Novel automatic posture detection for in-patient care using IMU sensors

Abstract

Posture detection using Inertia Measurement Unit (IMU) has recently attracted great interests in healthcare research community. However, very few studies focus on the applications of this technology in the care of inpatients. This specific group of users, who are moderately to severely ill, have a distinct set of postures and activities that require special attentions and continuous monitoring from clinicians. In this paper, we present a novel methodology for automatic detection of postures for hospitalized patients using two wearable IMU sensors, with tri-axial accelerometers, attached at the chest and the abdomen respectively. The data were collected from participants who were carefully instructed to perform activities and attain postures that simulate those of hospitalized patients in real life. From the data retrieved, we performed orientation analysis for acceleration vectors and transition analysis for transitional activities between various postures. Both rule-based detection and Artificial Neural Network (ANN) for transition recognition achieved high accuracy. The results also showed that a combination of orientation and transition study could enhance the robustness of the detection algorithm. Due to its efficiency and simplicity, the proposed method could find its way into many applications that aim to improve the current state of inpatient healthcare.