Evaluation and significance of a digital assistant for patient history-taking and physical examination in telemedicine.

Introduction: Assisted history-taking systems can be used in provider-to-provider teleconsultations to task-shift the collection of evidence-based medical history and physical exam information to a frontline health worker. We developed such a task-shifting digital assistant, called 'Ayu', for nurses in rural India to collect clinical information from a patient and share it with a remote doctor to arrive at an accurate diagnosis and triage decision. Materials & Methods: We evaluated the ability of the task-shifting digital assistant to collect a comprehensive patient history by using 190 standardized patient case studies and evaluating the information recall of the assistant by a skilled clinician. Following this, we tested the ability of nurses to use the system by training and evaluating the system's accuracy when used by 19 nurses in rural Gujarat, India. We also measured the diagnostic and triage accuracy based on the generated history note. Finally, we evaluated the system's acceptability by using the Technology Acceptance Model framework. Results: Ayu could capture 65% of patient history information and 42% of physical exam information from patient case studies. When used by nurses, the mean accuracy of the generated clinical note was 7.71 ± 2.42. Using the information collected by a nurse using Ayu, a primary care physician could arrive at the correct diagnosis in 74% of cases, and correct triage decision in 88% of cases. Overall, we saw a high acceptability from nurses to use the system. Conclusions: Ayu can capture an acceptable proportion of clinical information and can aid in collecting an evidence-based medical history by task-shifting some of the early investigational steps. Further development of Ayu to increase its information retrieval ability and ease of use by health workers is needed.