Hitachi Develops AI Technology to Support Pre-preparation for Expert Panels in Cancer Genomic Medicine

Tohoku University Hospital, Hitachi, and Hitachi High-Tech have developed AI technology to support the pre-preparation of "Expert Panels (EP)"*, a specialized meeting for the medical interpretation of genetic test results in cancer genomic medicine*1. This technology refers to past physician comments, related knowledge, and literature databases to present points for confirmation and discussion at the EP, including potential treatment options, along with supporting evidence*3. This is expected to facilitate smoother confirmation and discussion at the EP, reduce the time required for information gathering and organization during pre-preparation, and alleviate the workload of physicians. Furthermore, the system is designed to operate on PCs within the hospital, assuming no transmission of sensitive information externally. In an evaluation using past cases from Tohoku University Hospital, over 80% of the points presented by this technology for EP confirmation and discussion, corresponding to treatment policies, matched the EP's review results. Moving forward, we aim to contribute to the widespread adoption and quality improvement of cancer genomic medicine by continuing verification and striving for technological advancement, thereby reducing the burden on medical facilities.

*1 Cancer Genomic Medicine: A type of personalized medicine that analyzes genetic mutations in a patient's cancer cells and selects the optimal treatment based on the results.

*2 Expert Panel (EP): In cancer genomic medicine, a specialized meeting where multiple specialists (such as medical oncologists, genetic diagnosticians, clinical genetic specialists, and pharmacists) gather based on a patient's genetic mutation results to comprehensively discuss and advise on optimal treatment policies and necessary additional tests. It plays a crucial role in supporting the realization of personalized medicine.

*3 The content presented by this technology is for discussion purposes only. The final decision on diagnosis and treatment policy rests with healthcare professionals. Treatment policies will not be determined solely based on the content presented by this technology.

Figure 1: Overview of the AI Technology Supporting EP Pre-preparation

*4 Knowledge Graph: A technology that structures and systematically utilizes knowledge extracted from physician comments and other sources by linking them.

Background and Challenges

Cancer Gene Panel Testing (CGP)*6 has been covered by insurance since 2019 for patients with advanced or recurrent solid tumors*5 for whom standard treatment is unavailable or has been completed. Consequently, the number of patients eligible for cancer genomic medicine is expected to expand approximately 2.5-fold over the five years from 2020 to 2025*7, leading to an increase in cases to be reviewed by EPs.

However, the pre-preparation for EPs involves not only reviewing test results but also extensive tasks such as collecting related information and organizing discussion points. Depending on the facility and case, this can take up to 1-2 hours per case and is sometimes performed after regular clinical duties, posing a challenge of increased workload for physicians. Furthermore, since EPs are operated by a limited number of specialists, including experts in molecular biology, there has been a demand for a system that can facilitate the review of evidence-based information while keeping the burden of pre-preparation manageable.

*5 Solid Tumors: A general term for cancers that form lumps (tumors) in the organs or tissues of the body. *6 Cancer Gene Panel Testing (CGP): A test that examines the DNA of cancer cells to comprehensively analyze multiple cancer-related gene mutations simultaneously.

*7 "Cancer Genomic Medicine and Cancer Gene Panel Testing" website (https://www.ncc.go.jp/jp/cancer_genome/index.html)