Speech Recognition for Medical Conversations Health Record (MCHR)

This paper provides an overview of major technical perspectives and recognizes the fundamental progress of the conversion of speech to text and also provides an overview of the techniques developed at each point of the speech-totext conversion classification, a system that automatically transcribes conversations between doctor and patient. MCHRsystems are knownto store clinical data to capture the patient's electronic medical health record (EMHR) across time (Rajkomar et al. in NPJ Digit Med 1, 18 (2018) [1]). It eradicates the need to manually monitor the previous paper medical records of a patient and helps ensure consistency and standardization of data. When extracting medical data, electronic medical health records are more efficient for analyzing potential patterns and long-term improvements in a patient's clinical condition. MCHR systems manage doctor-patient conversational data recording as both an audio and text for the use of doctors in health data analysis. These systems help in analyzing clinical data in later stages or match patterns for other patients with similar problems or symptoms. Additional usage is from the conversation we can extract diagnosis information, drug prescriptions and dosage details which can be presented to doctors for verification and approval. This will reduce doctors' time; nowadays in large hospitals, patient loads are high and doctors are unable to perform prescription entry using text input in hospital management systems, as it's a time-consuming process. This paper presents the state-of-the-art in voice recognition and discusses how to build a clinical prescribing system for healthcare domain.