🔍 Intelligence Synthesis JP/EN
Ethics · Regulation · Technology — Pharma Practice Notes
Article digest · Nov 20 AM

AI in Medicine — what each story says

Every story in this issue with a short summary, so you can decide what to read before leaving the site. Summaries are written by AI (Claude) only from each publisher's own description and checked by Jev for anything not in that description. They are not full translations. Follow "Read the original" for the full story.

33 stories16 with summaries8 topics

🩻 Diagnosis and imaging4

Bioengineer.org11/19 13:40 UTC

Systematic Review Evaluates Deep Learning Generative Models in Medical Imaging

A systematic review examines approximately 160 studies from 2021 to 2025. It evaluates how deep learning architectures synthesize clinical medical images and analyzes major performance trade-offs.

Read the original →
EMJ11/19 11:23 UTC

AI-Assisted Ultrasonography Advances Non-invasive Kidney Assessment

AI-assisted ultrasonography showed promising accuracy in detecting moderate-to-severe kidney fibrosis without invasive biopsy.

Read the original →
StudyFinds11/19 12:41 UTC

AI Beat Doctors’ Due Dates, Then Researchers Checked The Math

Researchers checked the due date accuracy of an AI ultrasound tool against fairer comparisons. The report describes what held up and what did not.

Read the original →

🩺 Clinical decision support4

🏥 Hospital operations and care delivery4

Quiver Quantitative11/19 13:00 UTC

Gix Internet Ltd. Launches Carelee, an AI Workforce Orchestration Platform for Hospital-at-Home Programs

Gix's Carelee platform improves caregiver coordination for hospital-at-home programs through AI-assisted tools.

Read the original →

💊 Drug discovery and clinical trials4

The University of North Carolina at Chapel Hill11/18 21:14 UTC

Leading the next era of AI-enabled drug discovery

The UNC Eshelman School of Pharmacy is advancing the possibilities of AI in drug discovery. Its work ranges from discovering new therapeutic compounds to developing open-source tools.

Read the original →

⚖️ Device approvals and regulation3

Bioengineer.org11/19 07:27 UTC

Comparative Analysis Reveals AI-Enabled Medical Device Landscapes in Australia and US

This cross-sectional study maps the AI-enabled medical device landscape in Australia and the United States. It highlights differences in market concentration, specialty distribution and regulation.

Read the original →
Medical Daily11/19 05:54 UTC

Utah lets AI renew—and in one pilot write—prescriptions, but federal FDA oversight remains unanswered

A new HHS roadmap promotes state testing of clinical AI. It leaves unanswered whether the FDA oversees Utah pilots that renew medications and issue some prescriptions.

Read the original →

⚠️ Safety, errors and bias4

HIT Consultant11/19 09:43 UTC

What Is Agentic AI in Healthcare and How Do You Keep It Safe?

XiFin CTO Sanjay Kumar explains what agentic AI is and why monolithic agents fail in healthcare revenue cycle workflows. He also describes how multi-agent orchestration and least-privilege governance keep AI safe.

Read the original →
Korea Biomedical Review11/19 07:12 UTC

[Column] Korea's medical AI future should start in neighborhood clinics

The columnist, a clinician, describes finishing morning patient visits and spending lunch breaks catching up on charts. They note that reviewing a hypertension patient's blood pressure over the past year can mean opening one record after another.

Read the original →
Deccan Herald11/18 20:29 UTC

Former WHO chief Swaminathan calls for stronger safeguards to regulate AI in healthcare

Speaking at Bengaluru Tech Summit 2026, former WHO chief scientist Dr Soumya Swaminathan urged the creation of national ethical frameworks to monitor AI products. She said the aim is to ensure patient safety.

Read the original →
Digital Today11/18 00:29 UTC

KT to pilot medical AI robots with Seoul National University Hospital to support care for child patients

KT will work with Seoul National University Hospital to pilot AI robots in a real medical environment, testing physical AI and developing patient-centered service models. The pilot focuses on non-clinical services for child patients, including conversation, play, explanations of tests and treatment, and guidance and escort inside the hospital. Participation will be voluntary with prior guardian consent, and the robots will not access medical records or store personal data.

Read the original →

🔬 Research results (major journals)5

NZ Doctor11/19 09:55 UTC

US clinicians turn to AI for treatment answers

US clinicians are increasingly using artificial intelligence to help answer their treatment and management queries, according to a study published by JAMA.

Read the original →
Onco'Zine11/19 03:08 UTC

Your AI App Says You Have Cancer. Now What?

Your phone may be the first thing to raise the possibility of cancer.

Read the original →

🇯🇵 Medicine and AI in Japan5

ZDNET Japan11/19 06:34 UTCJapanese

Ichinomiya Nishi Hospital Adopts SuguDesk; AI Chat Resolves About 80% of Inquiries

Ichinomiya Nishi Hospital, an advanced acute-care general hospital with 801 beds in Ichinomiya, Aichi Prefecture, run by the medical corporation Kyoreikai, has introduced SuguDesk. SuguDesk is an AI reception chat for medical institutions provided by exmore.

Read the original →
Medical Tech News11/19 09:34 UTCJapanese

Imageone Opens Physical AI Robot Demonstration Site to Test Support for Medical Facility Tasks

Imageone announced on November 18 that it opened "IMO Robotics LAB," a display and demonstration site for physical AI robots, inside its Shinagawa, Tokyo headquarters. It was set up with Big Hands, a systems integration company based in Chiyoda, Tokyo. The site will mainly examine which medical facility tasks robots can support and what challenges arise at introduction.

Read the original →
valuepress11/19 01:00 UTCJapanese

AI Diagnostic Engine Emits 0.003631 mgCO₂e per Decision; Big-O Self-Discloses Software Carbon Intensity

In a press release dated November 19, 2026, Big-O Information Intelligence Corp. said it calculated the software carbon intensity (SCI) of its "LLMO diagnosis" decision engine using the ISO/IEC 21031:2024 method and disclosed the result itself. The figure is 0.003631 mgCO₂e per diagnostic decision, measured at the vCPU boundary.

Read the original →
The analysis and visual summary are in the Nov 20 AM issue. Japanese version here.