MedAI: A Safety-Oriented Health Information Prototype

MedAI is an early-stage, safety-oriented health-information prototype designed to help users organize symptom context, retrieve biomedical information, and prepare for clinician visits. It is intended to support health-information access rather than diagnose medical conditions. The prototype combines symptom-context handling, biomedical information retrieval, and AI-generated responses within a health-information workflow. A developer-reported evaluation included 42 tests covering safety and input handling, retrieval and application behaviour, AI behaviour, privacy, security, and reliability. Two patient-report PDF files were also reviewed retrospectively. The developer reported that all 42 tests passed. However, detailed test inputs, expected and observed outputs, acceptance criteria, and execution logs were unavailable for independent verification. The limited review of two patient-report PDFs does not establish diagnostic accuracy, clinical validity, or patient benefit. This preprint describes an early-stage prototype. Reproducible testing, independent evaluation, privacy and security assessment, and appropriately governed clinical validation are needed before clinical effectiveness can be claimed.