AI-Driven Diagnostic Assistance in Medical Inquiry: ReinforcementLearning Algorithm Development and Validation

Background: For medical diagnosis, clinicians typically begin with a patient's chief concerns, followed by questions aboutsymptoms and medical history, physical examinations, and requests for necessary auxiliary examinations to gather comprehensivemedical information. This complex medical investigation process has yet to be modeled by existing artificial intelligence (AI)methodologies. Objective: The aim of this study was to develop an AI-driven medical inquiry assistant for clinical diagnosis that providesinquiry recommendations by simulating clinicians'medical investigating logic via reinforcement learning. Methods: We compiled multicenter, deidentified outpatient electronic health records from 76 hospitals in Shenzhen, China,spanning the period from July to November 2021. These records consisted of both unstructured textual information and structuredlaboratory test results. We first performed feature extraction and standardization using natural language processing techniquesand then used a reinforcement learning actor-critic framework to explore the rational and effective inquiry logic. To align theinquiry process with actual clinical practice, we segmented the inquiry into 4 stages: inquiring about symptoms and medicalhistory, conducting physical examinations, requesting auxiliary examinations, and terminating the inquiry with a diagnosis.External validation was conducted to validate the inquiry logic of the AI model. Results: This study focused on 2 retrospective inquiry-and-diagnosis tasks in the emergency and pediatrics departments. Theemergency departments provided records of 339,020 consultations including mainly children (median age 5.2, IQR 2.6-26.1years) with various types of upper respiratory tract infections (250,638/339,020, 73.93%). The pediatrics department providedrecords of 561,659 consultations, mainly of children (median age 3.8, IQR 2.0-5.7 years) with various types of upper respiratorytract infections (498,408/561,659, 88.73%). When conducting its own inquiries in both scenarios, the AI model demonstratedhigh diagnostic performance, with areas under the receiver operating characteristic curve of 0.955 (95% CI 0.953-0.956) and0.943 (95% CI 0.941-0.944), respectively. When the AI model was used in a simulated collaboration with physicians, it notablyreduced the average number of physicians' inquiries to 46% (6.037/13.26; 95% CI 6.009-6.064) and 43% (6.245/14.364; 95%J Med Internet Res 2024 | vol. 26 | e54616 | p. 1https://www.jmir.org/2024/1/e54616(page number not for citation purposes)Zou et alJOURNAL OF MEDICAL INTERNET RESEARCHXSL center dot FORenderX CI 6.225-6.269) while achieving areas under the receiver operating characteristic curve of 0.972 (95% CI 0.970-0.973) and 0.968(95% CI 0.967-0.969) in the scenarios. External validation revealed a normalized Kendall tau distance of 0.323 (95% CI 0.301-0.346),indicating the inquiry consistency of the AI model with physicians. Conclusions: This retrospective analysis of predominantly respiratory pediatric presentations in emergency and pediatricsdepartments demonstrated that an AI-driven diagnostic assistant had high diagnostic performance both in stand-alone use and insimulated collaboration with clinicians. Its investigation process was found to be consistent with the clinicians'medical investigationlogic. These findings highlight the diagnostic assistant's promise in assisting the decision-making processes of health careprofessionals