Building Electronic Health Record using Voice Recognition and Big Data Techniques

Abstract

Electronic Health Record (EHR) is a foundation for any intelligent diagnose and medical data analytics. In our previous research, we discovered that the existing EHR in clinic mainly manually created by physicians are seriously fraud. They are too sample, containing serious mistakes and far from accurate. It causes great difficulties for any intelligent diagnose and medical data analyses. This paper reports our efforts in an attempt to automatic and semi-automatic EHR creation. We have adopted an advance AI techniques in voice recognition and text mining in order to build a prototype of “doctor’s assistant”, which can record conversations between a doctor and a patient, then the recorded voice file is further converted into a text file. With text mining techniques, the key symptom described by a patient is abstracted and converted into structured data. EHR will be generated automatically for doctors’ approval. This will greatly reduce doctors’ workload and increase the accuracy and quality of EHR.