AI Clinical Decision Support System (CDSS) for Teleconsultations in eSanjeevani, India Telemedicine Service: A Prospective Implementation Study

BackgroundMost Western clinical decision support systems (CDSS) fail to contextualise to Indias healthcare realities, limiting their adoption by healthcare workers (HCWs) and patients. With the rapid evolution of Indias telemedicine ecosystem, there is an urgent need for an indigenous, validated, and scalable AI-based CDSS integrated within national platforms to improve diagnostic accuracy and care delivery. MethodsThis study, conducted between 2022-2024 by an AI Centre of Excellence of the Government of India, focused on developing, validating, and implementing a knowledge-based CDSS symptom entry Physician Assistance Form (PAF) within eSanjeevani--Indias national teleconsultation platform. The study was conducted in three phases, each of which was interdependent, overlapping, and dynamic in nature. Phase 1The AI system development utilised retrospective eSanjeevani 1.0 data (64 million consultations), extracting 0.22 million SNOMED CT-aligned records to identify 29,000 unique symptoms, which were refined to 115 for model training in the initial plan during 2022. The upgraded eSanjeevani 2.0 (2023) incorporated AI-based differential diagnosis, which was later expanded to include 300 symptoms (2025) with branching logic that integrated patient age, gender, language, and multidimensional symptom attributes. Phase 2Expert clinicians validated the symptom repository, logic flow, and AI-generated diagnoses. Phase 3The validated CDSS was implemented in eSanjeevani 2.0, providing real-time differential diagnosis and departmental recommendations during assisted and non-assisted teleconsultations. FindingsIntegration of AI-CDSS improved structured data capture, enhanced diagnostic precision, and streamlined patient triage within teleconsultations. InterpretationIndias AI-CDSS initiative represents the first government-supported, large-scale CDSS integration in a developing country, offering a replicable, ethical, and contextually grounded model for other LMICs to advance equitable and quality telehealth services.