Kansai Medical University and IBM Japan Co-develop Healthcare AI Common Platform

Details

Kansai Medical University and IBM Japan co-developed the "Healthcare AI Common ICT Platform" aimed at integrating systems for hospital operation and clinical support. The background includes overcoming difficulties such as the time-consuming and costly process of connecting individual AI engines to electronic health records (EHRs). Furthermore, concerns were raised regarding the risk of sensitive patient data being improperly handled on overseas servers, and questioning business models where vendor knowledge is not returned to medical institutions. Technically, recognizing the limitations of traditional firewall boundary security, the platform adopts a "zero-trust network" structure to safely connect external networks and hospital internal systems. It also built a "data lakehouse" to consolidate multiple siloed system data. By introducing IBM Japan's server compatible with international standards HL7/FHIR, it enables handling clinical data from different vendors under one standardized model. As the first application utilizing this platform, a "Generative AI Summary Creation Support Application" has been implemented. This reduces the time required for nurses to create nursing summaries from approximately 35 minutes of manual entry to just seconds with a button press, saving about 40 minutes per patient (estimated at 7,000 hours annually). It also streamlines physician handover tasks.