Handling uncertainty in eHealth sensors using fuzzy system modeling

In remote health multiple sensors are attached to a patient and collected data is transferred to the cloud, which is used by the physician for diagnosis. These sensors are prone to random errors. Challenge is to provide correct information since any inaccuracy in the information leads to the incorrect diagnosis of the patient. Any abnormal condition recorded by a single sensor can be cross-checked by matching it against another sensor recording different vital signs. The decisions of the doctors rely upon whatever the sensor is streaming currently, because matching against other vitals has to be done at real-time. Since fuzzy-based system works with the imprecise dataset and merges them in ranges, it favours the decision of a medical practitioner in remote health. Therefore, in this paper, we discuss how to reduce uncertainty from the remote health sensors using fuzzy modelling system. We discuss some use cases to simulate remote health scenario with fuzzy inferencing system and obtain acceptable output in presence of random errors. We also compare our proposed model with basic (statistical) and advanced (context-aware) models to show its performance exceeds the other two. First, the statistical model needs more data set than our proposed model and second context-aware model may not correctly detect random error from the current context.