neuromorphic and analog computing systems, making them promising candidates for edge-based human activity recognition (HAR) in healthcare. However, training CTRNNs remains challenging due to their continuous-time nature and the need to respect physical hardware constraints. In this work, we propose G-CTRNN, a novel gradient-based training framework for analog-friendly CTRNNs designed for embedded healthcare applications. Our method extends Backpropagation Through Time (BPTT) to continuous domains using TensorFlow’s automatic differentiation, while enforcing constraints on time constants and synaptic weights to ensure hardware compatibility. We validate G-CTRNN on the WISDM human activity dataset, which simulates realistic wearable sensor data for healthcare monitoring. Compared to conventional RNNs, G-CTRNN achieves superior classification accuracy with fewer parameters and greater stability—enabling continuous, real-time HAR on low-power platforms such as MEMS computing networks. The proposed framework provides a pathway toward on-device AI for remote patient monitoring, elderly care, and personalized healthcare in resource-constrained environments. |
| format | Article |
| id | doaj-art-6a2945922e664a95828d35a5ff13bc16 |
| institution | OA Journals |
| issn | 2076-3417 |
| language | English |
| publishDate | 2025-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj-art-6a2945922e664a95828d35a5ff13bc162025-08-20T02:35:46ZengMDPI AGApplied Sciences2076-34172025-07-011513750810.3390/app15137508G-CTRNN: A Trainable Low-Power Continuous-Time Neural Network for Human Activity Recognition in Healthcare ApplicationsAbdallah Alzubi0David Lin1Johan Reimann2Fadi Alsaleem3Durham School of Architectural Engineering & Construction, College of Engineering, University of Nebraska-Lincoln, Omaha, NE 68182, USAGE Aerospace Research, Niskayuna, NY 12309, USAGE Aerospace Research, Niskayuna, NY 12309, USADurham School of Architectural Engineering & Construction, College of Engineering, University of Nebraska-Lincoln, Omaha, NE 68182, USAContinuous-time Recurrent Neural Networks (CTRNNs) are well-suited for modeling temporal dynamics in low-power neuromorphic and analog computing systems, making them promising candidates for edge-based human activity recognition (HAR) in healthcare. However, training CTRNNs remains challenging due to their continuous-time nature and the need to respect physical hardware constraints. In this work, we propose G-CTRNN, a novel gradient-based training framework for analog-friendly CTRNNs designed for embedded healthcare applications. Our method extends Backpropagation Through Time (BPTT) to continuous domains using TensorFlow’s automatic differentiation, while enforcing constraints on time constants and synaptic weights to ensure hardware compatibility. We validate G-CTRNN on the WISDM human activity dataset, which simulates realistic wearable sensor data for healthcare monitoring. Compared to conventional RNNs, G-CTRNN achieves superior classification accuracy with fewer parameters and greater stability—enabling continuous, real-time HAR on low-power platforms such as MEMS computing networks. The proposed framework provides a pathway toward on-device AI for remote patient monitoring, elderly care, and personalized healthcare in resource-constrained environments.https://www.mdpi.com/2076-3417/15/13/7508Human Activity Recognition (HAR)wearable healthcaresmart health monitoringedge computing in healthcareMEMS for healthcare AI |
| spellingShingle | Abdallah Alzubi David Lin Johan Reimann Fadi Alsaleem G-CTRNN: A Trainable Low-Power Continuous-Time Neural Network for Human Activity Recognition in Healthcare Applications Applied Sciences Human Activity Recognition (HAR) wearable healthcare smart health monitoring edge computing in healthcare MEMS for healthcare AI |
| title | G-CTRNN: A Trainable Low-Power Continuous-Time Neural Network for Human Activity Recognition in Healthcare Applications |
| title_full | G-CTRNN: A Trainable Low-Power Continuous-Time Neural Network for Human Activity Recognition in Healthcare Applications |
| title_fullStr | G-CTRNN: A Trainable Low-Power Continuous-Time Neural Network for Human Activity Recognition in Healthcare Applications |
| title_full_unstemmed | G-CTRNN: A Trainable Low-Power Continuous-Time Neural Network for Human Activity Recognition in Healthcare Applications |
| title_short | G-CTRNN: A Trainable Low-Power Continuous-Time Neural Network for Human Activity Recognition in Healthcare Applications |
| title_sort | g ctrnn a trainable low power continuous time neural network for human activity recognition in healthcare applications |
| topic | Human Activity Recognition (HAR) wearable healthcare smart health monitoring edge computing in healthcare MEMS for healthcare AI |
| url | https://www.mdpi.com/2076-3417/15/13/7508 |
| work_keys_str_mv | AT abdallahalzubi gctrnnatrainablelowpowercontinuoustimeneuralnetworkforhumanactivityrecognitioninhealthcareapplications AT davidlin gctrnnatrainablelowpowercontinuoustimeneuralnetworkforhumanactivityrecognitioninhealthcareapplications AT johanreimann gctrnnatrainablelowpowercontinuoustimeneuralnetworkforhumanactivityrecognitioninhealthcareapplications AT fadialsaleem gctrnnatrainablelowpowercontinuoustimeneuralnetworkforhumanactivityrecognitioninhealthcareapplications |