New week, new signals.
In 15 seconds:
1. AI tools go team-wide beyond physicians
2. Consumer AI health companions are accelerating
3. Clinical trials meet clinical workflows through AI tools
3. And the latest startup funding and AI healthcare research published
Let’s dive in 👇
Signal of the week
Microsoft Expands Dragon Copilot To Nurses
Microsoft is extending Dragon Copilot from physicians to nursing workflows—automating flowsheet capture and note drafting to cut admin load. This is made possible through their partnerships with Elsevier, OpenEvidence and Wolters Kluwer UpToDate. General availability for nurses in the U.S. begins December 2025, with broader uptake likely through 2026.
Foresight: As clinician shortages persist, ambient + AI-CDS will equip nurses, pharmacists, and PAs to take on larger roles—if governance is clear. Microsoft’s demo even shows a nurse querying IV albumin dosing for SBP, a task traditionally handled by doctors—further validating this shift. I predicted this will take place last year in this LinkedIn post.
Solutions and Launches
Microsoft (US): Expanded Dragon Copilot with nurse-focused features like ambient flowsheet capture and AI drafting for notes. This reduces cognitive load for care teams, enabling focus on patients amid shortages. [Link]
Verily (US): Launched Verily Me, an app offering personalised care team recommendations, unified health data (wearables, lifestyle), and an AI companion to interpret data and provide tailored health insights. Consented data also supports the Lifelong registry for pharma research. [Link]
SonderMind (US): Launched clinically backed AI tools for mental health care delivery. These boost operational efficiency and accelerate patient improvement in virtual settings. [Link]
Innovaccer (US): Unveiled Galaxy, an AI-powered payer platform for risk adjustment and quality improvement. It transforms payer performance by streamlining workflows and enhancing accuracy. [Link]
Hello Heart (US): Introduced Nia, an AI heart health assistant for managing cholesterol and blood pressure. This supports at-risk individuals in preventing heart disease through personalised guidance. [Link]
TytoCare (US): Launched Smart Clinic Companion powered by a multi-modal health dataset. It tackles primary care shortages with AI-enhanced home diagnostics. [Link]
Governance, Policy, and Ethics
MHRA (UK): Phase 2 of the AI Airlock launched, naming seven tools. They also published reports on Phase 1 - capturing lessons for future regulation. It tightens the UK pathway for safely testing higher-risk AI before scale-up. [Link]
Research and AI Advancements
Multi-turn evaluation of medical LLMs: New benchmark shows how performance shifts in multi-turn consult-like dialogues vs. single-shot QA, with data/code on HuggingFace. Implication: procurement should ask for multi-turn evidence, not point-in-time scores. [Link]
Edge medical imaging models: “BitMedViT” reports ternary-quantised ViTs achieving 86% accuracy on MedMNIST across 12 datasets with 43× smaller models and ~41× energy efficiency—relevant for near-patient and wearable use-cases. [Link]
Enabling Doctor-Centric Medical AI with LLMs through Workflow-Aligned Tasks and Benchmarks: Introduces DoctorFLAN dataset with 92,000 Q&A across 22 tasks; improves open-source LLM performance in medical contexts. Aligns AI with physician workflows to enhance clinical guidance safely. [Link]
AI-Powered Clinical Workflow
AI-Based Behavioral Health Safety Filter and Dataset for Identifying Mental Health Crises: VBHSF achieves 0.990 sensitivity for crisis detection in text conversations. Prioritises minimising missed emergencies in behavioural health apps. [Link]
Speculative Model Risk in Healthcare AI: Using Storytelling to Surface Unintended Harms: Framework uses stories to identify broader harms like bias and privacy issues; participants recognised harms across 13 types evenly. Reduces unintended risks in tools like wellness trackers before deployment. [Link]
AI-Powered Early Diagnosis of Mental Health Disorders from Real-World Clinical Conversations: Models achieve over 80% accuracy for depression, anxiety, and PTSD screening. Enables low-barrier early detection in clinical workflows. [Link]
Implementation of AI in Precision Medicine: Scoping review identifies barriers in data quality and governance; proposes ecosystem framework for sustainable translation. Supports trustworthy AI integration in clinical settings. [Link]
The Ecosystem Shaping AI Adoption in Precision Medicine
Multimodal AI Agents in Healthcare (scoping review, 37 studies): Maps four use-cases—clinical decision support, documentation/reporting, monitoring/health management, and medical education; flags evaluation gaps (robustness, fairness, safety, usability). Useful checklist for buyers to look beyond accuracy when assessing agentic systems. [Link]
Partnerships, Collaborations, and Implementations
OpenEvidence & Veeva Systems (US): Long-term partnership to co-create Open Vista, embedding trial eligibility during consultations with patients; first offerings expected in 2026. [Link]
MedicubeX & Istekki (Finland): €4M public-tender to deploy eHealth Stations™ across Istekki’s public-sector network. On the ground: self-measurement kiosks for vitals and diabetes/CVD risk plus AI-analysed 1-lead ECG (multi-arrhythmia detection), feeding data into client systems. [Link]
MedicubeX eHealth Station
Microsoft & multiple partners (US): Collaborated with Artisight, Atropos Health, Canary Speech, and others to build third-party AI solutions within Dragon Copilot. Enables automated prior authorisation and diagnostics, streamlining clinical and revenue workflows. [Link]
UPMC & Abridge (US): UPMC will scale Abridge’s ambient AI across the system (aiming >12,000 clinicians by 2026), signalling enterprise-wide scribe normalisation in a flagship academic network. [Link]
Northwell Health & Abridge (US): Northwell (28 hospitals) to deploy Abridge system-wide to ease documentation burden and standardise note quality across specialties. [Link]
Bets on the Next Health System
Investments + M&A:
OpenEvidence (US): $200M Series C at a $6B valuation to scale its clinical AI-CDS platform and for hiring. [Link]
Counsel Health (US): $25M Series A to scale its physician-supervised AI care platform and expand from urgent care into direct-to-consumer services. Funds go to clinician hiring and AI capability build-out. [Link]
Oura (Finland): $900M growth round led by Fidelity to expand hardware, biomarkers and software ecosystem. [Link]
ABK Biomedical (Canada): $35M Series D to progress imageable embolic devices toward commercialisation-interventional oncology toolchain plays continue to attract capital. [Link]
OneImaging (US): $38M funding to build lower-cost imaging footprint; claims affordability via ops/AI efficiency—track payer/provider pilots for proof on throughput and cost per study. [Link]
Marble (US): $15.5M Series A to scale youth mental-health infrastructure in schools, including AI-enabled eligibility and activation workflows. [Link]
MD Integrations (USA): $77M Series A. Accelerates telehealth network innovations. [Link]
Brook.ai (USA): $28M Series B. Scales remote patient care tech. [Link]
Medmo (USA): $15M Series A. Improves radiology appointment orchestration. [Link]
R1 acquires Phare Health (USA): Acquires AI platform for inpatient coding and documentation. Enhances revenue cycle automation. [Link]
Cairns Health acquires Together by Renee (USA): Acquires AI lifestyle assistant for seniors. Advances chronic care management. [Link]
Hazel Health acquires Little Otter (USA): Acquires youth mental health provider. Scales school-based services. [Link]
That’s a wrap for Edition #3 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.