Health AI in the 2nd Week of March

AWS (US): Launched Amazon Connect Health to tackle administrative burden. This agentic AI integrates directly with electronic health records to automate high-volume tasks like appointment scheduling, ambient documentation, and medical coding. [Link]

Sword Health (US): Launched Dawn, a direct-to-consumer AI mental health application offering continuous conversational support. It is powered by a proprietary foundational model trained on hundreds of thousands of hours of clinician data. [Link]

CVS Health (US): Unveiled Health100 in partnership with Google Cloud. Powered by Gemini AI, the platform connects a patient's different health services into one place to eliminate the stress and "homework" of managing their care.[Link]

Google Cloud (US): Showcased enterprise-wide Gemini-powered AI agents at HIMSS26. Moving past traditional tools, these autonomous agents execute multi-step administrative workflows alongside partners like Humana and Waystar. [Link]

Optum (US): Expanded its Optum Real claims platform using Microsoft Azure. The AI infrastructure predicts coverage and assists prior authorization, reportedly cutting avoidable denials by up to 80% in pilot programs. [Link]

Salesforce (US): Deployed six new Agentforce Health agents to automate digital work across providers, payers, and public health. Early adopter MIMIT Health saved $1.5 million and saw a 459% ROI by cutting administrative tasks. [Link]

Florida House of Representatives (US): Unanimously passed HB 527. The bill mandates that decisions to deny or reduce an insurance claim must be made by a qualified human professional, legally preventing AI from being the sole basis for coverage decisions. [Link]

Proofpoint (US): Warned that ambient AI systems, like documentation assistants, silently expand cybersecurity attack surfaces. Strong data governance is now the front line for protecting protected health information. [Link]

Colorado State Legislature (US): Advanced two healthcare AI bills, including one restricting chatbot use in psychotherapy and another preventing AI-only insurance denials without human review. [Link]

Doctronic (US): Researchers showed that the medical chatbot Doctronic could be jailbroken to generate unsafe advice and misleading AI-generated clinical notes. [Link]

UK House of Lords (UK): Raised urgent questions regarding data protection risks when NHS and public sector employees use generative AI tools. [Link]

Health Foundation (UK): Published survey data showing strong public concern over AI data privacy and bias. Earning patient trust requires strict governance and human oversight alongside clear communication. [Link]

Policy Connect / APHG: Said healthcare AI regulation should be proportionate enough to protect patients and professionals without blocking responsible innovation. It also stressed that trust, accountability, workforce readiness, and interoperable infrastructure remain unresolved foundations for NHS digital transformation. [Link]

Sepsis Prediction Under Distribution Shift: A multi-centre ICU study found that sepsis prediction models often lose generalisability when deployed in new hospital environments because of distribution shift. It also showed that fine-tuning is not always the best adaptation strategy. [Link]

Enhancing quality of antimicrobial prescribing through ‘Ask Eolas’: . Across 45 prescribing cases, it improved prescribing accuracy versus comparator guidance tools, although the findings were limited to simulated rather than real-world use. [Link]

Safety of Ophthalmology Scribes: A study evaluated the performance and safety of LLM-based ophthalmology scribing tools using simulated clinical encounters. [Link]

Validating AI-Enhanced ECGs: A UK Biobank study of 38,804 participants externally validated six AI-ECG models against cardiac MRI reference standards, showing strong diagnostic performance for detecting left and right ventricular dysfunction. [Link]

From Advice to Action — Real-World Behaviour of Patients Using an Integrated Diagnostic Decision Support System for Navigating the Health Care System: This prospective study evaluated how integrating an AI symptom checker into a Portuguese health network influenced actual patient decision-making. [Link]

Personalized Medical Devices: A systematic review examined how artificial intelligence is being integrated into medical devices to enable personalized healthcare, including AI-driven diagnostics, wearable monitoring systems, and smart prosthetics. The paper also discusses regulatory, safety, and ethical challenges in deploying AI-enabled medical technologies. [Link]

Factors for Patient Trust and Acceptance of Medical Artificial Intelligence: A study of 3,000 US adults found that trust in medical AI was associated with clinician involvement, representative training data, and governance mechanisms such as FDA approval or certification. [Link]

The "Model Medicine" Framework: Researchers published a pre-print proposing a new discipline to treat AI models as complex ecosystems prone to operational disorders and behavioral anomalies. [Link]

AI vs. Clinicians in Surgical Video: A systematic review and meta-analysis found that AI-assisted healthcare professionals outperformed unassisted clinicians in surgical and interventional video analysis, while AI-assisted experts performed comparably to AI alone.. [Link]

Preventing Venous Thromboembolism: A narrative review found that AI methods can improve VTE risk prediction over traditional models by capturing complex patterns in EHR data, while newer approaches also incorporate unstructured notes, imaging, and wearable time-series data for more dynamic assessment. [Link]

Beckman Coulter Diagnostics & Innovaccer (US): Partnered to modernize clinical lab operations using Innovaccer’s Gravity AI and data platform. [Link]

CVS Health & Google Cloud (US): Partnered to build the Health100 platform. [Link]

Regard & Microsoft (US): Integrated Regard’s clinical intelligence into Microsoft’s Dragon Copilot workflow. Sentara Health is piloting the setup to help clinicians identify relevant diagnoses and improve documentation in real time during patient encounters. [Link]

Harrison.ai (Australia): Expanded its Open Platform by adding four new AI partners - AIRAmed, Koios Medical, Lunit, and Nanox AI [Link]

Fujitsu & DT-Axis (Japan): Signed an MoU to support the development of Software as a Medical Device (SaMD), including regulatory approval and commercialization pathways for AI-powered diagnostic and therapeutic applications. [Link]

Procode AI (US): Raised $4 million in funding to expand its AI-powered revenue cycle management platform for private practice surgeons, including coding automation and related billing workflows. [Link]

Cognito Therapeutics (US): Raised $105 million in Series C funding to complete its HOPE pivotal trial and prepare regulatory submission for its Spectris neurotechnology therapy targeting Alzheimer’s disease. [Link]

KeyCare (US): Raised $27.4 million to scale its Epic-based virtual care platform for health systems, with plans to deepen investment in AI-enabled technology and expand coordinated 24/7 virtual care. [Link]

Antiverse (UK): Raised $9.3 million in Series A funding to design antibodies for difficult-to-drug targets using AI models and in-house lab validation. [Link]

Cent (India): Raised $5 million in seed funding to build a preventive healthcare platform combining advanced imaging, biomarker testing, and AI-powered analysis for early disease detection. [Link]

Sectra (Sweden): Acquired AI radiology company Oxipit to integrate its CE-certified autonomous chest X-ray analysis technology (ChestLink) into Sectra’s enterprise imaging platform [Link]