Over the past decade, there has been a marked expansion of electronic medical record systems (EMR) that has largely been driven by the Health Information Technology and Clinical Health Act (HITECH). As adoption of EMR has increased nationwide, studies have shown decreasing levels of physician satisfaction and productivity.1 Previous studies have suggested that increases in administrative tasks are associated with time pressure and lower physician job satisfaction rates while adequate time for patient-physician interaction are associated with higher physician satisfaction.2,3 In some practice settings, nearly as much time is spent on documentation as direct patient care.4-6 The greatest obstacle limiting workflow speed relates to the data input process. Most electronic record systems require manual input of historical information and physical examination findings into structured forms within the electronic health record system. This requires a significant expenditure of
physician time and carries substantial costs in lost productivity. Our proposed project aims to utilize current automatic speech recognition technology and natural language processing techniques to facilitate the construction of an electronic medical record note and improve the order entry system. Specifically, we will explore the potential for state-of-the-art speech and language processing techniques to automate the note writing and order entry process using ambient audio recordings of the doctor-patient consultation. Our approach contrasts with current dictation systems in which physicians typically dictate the note following the visit. At the University of Southern California, most physicians manually input data or dictate into Cerner,
our EMR system. This is also true at other hospital systems that employ the same or other EMR solutions. This process by nature is inefficient, as note preparation is time-intensive, and the majorcontents of the note were verbalized during the visit. Instead, we aim to automatically infer a substantial portion of the clinic note and order contents from automated recognition and understanding of the conversation between doctor and patient during the actual visit. The funds for
this project will support data collection and an initial technical feasibility analysis.

NIH Funding Acknowledgment: Important - All publications resulting from the utilization of SC CTSI resources are required to credit the SC CTSI grant by including the NIH funding acknowledgment and must comply with the NIH Public Access Policy.