AI in Medicine News — 2026-10-08 (Thu) Evening News

Overview

- The areas where the most notable movements were diagnostics/imaging (1 FDA clearance, 1 retracted paper), clinical decision support (integration during medical examinations, autonomous AI deployment in the NHS), operations (token cost, MyChart integration), drug discovery/clinical trials (test result prediction, shortened setup), and regulations (relationship between Utah's AI pilot and the FDA). - The information is only in the heading, and almost no numerical values such as accuracy or detection rate are shown. The stage of evidence is largely unknown.

Diagnosis/Image

- MRI2CT: AI-powered imaging platform reportedly receives FDA clearance (Imaging Technology News). The measurement target and performance level are unclear from the headline. - AI ECG (triage of acute myocardial infarction): European Medical Journal article discussing "promise, limitations and next steps". There is no specific accuracy in the headline. - Ovarian cancer AI study retracted: Study was retracted after it was discovered that the ultrasound dataset included thyroid images (Bioengineer.org). - Radiology training: Article on balancing clinical skills and AI literacy (Diagnostic Imaging). - Cardiac Wire: Articles summarizing cardiac rehabilitation, LAAC data, and diagnostic AI. The content is unclear, just the headline.

Clinical decision support

- Suki×UpToDate: Incorporating AI-based clinical decision-making support into medical examinations (Newsweek). - Skin Analytics (NHS England South East): Rolling out AI skin cancer check. The headline describes it as "the largest rollout of autonomous medical AI in NHS history". There are no performance numbers in the headline. - Breast Cancer Risk Prediction: Startup Direct to Patients Policy (STAT). - Patient safety in India: Book on AI and patient safety published as 1 in 10 patients are harmed during medical treatment (ANI News). - Triage Ethics: An essay that applies the ideas of 13th century theologians to AI triage ethics (Bioengineer.org).

Generation AI and medical documents/medical records

- Teladoc Health: Adds contactless vitals measurements and ambient AI scribe to virtual care platform (MedCity News). - Advocate: 7,000% increase in patients connecting MyChart and ChatGPT. Who pays for the costs is an issue (RamaOnHealthcare). - Healthleap: Funding for clinical AI screening for often overlooked conditions in hospitalized patients. There are two types, $38 million and $30 million (Series A), depending on the headline, and the amounts do not match. - Effectiveness/accuracy data is not included in the heading.

Hospital management/medical provision system

- Coverage of how healthcare systems are managing AI token costs (Modern Healthcare). - MedPal AI reports record daily prescription orders and liquidity balance of £6.4m (Yahoo Finance UK). - Delta CEO's article on fuel costs and AI will not be covered as it has nothing to do with medicine.

Drug discovery/clinical trials

- IQVIA: Releases an AI model that predicts clinical trial results. The stage of the test and prediction accuracy are not listed in the headline. - Suvoda: Reduce clinical trial setup time with AI (MyChesCo). The width of the shortening is unknown. - Carterra x AstraZeneca: Collaboration on “lab-in-the-loop” type AI drug discovery. Stage unknown. - Bay Area biotech company: Report on using AI to make cell therapy more accessible (CBS News, KION). The stage is unknown, and it is not possible to determine whether it is preclinical or clinical. - Article discussing retinal AI: as a biomarker for screening and stratification in trials involving pregnant women (Clinical Trial Vanguard).

Medical device approval/regulation

- MRI2CT: FDA clearance (clearance, not approval). - Utah's AI pilot: Reports that it could be subject to FDA regulation or conflict with FDA (STAT, Nacionale News). It is unclear from the headline whether it is a guideline or approval. - CARPL.ai x HOPPR: Partnership on AI-based model. Regulatory status is unknown. - Skin Analytics: Deploying autonomous AI. The regulatory classification is not in the heading.

Safety/Error/Bias

- Retraction of ovarian cancer AI study: due to dataset contamination (thyroid images mixed in). - Book on patient safety in India: It is estimated that 1 in 10 patients are harmed during medical treatment. - The "limitations" of AI electrocardiograms are discussed (specific details are unknown). - Ethical essay on triage.

Research results (major journals)

- Symptom self-assessment: AI-assisted symptom self-assessment has made decisions about medical treatment more appropriate (2 Minute Medicine). The subject size and research design are not listed in the headline, and the level of evidence is unknown. - Harvard Medical School: A new dataset that aims to "show its work" to AI with CT scans. This is not an evaluation of clinical performance at the dataset release stage.

Japanese movements

- Teradata: Announced results of AI utilization and disease prediction in the medical and insurance fields (Infoseek). There are no numbers or diseases in the headlines. - The business usage demonstration of "Tsunagi AI" and the vulnerability diagnosis of AeyeScan are not directly related to medical care. - The article about cancer diagnosis using AI in Vietnam is a movement outside the country. - Diamond Online's Stripe article has nothing to do with medical care.

Remarks from VIPs

No applicable information (tracked list is empty).