Partnered with
Welcome back HAIFer - hope you’re enjoying some well-earned time off.
Over the holidays, I’m pulling together every prediction I made for 2025 and turning it into a short reflection. Early read: a lot of it landed faster than I expected. I’ll share that recap soon - and then publish my 2026 predictions.
For now, here’s this week’s healthcare AI roundup.
And of course, thank you to Doximity for supporting this issue.
Sponsored by: Doximity
DoxGPT is a HIPAA-compliant AI assistant designed for U.S. healthcare professionals. It streamlines clinical and administrative tasks by answering clinical questions with cited, evidence-based references - including integrated drug information -generating chart notes and documentation, and creating patient education materials.
DoxGPT integrates seamlessly with Doximity tools and ensures robust data security. Free for verified clinicians, DoxGPT reduces documentation burden and improves access to up-to-date medical knowledge, so you can focus more on patient care.
Your shortcut for this week:
Read on for the foresight of the week and the full breakdown of what shaped healthcare AI 👇
Signal of the week
Clinical AI Governance Enters Its “How” Phase
This week, both UK (MHRA) and US (HHS/ONC) regulators opened public consultations on what safe clinical AI deployment should look like. It’s a direct response to the “regulators are behind” critique on socials - a signal they’re starting to listen to the deployment layer, not just publish principles.
Solutions and Launches
Mass General Brigham (US): Spun out AIwithCare to launch RECTIFIER, an AI tool for clinical trial screening. It analyses unstructured EHR notes and reportedly doubles patient enrolment rates compared to manual methods.[Link]
Pyx Health (US): Introduced an AI navigator to assist Medicaid and SNAP beneficiaries with the "One Big Beautiful Bill Act." It streamlines eligibility documentation and connects members to social support resources. [Link]
McCrae Tech (New Zealand): Launched Orchestral, a health-native AI orchestrator data platform that connects diverse data sources with AI agents and workflows for scalable, governed AI deployment. [Link]
MediRecords (Australia): Announced Evolve Direct, an AI agent for processing inbound clinical documents. Early betas show a reduction in clicks from 18 to 3, saving clinics up to 10 hours per week.[Link]
Governance, Policy, and Ethics
HHS & ONC (US): Issued an RFI on harnessing AI to "deflate health care costs" and advance the "Make America Healthy Again" agenda. It seeks input on using regulatory and reimbursement levers to accelerate clinical AI adoption. [Link]
MHRA (UK): Launched a national call for evidence to inform the new National Commission on AI regulation. It seeks views on liability, post-market surveillance, and the adequacy of current laws by February 2026. [Link]
Finnish Government (Finland):: Approved legislative amendments to enforce the EU AI Act, effective January 1, 2026. Authorities will supervise AI systems to ensure they do not endanger health, safety, or fundamental rights. [Link]
European Commission (EU): Proposed the Biotech Act to accelerate the transition of health innovations from lab to market. [Link]
Research and AI Advancements
Just launched: MedASR, a Health AI Developer Foundations, is a 105M-parameter Conformer model designed for high-accuracy medical dictation and serves as an acoustic front end for generative pipelines. [Link]
Concurrent prediction of in-hospital mortality and length of stay using single-task, multiclass and multitask machine learning. [Link]
Mount Sinai’s NutriSighT model predicts the risk of underfeeding in ventilated ICU patients hours in advance. [Link]
The 'Empathy AI in Healthcare' study found that AI models scored 4.1/5 on compassion metrics, significantly outperforming human clinicians who scored 2.6/5. [Link]
Harvard Medical School: Researchers identified that cancer diagnostic models were learning demographic shortcuts and introduced the FAIR-Path debiasing framework. [Link]
University of Oxford (UK): Secured ethical approval to pilot TrustedMDT, a multi-agent AI system integrated into Microsoft Teams for cancer care. [Link]
Partnerships & Adoption
Nabla (France/US) & Advanced Machine Intelligence: Announced an exclusive strategic partnership to deploy Yann LeCun’s "world model" technology in healthcare. [Link]
Medicus Pharma & Reliant AI: Partnered to build a generative AI platform that optimizes clinical trial execution, focusing on dynamic site selection and patient stratification. [Link]
Horizon Health & Qure.ai (Canada): Partnered to evaluate and implement an AI suite for lung cancer detection on chest X-rays and CTs, preparing for a province-wide screening program. [Link]
Sanome & NHS Trusts (UK): Partnered with East Kent Hospitals and the Royal Hospital for Neuro-disability to deploy the MEMORI tool, designed to predict hospital-acquired infections up to three days early. [Link]
Bets on the Next Health System
Investments:
Valerie Health (US): Raised $30M Series A to bring its AI front office to independent provider groups. [Link]
Dentira (US): Secured strategic investment from Vista Equity Partners to scale its AI spend management platform for multi-site healthcare procurement. [Link]
Mindoo (Belgium): Raised €5M to build an "AI workforce" for hospitals. Their configurable agents integrate with EHRs to automate intake and follow-up. [Link]
M&A:
Affluent Medical (France): Affluent Medical: Agreed to acquire Caranx Medical and Artedrone to form "Carvolix," alongside a simultaneous €30M fundraising led by Truffle Capital and Edwards Lifesciences. [Link]
That’s a wrap for Edition #11 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.