An effective and robust machine learning approach for automated human posture detection from IoTs module

People often do not notice their posture disorders. However, over time, poor posture can cause arm, head, waist, and back pain, nerve compression, muscle fatigue, and weakness. IoTs and machine learning based-applications that instantly detect posture disorders and provide information to the user can prevent such disturbances from occurring over time. In this study, healthy and unhealthy posture was automatically detected from posture position information obtained from an IoTs-based sensor module. Axis information obtained from the human knee and chest was used as the feature set. The size of the feature set was decreased with Chi-square and Decision Tree algorithms. Sleeping, sitting, and standing postures were classified as healthy and unhealthy with Support Vector Machine (SVM), k-Nearest Neighbor (KNN), and Naïve Bayes algorithms. The best accuracies were 100% for all situations.