A Vision-Based System for Early Screening and Monitoring of Chronic Disease Risk Using Emotion and rPPG Signals

Abstract

Chronic diseases represent a growing global health burden and often progress unnoticed before clinical diagnosis, creating a need for accessible early screening solutions. This paper presents a vision-based framework for early screening and monitoring of chronic disease risk using physiological and emotional signals captured through consumer-grade smartphone cameras. The proposed system combines facial emotion recognition and remote photoplethysmography (rPPG) within a step-based assessment pipeline. Facial analysis using the front camera estimates emotional states associated with stress, while fingertip-based rPPG signals acquired through the rear camera with flashlight assistance enable heart rate estimation and mean green channel analysis. These physiological and affective features are then integrated to generate a consolidated health risk profile suitable for longitudinal monitoring. Experimental observations indicate that combining emotional probabilities with rPPG-derived heart rate metrics improves the reliability of early health risk assessment. The framework provides a non-invasive, real-time, and scalable approach for continuous preventive health monitoring.

Introduction

The text presents a vision-based, non-invasive health monitoring system aimed at early detection of chronic disease risks such as cardiovascular and stress-related disorders. It highlights that many chronic conditions develop silently over time, and traditional monitoring methods like wearable devices and clinical tests can be inconvenient, costly, and unsuitable for continuous tracking. As an alternative, the study focuses on using camera-based health assessment through computer vision and machine learning.

The proposed system combines two key approaches: facial emotion recognition (to estimate stress and psychological states) and remote photoplethysmography (rPPG) using video signals (to estimate heart rate from skin color variations). These two streams are fused to generate a unified health risk profile for longitudinal monitoring. The system is designed to work using standard smartphone or webcam cameras, making it scalable and accessible.

The literature review shows that while rPPG and facial emotion analysis have individually advanced significantly, most existing systems lack integration of physiological and emotional signals. Challenges such as noise, lighting variations, motion artifacts, and the absence of multimodal fusion remain key limitations in current research.

The proposed method addresses these gaps by introducing a modular framework consisting of face detection, signal enhancement, rPPG extraction, emotion recognition, and a stress-mapping mechanism. Emotion outputs are converted into a stress index and combined with physiological signals to support early health risk prediction.

Conclusion

This work presents a vision-based framework for early screening and continuous monitoring of chronic disease risk by jointly analyzing facial emotion cues and remote photoplethysmography (Rppg) signals acquired through commodity cameras. The system integrates three core components: real-time face detection and emotion recognition, fingertip-based Rppg signal extraction using camera and flashlight interaction, and consolidated physiological–behavioral result visualization. Experimental observations indicate that emotion probabilities and stress-related indicators provide valuable contextual information, while Rppg-derived heart rate and waveform characteristics offer direct insight into cardiovascular state. Although simple intensity-based Rppg methods enable efficient real-time estimation, their reliability depends strongly on acquisition conditions, motivating the need for confidence-aware measurement and signal-quality validation.

References

[1] S. N. Kobel, C. Wuerich, A. L. Ernst, E. Fusshöller, J. N. Grüter, J. Händeler, C. Wiede, and K. Seidl, “Influences on rPPG-Based Spatial Blood Perfusion Maps,” Current Directions in Biomedical Engineering, vol. 10, no. 2, pp. 123–127, 2024. [2] J. Sun, Z. Zhang, J. Liu, L. Zhou, and S. Hu, “Heart Rate Estimation Algorithm Integrating Long and Short-Term Temporal Features,” Mathematics, vol. 12, no. 21, Art. no. 3444, 2024. [3] G. de Haan and V. Jeanne, “Robust pulse rate from chrominance-based rPPG,” IEEE Transactions on Biomedical Engineering, vol. 60, no. 10, pp. 2878–2886, 2013. [4] W. Wang, A. C. den Brinker, S. Stuijk, and G. de Haan, “Algorithmic principles of remote PPG,” IEEE Transactions on Biomedical Engineering, vol. 64, no. 7, pp. 1479–1491, 2017. [5] X. Niu, H. Han, S. Shan, and X. Chen, “Continuous heart rate measurement from face: A robust rPPG approach with distribution learning,” IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019. [6] Z. Yu, X. Li, and G. Zhao, “Remote photoplethysmograph signal measurement from facial videos using spatiotemporal networks,” British Machine Vision Conference (BMVC), 2019. [7] R. Li, Z. Wang, and J. Wu, “Non-contact physiological monitoring using facial videos for health assessment,” IEEE Access, vol. 9, pp. 113452–113465, 2021.

Copyright

Copyright © 2026 Pranav Jyotish Maurya, Komal Chandel, Sharvari Sudhir Kawale, Harshal Hemantkumar Koli, Prof. Sanjay Jadhav. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.