Triaging Patients With Artificial Intelligence for Respiratory Symptoms in Primary Care to Improve Patient Outcomes: A Retrospective Diagnostic Accuracy Study

Given the rise of widespread digitalization and increased reliance on specialized artificial intelligence (AI) modeling, your study — Triaging Patients with Artificial Intelligence for Respiratory Symptoms in Primary Care to Improve Patient Outcomes — presents a means of revolutionizing primary care delivery, striving towards both optimal efficacy and efficiency. Creation and validation of a successful pre-appointment triage system between the patient and physician would enable physicians to immediately direct their attention to patients presenting with highest urgency, while it would also spare non-urgent patients from the hassle and possible anxiety from an unnecessary medical visit. The respiratory symptom triage model’s (RSTM) precision in symptom-based triage is strongly confirmed by the ability to correctly identify a substantial number of patients as low- versus high-risk, using physician notes. As the authors address in the Clinical Implications section, an opportunity exists for patients to provide responses to symptom-based questions to predict their likelihood and/or urgency of needing a face-to-face appointment. Have the authors considered how such a patient generated RSTM platform would be developed, tested, and validated? With the acknowledgement that this research field is pioneering and currently unexplored in daily practice, perhaps further research focusing on methods of input assessment and improvement would bring key insights for integration of the triage web-tool in routine primary care practice.