AI in Medicine — Cancer & Rare Disease — 2026-09-29 (Tue)

Latest news on AI and cancer treatment

Overview

Movements were noticeable in the field of image diagnosis and screening support (FDA approval of breast cancer screening AI, glioblastoma recurrence prediction, and endoscopic capsule AI). In the drug discovery field, we continue to invest in AI drug discovery platforms, such as utilizing Roche's autonomous AI drug discovery lab and raising funds for HITS from KAIST in South Korea. On the regulatory front, FDA approval for endoscopic capsule AI and Osaka University's oral cancer detection system were confirmed.

Breast cancer

- Lunit's INSIGHT MMG is being introduced for mammography interpretation in hospitals throughout the Stockholm region (PR Newswire/Korea Biomedical Review/Trend Hunter, multiple media outlets reported on the same incident). As the project is positioned as a region-wide screening program, numerical values such as accuracy and detection rate are not listed in the headline. - A research presentation by Sylvester, a cancer researcher at the University of Miami, on "Using AI in Cancer Survivorship" (target cancer types are not specified in the headline). - Boston University "AI Prognostics with Pathologist-in-the-Loop" = Research on AI prognostic prediction model with pathologist participation (target cancer types and evidence level are not listed in the heading).

lung cancer

- Roche focuses on leveraging autonomous AI drug discovery lab to develop KRAS G12C inhibitor divarasib (GenomeWeb). The development stage and clinical trial phase are not listed in the heading. *Headings related to esophageal cancer CT detection (Ynetnews) have been organized and listed under "7. Colorectal Cancer/Gastrointestinal System".

Malignant brain tumor

- A new AI tool has been reported for predicting the areas where glioblastoma is likely to recur after initial surgery (Medical Xpress). Whether the study was a single-center study or a multi-center study, or prospective/retrospective, is not stated in the heading.

Leukemia/lymphoma

No applicable information

Colon cancer/digestive organ

- AI study (Ynetnews) claims to have detected esophageal cancer from CT scans several months before diagnosis. The lead time for detection is only suggested, and the number of target patients, sensitivity specificity, and study design (preclinical/retrospective/prospective) are not listed in the heading. - FDA approves AI that presents abnormality candidates on swallowable camera (capsule endoscope) images (Stock Titan). See also “11. Regulations and Approvals” for details. - A report on the use of AI in hospitals (Newsweek) states that "if AI had not been introduced, patients would have died." Target cancer types and specificity of evidence are not listed in the heading.

Rare cancer

No applicable information

Diagnosis/Pathology/Image

- Improving the accuracy of skin cancer detection tools using AI and pointing out that the benefits are not uniform (Fast Company Middle East). Specific sensitivity, specificity, and target population are not listed in the heading. - GuideAI and vRad renew exclusive agreement for vascular AI image data (AuntMinnie). This is a business partnership related to image diagnosis infrastructure, rather than specific cancer types.

Treatment selection/drug discovery

- Roche utilizes autonomous AI drug discovery lab in development of divarasib (GenomeWeb). The stage (preclinical/clinical phase) is not stated in the heading. - Rakovina Therapeutics continues to design cancer drugs using AI (kalkine.ca). Details of the development stage are not listed in the heading. - HITS, an AI drug discovery company from KAIST, has raised approximately 18.3 billion won (equivalent to approximately 1.34 billion yen) and plans to start operating "AI Scientist" in November (finance.biggo.com). The story is about the platform stage, and there is no mention of specific drugs or clinical phases. - Efforts to build an AI infrastructure for cancer care in Somaliland (ASCO Connection). The main theme is access improvement in low resource environments. - OpenEvidence, Anthropic, and University of Pennsylvania Medical Institutions partner to expand AI access for doctors in Africa and Asia (Fierce Healthcare). The theme is not about specific cancer types, but about expanding access to clinical decision-making support AI.

Regulations/Approval

- FDA: Approved (clearance) AI that presents abnormality candidates in swallowable camera (capsule endoscopy) images. Details of item name and approval category are not listed in the heading (Stock Titan). - Osaka University's oral cancer early detection AI system receives approval (public recruitment guide). Approval entity and approval category (medical device approval under the Pharmaceutical and Medical Device Act, etc.) are not listed in the heading.

