Our latest journal metrics are now available! https://t.co/r2a94kxK2i

🇪🇺 Navigating uncharted waters: select practical considerations in radiology #AI compliance with the #EU AI Act | @npjDigitalMed https://t.co/b2am3OD9Uh #MedTwitter #AImedicine #AI #HealthAI #AIinHealthcare #MedTech

Our new study in @npjDigitalMed explores how combining AI with human review—“human-in-the-loop”—can improve translation of patient discharge instructions across 6 languages. 🧵👇

Our new paper with Sonali Sharma and @RoxanaDaneshjou is out in @npjDigitalMed! We examine how medical safety and disclaimer messages in public LLMs have changed over time when answering patient questions.

Call for Papers — npj Digital Medicine Exploring how digital tools can bridge—or widen—health gaps. Submit research or perspectives on equitable AI, telehealth, and inclusive innovation to ensure tech benefits everyone, not just those with access. Deadline: Dec 3, 2025

New today in npj Digital Medicine: Researchers tested GPT-4o, Llama 3 & others on personalized, biomarker-based longevity advice. Across 56K cases, models were safe—but lacked depth, validation & showed age bias. Is AI ready to be your longevity coach? https://t.co/KmoRrmUhKm

Cases are emerging of harm or suicide resulting from people’s dependent relationships with AI chatbots. Are these warning signs of a larger hidden problem? And, if so, what should regulators do? https://t.co/DSQlbFfOj0

AI alone struggles to accurately translate hospital discharge instructions—especially for underrepresented languages like Somali & Armenian. But adding a human-in-the-loop made translations faster, safer & often better than professional versions. https://t.co/x6DSWApT11

Absolutely. Dr. Kostick-Quenet’s piece in npj Digital Medicine is such a timely warning — personalization in AI may improve “user satisfaction,” but in medicine, it risks filtering out the very friction clinicians need to make sound, evidence-based decisions.

A systematic review finds mobile-based active remote monitoring (apps where patients regularly log symptoms) is cost-effective in 6 of 7 studies for long-term conditions like arthritis, MS, and cancer—yet uncertainty remains for scale-up. https://t.co/IX7RVB03Ki

NEW in npj Digital Medicine: A deep learning model can spot fluid buildup & air leaks in ventilator circuits in real time — detecting subtle waveform changes invisible to the eye. 99% accuracy → potential to prevent ventilator-associated events. https://t.co/oquqrKPa3T

Meet HONeYBEE: an open-source AI that unifies clinical, imaging, and molecular data to predict cancer outcomes. Tested on 11,400+ patients across 33 cancers, it reached 98.5% accuracy in classification and improved survival prediction. https://t.co/a3jqYKnrTN

Researchers found GPT-4 and Llama 3 often comply with false or unsafe prompts — even when they “know” the correct answer. This “AI sycophancy” reveals a deeper risk: when models prioritize politeness and compliance over clinical accuracy, patient safety is on the line.

An AI system called UMORSS integrates ultrasound + clinical data to assess ovarian cancer risk—with uncertainty awareness built in. Across 7,352 patients, it matched or beat expert radiologists — while cutting unnecessary exams and improving workflow. https://t.co/YdFydrNVPr

Northwestern scientists have created a new #LifestyleMedicine program that uses three #wearable sensors to capture real-world eating behavior in unprecedented detail, per @npjDigitalMed. 🔗 https://t.co/L45QZjBNOe

That’s a fascinating step forward for Digital Health and Lifestyle Medicine! Using wearables to track how and when people eat objectively could finally close the gap between self-report and reality: paving the way for more personalized nutrition, obesity prevention, and

Generative AI models are giving fewer medical disclaimers over time. 📉 In 2022, ~26% of AI health answers had a disclaimer. By 2025? <1%. As models get smarter, they’re getting less safe. Patients may take outputs as medical advice. https://t.co/2OYQvKdezT

Sepsis is the #1 killer of burn patients, but spotting it early is tough. A new @npjDigitalMed study trained an ML model on 6 simple admission variables (age, burn size/depth, inhalation injury, etc.). Results show reliable early risk stratification. https://t.co/qlFoX4guLQ

Big step for AI in eye care: a deep learning model (MORG) can auto-generate retinal OCT reports nearly as accurate as ophthalmologists—while cutting reporting time by ~59%. Could reshape how we manage Diabetic Retinopathy & more https://t.co/Btim3RIU0r

Important reminder: most AI in healthcare fails not because of weak algorithms, but because of poor fit with clinical workflows + missing data infrastructure Read here:

Surgical site infections are among the most common hospital infections—but tracking them is labor-intensive. A new @npjDigitalMed study shows ML + rule-based models can detect SSIs with up to 90% sensitivity while cutting manual workload by >90%. https://t.co/T4yXgn4OsT