talk2care

• Healthcare systems increasingly rely on digital records, yet most clinical interactions between doctors and patients remain unstructured and underutilized.

• This work proposes an Agentic AI–based clinical intelligence system that captures doctor–patient conversations (with explicit patient consent), converts speech into structured medical knowledge, and generates clinically meaningful summaries and personalized insights.

• The proposed system employs multiple specialized AI agents, each responsible for a distinct task, including consent validation, audio processing, speech-to-text transcription, medical entity extraction, clinical reasoning, summarization, personalization, and secure data storage.

• By decomposing the workflow into autonomous yet cooperative agents, the system improves reliability, auditability, and safety compared to monolithic AI approaches. The generated outputs include structured clinical notes for healthcare providers and simplified summaries for patients, enabling improved continuity of care and reduced documentation burden.

• Furthermore, the agentic framework enables longitudinal analysis of patient interactions, supporting early risk detection, personalized care recommendations, and proactive follow-up management while maintaining compliance with healthcare privacy and security standards. This approach demonstrates how agentic AI can transform unstructured clinical conversations into actionable medical intelligence, enhancing decision support without replacing clinician authority.