Patient Monitoring Framework for Smart Home Using Smart Handhelds

Nowadays, smart handhelds are becoming essential for activity monitoring due to its ubiquitous nature. The Smartphone sensors include accelerometer, gyroscope and wireless interfaces making the task more affordable and user friendly. Though a few patient monitoring frameworks are available, most of them take images to monitor activity while taking images depend heavily on lighting conditions. Only a few works can be found where input from ambient sensors is taken into account when body sensors give anomalous result and even fewer that collects data from inertial sensors of smartphones. However, in this work we propose a patient monitoring framework, that combines sensor data from smartphones and body sensors and applies rule based classification technique to detect abnormal activities. It generates alarm and notifies the care giver if any abnormal or critical situation occurs. Major challenges for this framework include combination of data from several types of sensors to identify activities using inexpensive time domain features to make the system cost effective and convenient to use. Experimental results show that incorporation of body sensors improve accuracy of activity monitoring.