How to unlock digital health with AI evaluation frameworks

Artificial intelligence and digital health offer immense promise for the healthcare industry. However, systemic barriers continue to hinder their full potential. Discover how new AI evaluation frameworks can help realize the promise of these technologies in our latest Healthcare Transformers article. https://spkl.io/6044AVWvi #Roche #HealthAI #DigitalHealth #HealthcareTransformers

Artificial intelligence and digital health offer immense promise for the healthcare industry. However, systemic barriers continue to hinder their full potential. Discover how new AI evaluation frameworks can help realize the promise of these technologies in our latest Healthcare Transformers article. https://spkl.io/6049ATN7R #Roche #HealthAI #DigitalHealth #HealthcareTransformers

Artificial intelligence and digital health offer immense promise for the healthcare industry. However, systemic barriers continue to hinder their full potential. Discover how new AI evaluation frameworks can help realize the promise of these technologies in our latest Healthcare Transformers article. https://spkl.io/6042Ap6dq #Roche #HealthAI #DigitalHealth #HealthcareTransformers

Artificial intelligence and digital health offer immense promise for the healthcare industry. However, systemic barriers continue to hinder their full potential. Discover how new AI evaluation frameworks can help realize the promise of these technologies in our latest Healthcare Transformers article. https://spkl.io/6041ATLZS #Roche #HealthAI #DigitalHealth #HealthcareTransformers

Here is the evolution of using AI in healthcare from a zoomed-out perspective. 1) First, there were data digitally through electronic medical records. 2) Then, the rise of AI has led to the analysis of vast amounts of data. 3) The third step could be using AI not to analyze but to predict disease progression and, even more importantly, the onset of diseases in the future. This is what this articles describes focusing on Delphi-2M, an AI system trained on millions of health records that can estimate the likelihood and timing of more than 1,000 diseases throughout a person’s life. Source: https://lnkd.in/eKakxWyr

Artificial intelligence is transforming health and medicine. But without shared values, even well-intentioned technology can deepen divides or compromise care. The National Academy of Medicine’s special publication called Artificial Intelligence Code of Conduct in Health and Medicine offers a blueprint for trustworthy AI — grounded in six key commitments that place human dignity, access, and safety at the center of every decision, from code to clinic. In this clip, Dr. Suchi Saria of Bayesian Health and The Johns Hopkins University shares how her team is applying these principles in practice — ensuring that AI tools improve outcomes and truly serve the people they’re designed for. Explore the full Code of Conduct and how to put it into action: https://buff.ly/3iGXAzI #AIinHealth #ResponsibleAI #HealthInnovation #NAMedicine

Will LLMs ever be truly safe for medicine or are we measuring the wrong things? A new arXiv release, “MedOmni-45°: A Safety Performance Benchmark for Reasoning-Oriented LLMs in Medicine,” reframes this debate. Unlike traditional benchmarks, MedOmni-45° looks beyond accuracy to probe whether AI models reason safely, resist misleading prompts, and perform reliably under clinical stress. It asks the tough question: Not just, “Did the AI get it right?” but “Did it reach the answer for the right reasons?” Just as no single test captures real intelligence, we shouldn’t assume leaderboard rankings equal clinical readiness. This work moves the bar putting safety and sound reasoning at the center, where they belong. What stands out? Early results show even top rated models can fail when facing ambiguous or adversarial scenarios, underscoring that clinical trust demands more than brute force accuracy. This is a valuable shift for anyone building or evaluating healthcare AI. #MedOmni45 #MedicalAI #AIinHealthcare #MedicalSafety #ResponsibleAI 📄 Credit: MedOmni-45°: A Safety–Performance Benchmark for Reasoning-Oriented LLMs in Medicine, arXiv (August 21, 2025).

🤖💙 ARTIFICIAL INTELLIGENCE AND MEDICINE: BETWEEN ENTHUSIASM AND CAUTION 💙🤖 Artificial Intelligence leaves me with a dual feeling: enthusiasm for what it promises and caution for what it may change. The question is not if it will change medicine, but how (and who) will truly lead this transformation. ❤️ It is not technology that heals. It is people who heal! AI amplifies what we already are: more prepared if prepared, more distracted if distracted. More empathetic if empathetic, colder if cold. More ethical if ethical, more complacent if complacent. 💭 The future will not be less human, but more demanding of our humanity. It will require more competence, more clinical judgment, and a greater ability to distinguish between data and wisdom. 🩺 Artificial Intelligence will truly be useful only when it makes every physician free to do what no algorithm will ever do: listen, decide, and care. For further reading, see: Angus DC et al. AI, Health, and Health Care Today and Tomorrow: The JAMA Summit Report on Artificial Intelligence. JAMA. 2025; doi:10.1001/jama.2025.18490 #AIinMedicine #DigitalHealth #CardiacSurgery #FutureOfMedicine #HumanCenteredCare #ConnectedCare #MedicalEthics #HealthcareLeadership

IARC scientists are participating in PERIFORMANCE, a European project to ensure #CancerResearch & developments, in line with tech advances such as #AI & the European Health Data Space (EHDS), are applied ethically, equitably & transparently. BBMRI-ERIC https://lnkd.in/d6UUTtid