Clinical trial implementation status

No applicable information as individual data from ClinicalTrials.gov is not included in this message. No estimates will be made regarding the number of cases or registration status.

Japanese movements

- The AI system for early detection of oral cancer developed by Osaka University received approval (public recruitment guide). Details such as approval category, target patients, accuracy, etc. are not listed in the heading. *Other headlines from CHOSUNBIZ (Okuma Pharmaceutical's controlled release formulation design, Korean PE's AI infrastructure investment, TelePIX's satellite AI contract, Frontier AI resource intensive study) were excluded because they are related to South Korea and have nothing to do with cancer medicine.

Remarks from VIPs

- Khurram Khaliq Bhinder: Scholar-in-Training Award from AACR (American Association for Cancer Research), associated with research project “AI-PACED” (oncodaily.com). Details of target cancer types and research content are not listed in the heading. - Chandler Park, published a paper titled "What does AI need to do to improve oncology care?" (oncodaily.com). The title is not listed in the heading.

Excluded headlines (determined to be unrelated to cancer medical AI): Nvidia's share buyback, UltraSight (cardiac ultrasound AI), heart health monitoring AI article, Mitratech (legal AI acquisition), TelePIX satellite AI contract, Korean PE's AI infrastructure investment base, Korean government's frontier AI resource concentration consideration, Sheinelle Jones' personal article.

Latest news on AI and rare and incurable diseases

Overview

There has been a noticeable movement in the use of AI in the diagnostic field, with reports of AI applications in identifying causative gene candidates for rare neurodevelopmental diseases, classifying vasculitis, and screening for heart disease. In Japan, Kobe University reported on AI diagnosis of intractable diseases based on photographs of hands. There were no headlines regarding therapeutic drug development or systems.

Diagnosis of rare diseases

A team at Kobe University reported on their research using AI to diagnose intractable diseases from photographs of hands in a privacy-friendly manner (Sanyo Shimbun). Details such as target disease name, diagnostic accuracy, and number of target patients are unclear from the headline. Additionally, Tempus announced that it has received FDA clearance for its ECG-MR AI product, which is intended to identify undiagnosed mitral regurgitation (Business Wire). However, this is not for diagnosing rare diseases per se, but for screening for heart disease, and is still at the stage of obtaining clearance from regulatory authorities.

Gene analysis/genome

There was a report that a candidate gene for the cause of a rare neurodevelopmental disease was identified by combining genome analysis and AI (News-Medical). The research stage, number of cases, and target disease name are unclear from the headline, and it may be a case report or research stage. Codebreaker Labs announced seed funding to prepare data for genomic AI (AiThority). This is a business/financial move and does not indicate clinical results.

Development of therapeutic drugs

No applicable information

Nerve/muscular diseases

Regarding the ``rare neurodevelopmental disease'' reported in News-Medical, it is not possible to confirm from the headline that it corresponds to the list of 348 designated intractable diseases. Since there is no mention of the specific disease name that falls under the Ministry of Health, Labor and Welfare's group of neurological and muscular diseases (87 cases), it is not possible to confirm whether the disease is a designated intractable disease or not.

Immunology/Collagen disease

A study reported that AI classified rare vasculitis into three different types (ANCA Vasculitis News). The number of target patients and the stage of verification of the validity of the classification (single facility or multiple facilities, etc.) are unclear from the headline. Vasculitis is a disease that falls within the group of immune diseases designated by the Ministry of Health, Labor and Welfare as an intractable disease (Takayasu's arteritis, polyarteritis nodosa, microscopic polyangiitis, granulomatosis with polyangiitis, granulomatosis with eosinophilic polyangiitis, etc.), but since the specific name of the disease is not listed in the heading, it is not possible to confirm which designated intractable disease it corresponds to.

Rare metabolic/endocrine diseases

No applicable information

System/Patient Support

No applicable information

Japanese movements

A Kobe University team announced a privacy-friendly method for AI diagnosis of intractable diseases using hand photos (Sanyo Shimbun). The target disease and verification stage (case report or clinical trial) are unclear from the headline.

Remarks from VIPs

No applicable information