Sensor anomaly detection in wireless sensor networks for healthcare

Area of Science:

• Computer Science
• Biomedical Engineering
• Signal Processing

Background:

• Wireless Sensor Networks (WSN) are susceptible to sensor faults and inaccurate data, impacting critical applications like remote patient monitoring.
• Faulty sensor readings in healthcare can trigger false alarms, leading to unnecessary interventions and reduced quality of care.
• Reliable data is crucial for effective remote patient monitoring systems in healthcare.

Purpose of the Study:

• To develop and validate a novel approach for detecting sensor anomalies in WSN using physiological data.
• To effectively differentiate between true medical alarms and false alarms generated by faulty sensors.
• To enhance the reliability and accuracy of remote patient monitoring systems.

Main Methods:

• A novel sensor anomaly detection method is proposed, analyzing historical physiological data to predict sensor values.
• The predicted sensor value is compared against the actual sensed value.
• A dynamically adjusted threshold is used to identify anomalous sensor readings, distinguishing true from false alarms.

Main Results:

• The proposed approach was applied to real-world healthcare datasets.
• Experimental results show a high Detection Rate (DR) and a low False Positive Rate (FPR) compared to existing methods.
• The system effectively distinguishes between genuine and erroneous sensor data.

Conclusions:

• The novel sensor anomaly detection method significantly improves the reliability of WSN in healthcare applications.
• Accurate differentiation of alarms enhances the quality of remote patient monitoring and healthcare services.
• The proposed approach offers a robust solution for managing sensor faults in medical WSN.