AI in Medicine News — 2026-10-21 (Wed)

AI and Medical Latest News

Overview

The focus is on the actions of regulatory authorities. The FDA and MHRA discussed reviewing the regulatory framework for AI-generated medical devices, with FDA officials recognizing the need for "evaluation similar to that used by physicians." Among the approvals, it was reported that FDA clearance of an AI tool for electrocardiogram AI and an AI tool for evaluating breast cancer recurrence using MRI was reported. In drug discovery, OpenAI's acquisition of Orchestra Bio, AstraZeneca and Oracle's collaboration for clinical trial patient registration was announced, and in Japan, it was reported that Gunma University Hospital and Saiseikai Kawaguchi General Hospital have introduced generative AI and transcription.

Diagnosis/Image

- Spectral CT and AI: Applied Radiology has published an explanatory article that deals with the clinical benefits of spectral CT, the role of AI, and the future of CT images (at the commentary/review stage. The device name and accuracy figures are not listed in the headline). - 510(k) for ECG AI: MobiHealthNews brief reports on FDA 510(k) clearance for ECG AI (clearance coverage; item name not listed in headline). The brief includes information related to smart rings in China. - Rejection warning after heart transplantation: Cardiovascular Business reports that electrocardiogram evaluation using AI may lead to improved warning of rejection. At the headline stage, it only states "possibility" and does not specify accuracy figures or verification stage (preclinical/clinical). - The state of radiology: Inside Precision Medicine publishes an essay that goes beyond detection and reconsiders radiology in the AI era (opinion/discussion stage). - AI in Oncology: CancerNetwork publishes explanatory article addressing impact on clinical trials, workflow, and outcomes (review stage; no specific trial results reported).

Clinical decision support

- Pathological diagnosis support: Medical Xpress reports that AI will help pathologists detect harmful lesions that cannot be seen with the naked eye. The distinction between the stage of evidence (research/clinical implementation) is not stated in the heading. - Analysis of adverse reactions in traditional Chinese medicine: A research report (Bioengineer.org, research stage) that uses machine learning and multi-omics to extract toxic signals from adverse reactions in traditional Chinese medicine. - ATA Action's opinion to the FDA: It is reported that the response regarding generated AI medical devices was based on ``two-axis risk assessment'', ``PCCP (planned change control plan)'', and ``manufacturer's responsibility'' (HIT Consultant, an industry group's opinion statement). - Funding brief: MobiHealthNews reports on funding for cancer-related and general practitioner (GP) AI. The amount and company name are not listed in the headline. - Doximity: Kalkine Media reports on the company's AI-related business trends. Evaluation of stock market prices is outside the scope of this report.

Generation AI and medical documents/medical records

- MEDITECH and Lifemed AI partnership: Partnership and coverage (HIT Consultant) aimed at helping hospitals address insurer claims denials. The applicable area is billing and administrative documents. No indication of indication for diagnosis or treatment. - 6 AI Scrivers: MedicalResearch.com introduces 6 AI Scrive (medical record creation support) software products for clinicians (introduction/comparison article. No evaluation data included). - Workflow Commentary: Healthcare IT Today published an opinion piece arguing that purchasing additional individual AI tools (point solutions) alone will not solve medical workflow problems.

Hospital management/medical provision system

- NTL Healthcare's cervical cancer screening AI: Reportedly migrated to on-premises environment (remio). It has been pointed out that security depends on the management of the facility (clinic). Numerical values such as detection accuracy are not listed in the heading. - Sunlight AI's funding: reportedly raised $17.4 million to expand operational infrastructure and collections capabilities for independent practitioners (HIT Consultant).

