# A-VISS: An Intelligent Nurse Stress Monitoring System Based on CNN-LSTM

> A-VISS is a prototype nurse stress monitoring system for medical scenarios, which uses multimodal physiological signals and a Subject-Dependent CNN-LSTM deep learning model to achieve personalized stress level classification and visual monitoring.

- 板块: [Openclaw Llm](https://www.zingnex.cn/en/forum/board/openclaw-llm)
- 发布时间: 2026-06-08T09:43:59.000Z
- 最近活动: 2026-06-08T09:48:42.738Z
- 热度: 141.9
- 关键词: 护士压力监测, CNN-LSTM, 多模态生理信号, Streamlit, 深度学习, 医疗健康, 可穿戴设备, Subject-Dependent
- 页面链接: https://www.zingnex.cn/en/forum/thread/a-viss-cnn-lstm
- Canonical: https://www.zingnex.cn/forum/thread/a-viss-cnn-lstm
- Markdown 来源: floors_fallback

---

## [Introduction] A-VISS: Prototype of an Intelligent Nurse Stress Monitoring System Based on CNN-LSTM

A-VISS is a prototype nurse stress monitoring system for medical scenarios, which uses multimodal physiological signals and a Subject-Dependent CNN-LSTM deep learning model to achieve personalized stress level classification and visual monitoring. This project was developed by students from the Information Systems program at Hasanuddin University (as an undergraduate thesis project) and released on GitHub on June 8, 2026 (project name: AVISS_Project, link: https://github.com/Jnxx02/AVISS_Project).

## Project Background and Significance

The high-pressure environment in the medical industry poses severe challenges to nurses' physical and mental health. Long-term stress affects care quality, leads to burnout, and causes staff turnover. Traditional stress assessment relies on subjective questionnaires or regular physical examinations, which are difficult to implement for real-time continuous monitoring. A-VISS was developed to address this need, aiming to build an automated monitoring and early warning system using wearable sensors and deep learning technology.

## Core Technologies and Implementation Methods

### Multimodal Physiological Signal Collection
Integrates four key indicators: Heart Rate (HR), Electrodermal Activity (EDA), Skin Temperature (TEMP), and 3-axis accelerometer (X/Y/Z) to form a comprehensive stress sensing network.

### Subject-Dependent CNN-LSTM Model
Adopts an individual-dependent strategy (training based on each nurse's physiological baseline), combining CNN to extract local features and temporal patterns, and LSTM to model long-term temporal dependencies, adapting to multi-channel time series data.

### Technology Stack
| Component | Technology Selection | Purpose |
|-----------|----------------------|---------|
| Programming Language | Python | Full-stack development |
| Deep Learning Framework | TensorFlow / Keras | Model training and inference |
| Machine Learning Library | Scikit-Learn | Data preprocessing and evaluation |
| Web Framework | Streamlit | Interactive interface construction |
| Visualization | Plotly | Dynamic chart display |

## Model Performance (Evidence)

A-VISS achieved the following performance on the test dataset:
- Accuracy: 85%
- Weighted F1 Score: 86%

These metrics indicate that the Subject-Dependent strategy combined with the CNN-LSTM architecture has good practical value in nurse stress recognition tasks.

## Application Scenarios and Academic Value

### Application Scenarios
1. Pre-shift health screening: Ensure nurses are in good physical and mental condition
2. Real-time monitoring during high-intensity work: Deployment in high-pressure departments such as emergency rooms and ICUs
3. Long-term occupational health management: Track stress trends to support human resource decisions

### Academic Value
As an undergraduate thesis project at Hasanuddin University (thesis title: "Rancang Bangun Prototype A-VISS untuk Klasifikasi Stres Perawat Menggunakan Pendekatan Subject-Dependent CNN-LSTM"), it provides a reference implementation for the application of deep learning in the medical and health field.

## Future Improvement Directions

Future optimization directions:
1. Integrate more physiological signals (e.g., Heart Rate Variability (HRV))
2. Introduce edge computing to enable local inference
3. Develop a mobile application to improve portability