Intelligent Posture Training: Machine-Learning-Powered Human Sitting Posture Recognition Based on a Pressure-Sensing...

Area of Science:

• Biomedical Engineering
• Human-Computer Interaction
• Machine Learning

Background:

• Poor sitting posture is a prevalent issue in modern workplaces, leading to musculoskeletal discomfort and health problems.
• Existing posture monitoring solutions often lack accuracy, real-time feedback, or personalized recommendations.
• The integration of IoT devices and machine learning offers a promising avenue for intelligent health monitoring and intervention.

Purpose of the Study:

• To develop and validate an intelligent posture training system utilizing an IoT cushion for precise, real-time sitting posture monitoring.
• To assess the system's accuracy in recognizing various sitting postures and seated stretches using supervised machine learning.
• To propose a novel data-driven stretch recommendation system aligned with physiotherapy standards.

Main Methods:

• Development of the LifeChair IoT cushion equipped with pressure sensors to capture user body data.
• Application of supervised machine learning algorithms for analyzing sensor data and recognizing sitting postures and seated stretches.
• Validation of the system's performance across diverse real-world workplace environments and analysis of factors influencing accuracy, such as Body Mass Index (BMI) variations.

Main Results:

• The system achieved high accuracy, exceeding 98.82% for recognizing 15 distinct sitting postures and 97.94% for identifying six seated stretches.
• Significant impact of user BMI divergence on machine learning-based posture recognition accuracy was identified.
• Successful validation of the system's performance in five different workplace settings.

Conclusions:

• The developed system provides an accurate and effective solution for intelligent posture training and monitoring.
• The study highlights the importance of considering user-specific factors like BMI for optimizing posture recognition algorithms.
• The proposed smart posture data-driven stretch recommendation system represents a significant advancement in personalized physiotherapy and ergonomic interventions.