Patient Identity Verification Based on Physiological Signal Fusion

Patient identification is crucial in providing proper care in hospitals or other care-facilities. Failure to correctly identify patients can result in a variety of problems such as medication errors, transfusion errors, testing errors, and duplication of EHR records. The current solutions for identifying patients in a hospital setting rely on confirming the patient's identity several times, the use of labels, bar-codes, and RFID tags. However, these solutions are not always sufficient and patient identification errors are common. In this paper we present a patient identity verification approach that can complement existing patient identification solutions in the event that they cannot be relied upon. Our approach works by fusing common physiological signals (vital signs) already collected from the patients. The idea is to fuse two physiological signals electrocardiogram (ECG) with arterial blood pressure (ABP) to identify an individual patient. Existing identification/authentication solutions based on cardiac signals have been primarily tested on healthy patients whose signals exhibit normative rhythm and morphology. However, in our context not all patients can be expected to have normative cardiac signals as many of them may be suffering from ailments that affect the cardiac process. The fusion of multiple ECG and ABP allows us to ensure patient identification even when the patients have ailments that affect their cardiac rhythms. An evaluation of our approach showed that it is over 97% accurate in identifying patients with non-normative cardiac rhythms and morphology with over 99% accuracy for patients whose cardiac rhythms are normative. Further, our approach can identify a patient in as little as 3 seconds, which makes it practical in real-world scenarios.