A Multimodal Closed-Loop Framework for Vital Sign Monitoring and Intelligent Diagnosis of Amusement Ride Passengers...

Area of Science:

• Biomedical Engineering
• Physiological Monitoring
• Machine Learning in Healthcare

Background:

• Amusement rides generate severe motion artifacts, degrading vital sign quality and physiological state recognition.
• Existing passenger monitoring systems struggle with high-dynamic conditions, impacting safety and diagnostic accuracy.

Purpose of the Study:

• To develop an engineering-ready, closed-loop framework for robust passenger monitoring and intelligent diagnosis in high-dynamic amusement ride environments.
• To integrate multimodal physiological and kinematic data for enhanced safety and early warning systems.

Main Methods:

• A multimodal sensing pipeline combining physiological (heart rate, SpO2) and kinematic (acceleration, angular rate) data.
• Wavelet denoising, coordinate frame unification, and a strapdown inertial navigation system for signal preprocessing.
• Development of three learning modules: heart rate fusion model, TAPNet for anomaly screening, and HS-BANet for arrhythmia classification, incorporating standards-driven rules (GB 8408-2018).

Main Results:

• Accurate heart rate estimation (RMSE ~1.2 bpm) and strong correlation (0.9928 training, 0.9865 testing).
• TAPNet achieved high accuracy (96.9% validation, 98.2% test) for kinematic anomaly recognition.
• HS-BANet demonstrated multi-class arrhythmia classification accuracy (92.37%) with high F1-score (86.87%).

Conclusions:

• The proposed two-stage multimodal framework provides fast, interpretable early warning and reliable decision-making under strong motion artifacts.
• The system balances responsiveness and diagnostic credibility, supporting practical safety early warning in amusement rides.
• Demonstrates potential for deployment-oriented operational support and enhanced passenger safety in dynamic environments.