Ergoguard a low cost non wearable machine learning based posture enhancement system

Poor posture, particularly during prolonged sitting, is a significant health concern, contributing to musculoskeletal disorders, chronic pain, and reduced quality of life. With nearly 40% of children and many adults affected, there is a pressing need for accessible, user-friendly solutions to monitor and correct posture in real-time. Traditional systems often rely on costly or uncomfortable wearable devices, limiting their practicality. This paper introduces ErgoGuard, a cost-effective, non-wearable posture monitoring system designed to provide real-time feedback and promote healthier sitting habits. The study aims to develop a scalable solution using Force Sensing Resistors (FSRs) and machine learning to classify sitting postures accurately. ErgoGuard integrates ten FSRs on a chair’s seat pan and backrest, connected to a Raspberry Pi 4 for data processing. Pressure distribution data was collected from five participants of diverse age groups and body sizes in various sitting positions (upright, slouched, crossed-leg). The data was preprocessed, normalized, and analyzed using models like Support Vector Machine (SVM), Random Forest, and Gradient Boosting Classifier. Results show ErgoGuard achieves robust posture classification, with Random Forest achieving the highest macro-average F1 of 0.87, indicating strong performance across multiple posture classes. The non-wearable design ensures user comfort, while machine learning enables scalable, accurate monitoring. In conclusion, ErgoGuard offers a practical, innovative solution for posture correction, addressing limitations of existing systems. Its cost-effective design and real-time feedback make it suitable for ergonomics, healthcare, and workplace wellness. Future work will explore heat maps to enhance classification performance and provide deeper insights into posture transitions, laying the foundation for advanced, user-centric systems.