If you have 1 minute - here’s the summary
Chan Zuckerberg Initiative (US): Zuckerberg and Chan consolidate Biohub with EvolutionaryScale, forming one of the largest biology-AI research engines.
McKinsey (Global): 88% of firms now use AI, but only 6% see >5% EBIT gain - real impact comes from workflow redesign.
OpenAI (US): Exploring a consumer health assistant and unified health data hub; ex-Doximity founder joins to lead strategy.
Nvidia (US): Partners with Eli Lilly, J&J, and Verily to power AI-driven drug discovery, robotics, and genomics.
Microsoft (US): Forms Superintelligence Team for healthcare-grade diagnostic and planning AI.
Ambience (US): Launches Conditions Advisor to surface overlooked inpatient diagnoses.
Viz.ai (US): Debuts Viz Assist agents integrating imaging + EHR data for real-time patient summaries.
Mercy + Microsoft (US): Pilot nurse-focused ambient documentation cuts 24 min per shift.
Hippocratic AI (US): Raises $126 M Series C to scale safe agentic systems for healthcare.
EU Publishes report on AI deployment gaps - interoperability and governance remain the biggest barriers.
If you have more time, read on for the foresight of the week and the full breakdown of what shaped healthcare AI 👇
Signal of the week
AI in the Patient’s Hands, Not Just the Hospital’s
(Image credit: Getty / ullstein bild)
A US patient used Anthropic’s Claude to audit a $195,000 hospital bill - uncovering duplicate charges and coding errors that reduced it to $33,000 without any legal help. The model parsed complex CPT and billing codes, surfacing violations and inconsistencies that forced the provider to amend its invoice. [link]
Foresight: AI’s centre of gravity is shifting from systems used on patients to tools used by patients. As large models get better at understanding medical language, we’re seeing the rise of “patient-side AI” - LLMs that help people audit their hospital bills, alongside point solutions built specifically for them such as AI doctors or AI advocates that explain their own records and lab results.
The best era for patients is quietly beginning. More point-of-care tools are being designed directly for them, and the balance is starting to shift from healthcare delivered to patients to healthcare shared with them.
Solutions and Launches
OpenAI GPT-5 (US): Enhanced crisis-response safeguards after input from 170 clinicians, improving adherence to mental-health safety protocols to 91%. [link]
Edison Scientific (US): Launched Kosmos, an autonomous “AI scientist” reviewing 1,500 papers and 42K lines of code per session with 79% validated accuracy. [link]
Microsoft (US): Formed a Superintelligence team under Mustafa Suleyman to build expert-level AI for medicine and diagnostics. [link]
Microsoft Dragon Copilot (US): Now powering documentation across 650+ healthcare organisations and 17 million patient encounters last quarter - nearly five-fold growth year-on-year. [link]
Chan Zuckerberg Initiative (US): Integrated its Biohub research network with EvolutionaryScale, an AI lab specialising in protein and cell modeling, to create foundation models that accelerate biological discovery and therapeutic research.[link]
Ambience Healthcare (US): Released Conditions Advisor to synthesize patient data and highlight missed diagnoses. [link]
Atropos Health (US): Debuted Evidence Agent inside EHRs to auto-generate real-world evidence; Stanford Health piloting. [link]
Amera launches: YC-backed startup automating claims processing in minutes, cutting insurer admin costs by ~30%. [link]
Governance, Policy, and Ethics
NHS (UK): Study finds 10,000+ digital health tools active, 70% lacking documented safety assurance - a significant governance risk. [link]
European Union: A new report maps healthcare AI adoption and underscores gaps in data sharing, governance and clinical assurance. [link]
FDA (US): Advisory panel debates status and use of therapy chatbots as medical devices. [link]
Research and AI Advancements
AI VS Clinicians in empathy: Meta-analysis of 13 studies found LLMs often rated more empathetic than clinicians - prompting debate over “synthetic bedside manner.” [link]
Transformer model using 9.7 million EHRs predicts biological age across lifespan, linking early life markers to adult disease risk. [link]
LLM Phenotyping: A model extracted ~400 clinical signs from 100,000+ admission notes, enabling syndrome-based infection risk stratification. [link]
McKinsey State of AI 2025: 88% of companies use AI but only around 6% achieve >5% EBIT improvement; agentic systems adoption at 23% in scaling phase. [link]
Partnerships, Collaborations, and Implementations
UCSF + OpenAI (US): Deploying ChatGPT Enterprise across clinicians, researchers, and admin staff for HIPAA-secure AI use. (link)
Mercy Health + Microsoft (US): Co-develop ambient AI for nursing workflows; pilot shows 21% reduction in documentation time. [link]
AdventHealth + HelloCare (US): Deploying AI-enabled virtual smart rooms with monitoring, digital whiteboards and remote rounds. [link]
DRH Health + Commure (US): Integrating ambient documentation AI within Meditech Expanse across hospitals and clinics. [link]
Bets on the Next Health System
Investments + M&A:
Bevel (US): Raised US $10M Series A from General Catalyst to build a consumer AI health companion. [link]
Tala Health (US): Raised US $100M seed to build an agentic platform for healthcare workflows. [link]
Arya Health (US): Raised US $18.2M Series A to develop post-acute-care AI agents. [link]
Hippocratic AI (US): Raised US $126M Series C to scale agentic AI systems in healthcare settings. [link]
Synchron (US): Brain-computer interface startup secured US $200M Series D led by Double Point Ventures. [link]
That’s a wrap for Edition #5 of Health AI Foresight.
My goal with this newsletter is simple - to connect the present, the emerging, and the future of healthcare AI.
See you next week.
- Dr. Aboufandi
📩 Something stand out? Or have a suggestion? Hit reply - I read every message.