Global Digi-Health Monitor: Calendar Week 11, 2025
Reimagining Health Records Through Transmedia Storytelling
Social media has transformed how we share our lives, yet our medical records remain stagnant, capturing only isolated snapshots from brief clinical encounters. Kirsten Ostherr, founding director of Rice University’s Medical Humanities Research Institute, proposes merging these worlds through “medical transmedia” – a multifaceted approach to health documentation that weaves patient-generated content with clinical data. This would combine the engaging aspects of social platforms with medical validity, allowing patients (especially those with chronic conditions) to document stress levels, medication effects, and daily experiences in formats like photos, videos, and narratives. Unlike electronic health records that reduce patients to numbers, transmedia documentation would reveal patterns between lifestyle and symptoms, potentially improving diagnosis and treatment. For example, a patient with Parkinson’s could share meal photos and videos to document nutrition, tremors, and speech patterns, while a Black patient’s blood pressure reading might be better understood alongside their documented experiences with systemic racism. While privacy concerns must be addressed through patient ownership of data, this approach could transform healthcare by humanizing patients while simultaneously enriching medical science.
WHO Establishes AI for Health Governance Collaborating Centre at Delft University
The World Health Organization (WHO) has designated the Digital Ethics Centre at Delft University of Technology in the Netherlands as its Collaborating Centre on artificial intelligence for health governance. This partnership aims to ensure the ethical and responsible use of AI in healthcare while maximizing its potential to transform health systems globally. The centre will advance research on priority topics, provide expert input for WHO guidance, serve as an education hub, and facilitate knowledge-sharing through regional workshops. Led by Professor Jeroen van den Hoven, the centre brings two decades of experience in digital ethics and responsible innovation to help operationalize ethical values into AI design requirements. WHO officials emphasized that this collaboration strengthens their ability to help Member States navigate AI opportunities and challenges while upholding ethical standards and ensuring that benefits reach everyone equitably and safely.
AI Plus Telemedicine: A Powerful Combination for Healthcare Transformation
The American healthcare system faces critical challenges in managing chronic diseases, specialist shortages, and affordability that an innovative partnership between generative AI and telemedicine could solve. This combined approach creates synergy by leveraging AI’s ability to monitor conditions and analyze symptoms while using telemedicine to provide immediate clinical intervention when needed. For chronic disease management, AI can continuously track patient metrics through wearable devices, flagging concerning trends and scheduling telemedicine appointments when treatment adjustments are necessary. Meanwhile, specialist access improves as AI analyzes symptoms and connects patients with remote experts, eliminating geographical barriers. Studies show virtual specialty consultations resolve 40% of cases immediately, with another 30% handled after additional testing. Despite these promising capabilities, adoption remains limited because the current fee-for-service payment model incentivizes treatment over prevention. A shift to value-based care would reward providers for keeping patients healthy, naturally encouraging the integration of these technologies. With proper implementation, this combined approach could prevent up to half of all heart attacks, strokes and kidney failures while saving approximately $1.5 trillion annually – more than a quarter of total U.S. healthcare expenditures.
Medical AI Tools Need Better Testing Methods, Experts Argue
Researchers led by computer scientist Deborah Raji argue that current evaluation methods for healthcare AI are fundamentally flawed and misleading, potentially endangering patient outcomes. While artificial intelligence is being rapidly integrated across healthcare applications—from breast cancer screenings to clinical note-taking and virtual nursing—a systematic review revealed that only 5% of studies evaluating medical large language models (LLMs) used real patient data, with most relying on medical exam questions like the MCAT that poorly represent actual clinical tasks. Writing in the New England Journal of Medicine AI, Raji and colleagues contend these benchmark tests fail to capture the complexities of real-world medical decision-making, creating false confidence in AI systems that may perform differently when deployed in hospitals. The researchers advocate for more contextually appropriate evaluation frameworks developed through interviewing domain experts, collecting naturalistic datasets from pilot interactions, conducting “red team” adversarial testing, and requiring greater transparency from both healthcare institutions and AI vendors about their actual implementation and testing practices.
Study Reveals AI Can Detect Gender Bias in Emergency Triage Decisions
Researchers at Bordeaux Population Health Research Center have demonstrated how large language models (LLMs) can uncover cognitive biases in emergency medicine decision-making. Using an innovative AI system trained on 480,000 emergency department records from Bordeaux University Hospital, the team led by Emmanuel Lagarde found significant gender bias in triage assessments. When researchers altered patient gender references in clinical texts, the AI revealed that women’s conditions were underestimated compared to men’s, with approximately 5% of female cases classified as “less critical” than identical male cases. The bias was more pronounced among less experienced nursing staff. This study, published in Proceedings of Machine Learning Research, demonstrates that generative AI can not only reproduce human biases from clinical data but potentially help identify and mitigate them, opening possibilities for fairer emergency care and future research on age and ethnicity-related biases in medical decision-making.
