Joint Patent Acquired for Treatment Planning Support System Utilizing Multi-AI Agents

Qsol Corporation (Fukuoka City, Fukuoka Prefecture; President: Ken Hirowatari) and Kumamoto University (Kumamoto City, Kumamoto Prefecture; President: Hisao Ogawa) have announced the joint acquisition of a patent (Patent No. 7835948) for a 'Treatment Support System.' This system, a result of joint research on generative AI in the medical field, uses multiple AI agents to support the formulation of treatment plans through collaboration.

Overview of the Patented Technology This patent covers a system that uses patient medical history and test results to configure multiple AI agents, each virtually assuming the roles of medical professionals such as internists, surgeons, pharmacists, and clinical engineers. These agents evaluate and output treatment proposals from their respective professional perspectives before integrating them into a final proposal. The agents collaborate in a workflow-oriented manner, replicating the検討 (deliberation) process seen in multi-professional medical conferences. The system is designed with the premise of physician verification and modification, aiming to support decision-making rather than replacing it.

Features - Multi-AI agent structure with clearly defined medical professional roles - Workflow-based design mimicking multi-professional clinical conferences - Use of RAG (Retrieval-Augmented Generation) to reference medical guidelines, drug inserts, and past cases - Safety considerations through 'Human in the Loop' verification by physicians

Future Prospects This patent is a product of joint research between the Kumamoto University Hospital Medical Information Management Planning Department and Qsol. Future plans include evaluating potential applications in medical operations, such as treatment planning support, based on real-world clinical needs. Qsol and Kumamoto University will continue to collaborate to address societal challenges in healthcare, supporting physician workload reform.

(Diagram: Image of the Treatment Planning Support System, showing multiple AIs like physician-AI and pharmacist-AI deliberating to propose a plan to the user/physician.)

Key Generative AI Technologies - RAG: Provides expert information to generative AI to enable professional responses. - AI Agent: Autonomously plans tasks, makes decisions, and takes actions to solve problems. - Multi-AI Agent: Multiple agents work together to solve complex problems.

Development was conducted in compliance with relevant regulations, including ethical guidelines for life science and medical research, and safety management guidelines for medical information systems, following approval by an ethics committee. Electronic health records are used only as input for support and are never used to train the generative AI.

About Kumamoto University A comprehensive university with high global recognition in education and research, contributing to improved healthcare quality through advanced research and community collaboration.

About Qsol Corporation Since its founding in 1987 as a member of the Kyushu Electric Power Group, Qsol has focused on system development and operation for stable power supply, offering unique solutions for various challenges using its expertise and new technologies.