Drug discovery/clinical trials

- OpenAI's Acquisition of Orchestra Bio: Objectives and Coverage to Strengthen Biotech Business (TechCrunch, TodayPress.tv). The development stage (preclinical/Phase I, etc.) of the pipeline to be acquired is not stated in the headline. - AstraZeneca and Oracle collaboration: Launch and coverage of global AI collaboration to speed patient enrollment for clinical trials (Dealroom). The initiative is in the clinical trial implementation stage. The phase of the target clinical trial is not listed in the heading. - Sinopia Biosciences: Reportedly received a $2.5 million grant from the NIH (National Institutes of Health) to promote AI drug discovery (citybiz. Grant acquisition = research stage; preclinical/clinical distinction not stated). - Patient Recruitment Explanation: 36Kr has published an explanatory article dealing with the use of AI in patient registration for clinical trials.

Medical device approval/regulation

- AI tool to assess breast cancer recurrence with MRI — FDA clearance (approval): Cure Today reports. Item name and classification code are not listed in the heading. - EKG AI — 510(k) Clearance (Approved): Reported by MobiHealthNews Brief. The item name is not listed in the heading ("2.

Diagnosis/Image

”). - FDA/MHRA — Regulatory framework for generative AI (guidance under consideration): RAPS reports that officials from both agencies are eyeing a streamlined regulatory framework for generative AI. No formal guidance issued. - FDA staff opinion (opinion, not guidance or approval): MedCity News reports that FDA staff recognizes that generative AI-powered tools need to be evaluated in the same way as doctors. Name is not listed in the heading.

Safety/Error/Bias

- AI chatbots and unnecessary tests: EurekAlert! introduces research. Although the AI chatbot can recognize unnecessary medical tests, it is reported that the response is unstable in situations where human doctors have differing opinions (research stage; quantitative values are not listed in the headline). - Biosecurity report: News9live presents a safety report on whether AI can create dangerous viruses. The issuer of the report is not mentioned in the heading. - Demonstration status in the UAE: Fast Company Middle East analyzes that while the UAE has become a testing ground for medical AI, the actual verification will begin after the pilot ends. - Delay in safety measures: AfroTech reports the opinion of Dr. Tiffani Bright, who points out that medical AI is progressing faster than the development of safety measures (see "11." for details).

Remarks from VIPs

”).

Research results (major journals)

- Scientific American: A report highlighting research into whether it is possible to predict death from cancer from a single facial image. Numerical values such as prediction accuracy are not listed in the heading. *The details of cancer research are covered in the research section, so only a summary of the headings is provided here. - USA Today: Report on the development of an AI photo tool that estimates age and cancer survival. *Same as above, details are covered in the research section. - Motor Illustrated's AI articles (related to Japanese automakers) are outside the medical field and are therefore not covered by this section.

Japanese movements

- Gunma University Hospital: Introduced generative AI to streamline medical document creation (Nico Nico News). The goal is to reduce the burden on doctors and create more time for them to "face-to-face with patients." Quantitative effect values are not listed in the heading. - Saiseikai Kawaguchi General Hospital: Introduced AI transcription "Notta" and reduced the time required to create minutes from 90 minutes to 10 minutes (Medical Tech News). It is intended for meeting minutes, not medical records. - Research institutes participating in Lapidus: Reported to be developing medical AI (Shimotsu Shimbun). Details of the development stage and target area are not listed in the heading. - Excluded: Hitachi's physical AI solution for the process manufacturing industry (Cloud Watch) and advance sales (PR TIMES) of the palm-sized AI server "miraiGen.Cube" are not covered in this section as they are not directly related to medical treatment and delivery systems.

Remarks from VIPs

No registration on the list of VIPs being tracked. The statements that can be confirmed from the headline are as follows.

- FDA official (name not listed in headline): Recognizes that medical tools equipped with generative AI need to be evaluated in the same way as doctors (reported by MedCity News). - FDA and MHRA officials: looking to apply streamlined regulatory framework to generated AI medical devices (RAPS coverage). - Dr. Tiffani Bright (Introduced as an AfroTech speaker. Title details are not listed in the headline): He is of the view that there is an opportunity to be involved in the current situation where medical AI is advancing faster than the development of safety measures (AfroTech report).

*This report is based on information derived from the headline. Details of accuracy, approved content, and development stage require confirmation of each original text and paper in each research section.