Using Hidden Markov Models and Rule-based Sensor Mediation on Wearable eHealth Devices

Improvements in sensor miniaturization allow wearable devices to provide more functionality while also being more comfortable for users to wear. The Samsung Simband©, for example, has 6 different sensors Electrocardiogram (ECG), Photoplethysmogram (PPG), Galvanic Skin Response (GSR), Bio-Impedance (Bio-Z), Accelerometer and a thermometer as well as a modular sensor hub to easily add additional ones. This increased number of sensors for wearable devices opens new possibilities for a more precise monitoring of patients by integrating the data from the different sensors. This integration can be influenced by failing or malfunctioning sensors and noise. In this paper, we propose an approach that uses Hidden Markov Models (HMM) in combination with a rule-based engine to mediate among the different sensors’ data in order to allow the eHealth system to compute a diagnosis on the basis of the selected reliable sensors.
We also show some preliminary results about the accuracy of the first stage of the proposed model.