EER Calculation and DET Approximation in a Multi-Threshold Biometric System

Identity security in Health Information Systems is a major concern in order to guarantee patients confidentiality. Biometrics has been used in order to improve traditional identity authentication. In this work, we present a method to get the Detection Error Trade-off—DET—graph when two or more thresholds are involved in a biometric system. Consequently, we also provide a method to calculate the Equal Error Rate—EER—in a multi-threshold system. In general, we can calculate the EER visually through the DET graph or with a formula. However, the formula is applicable only with normal distributed data. Biometric data distribution is usually not normal; therefore, we cannot always use the formula to calculate the EER. We will be able to use this method in a biometric system with any type of data distribution and with any number of thresholds. The obtained results show that the difference of EER calculated with our method and the one calculated with the formula have a standard deviation of 0.62%. These findings will contribute in the medicine field to better ensure the health information privacy of the patient.