Health AI in the 4th Week of April

OpenAI (US): Released GPT-Rosalind, a life sciences reasoning model for biology, drug discovery, and translational medicine. It also launched a Codex plugin that connects scientists to 50+ scientific tools and data sources. [Link]

Hippocratic AI (US): Unveiled clinical AI agents for patient scheduling and inpatient nursing workflows. AI Front Door is powered by 31 coordinated large language models, while both tools run on the company’s Polaris safety architecture. [Link]

Cera (UK): Launched an AI lab to develop and scale tools for care capacity challenges, using anonymised data from millions of home care visits each month and building on its existing predictive technology. [Link]

Bupa (UK): Integrated an AI medical device into its at-home skin assessment pathway, delivering remote risk assessment within minutes and reporting 99.9% accuracy in ruling out melanoma. [Link]

Health Sector Coordinating Council (US): Released a guide on third-party AI risk and AI supply chain transparency, highlighting gaps in discovery and disclosure processes that make healthcare AI supply chain risks harder to manage. [Link]

National Health Authority (India): Announced development of a national AI policy for healthcare to scale AI adoption and provide a roadmap for its application across health services. [Link]

AI Governance and Economic Group (India): Formed as a high-level inter-ministerial body to guide national AI governance, strengthen coordination across sectors, and support responsible AI policy in India. [Link]

Public use of a generalist LLM chatbot for health queries: Analysed over 500,000 Copilot health conversations, showing higher proportions of personal health queries on mobile and increased activity in the evening and night time, while desktop use is more associated with research and medical paperwork tasks. [Link]

New Evidence Backs Hologic’s AI-Powered Mammography Technology for Detecting Challenging Cancers: Presented new SBI data showing Hologic’s Genius AI Detection identified and localized close to 90% of confirmed invasive lobular cancers, including 43% of cases initially interpreted as negative at screening. [Link]

A&A Practice: Used NLP and machine learning to identify cardiopulmonary disease cases from free-text preoperative documentation, surfacing perioperative risk signals aligned with clinical guidelines. [Link]

Generative artificial intelligence-driven chatbots and medical misinformation: an accuracy, referencing and readability audit: Audited five popular consumer chatbots across misinformation-prone medical queries. Found 49.6% of responses were problematic, with frequent inaccuracies, incomplete information, and unreliable citations. [Link]

Rapid Evaluation of Artificial Intelligence Technology Used for Ambient Dictation in Primary Care: Comparing the Quality of Documentation of Artificial Intelligence-Generated and Human-Produced Clinical Notes: Evaluated 11 ambient AI scribes against 18 human note takers across standardized primary care cases, finding that blinded raters consistently scored human-produced notes higher in quality. [Link]

Explainable Machine Learning for Predicting Infections That Require Hospitalization in Patients With Systemic Lupus Erythematosus: Developed an explainable ML model to predict infection-related hospitalization in SLE, identifying key risk factors such as corticosteroid use. [Link]

Fact-Checking Large Language Model Responses to a Health Care Prompt: Comparative Study: Found LLMs achieved near-expert-level accuracy in medical fact-checking but showed gaps in communicating key safety details in certain scenarios. [Link]

A Deep Learning-Based Digital Biopsy for Predicting Early Recurrence in Gastric Cancer: Developed a multimodal model that analyses routine histopathology slides alongside clinical variables to predict recurrence risk. [Link]

Gallup: Found that around 14 million Americans skipped a provider visit after using AI-generated health advice, suggesting AI is starting to shape care-seeking behaviour at scale. [Link]

Hippocratic AI (US): Launched two voice AI tools for patient access and nursing workflows, with early deployment at Cincinnati Children’s and collaboration from Cleveland Clinic and OhioHealth. [Link]

GNQ Insilico (US): Collaborated with IBM to deploy causal AI and digital twin platforms for precision medicine and clinical AI adoption. [Link]

Caris Life Sciences (US): Added UCSF to its Precision Oncology Alliance to advance biomarker-driven cancer research using AI and multimodal molecular data. [Link]

Rapid Health (UK): Integrated its AI-powered Smart Triage tool into the NHS App, enabling more than 1 million patients to receive instant triage and book appointments directly with clinicians. [Link]

Lunit (South Korea): Announced full deployment of its breast cancer AI ecosystem at Lexington Clinic, covering risk assessment, detection, quality, and patient tracking. [Link]

McKinsey & Company (Global): Reported that 50% of surveyed US healthcare organizations have implemented generative AI, up from 25% in McKinsey’s fourth-quarter 2023 survey. [Link]

Click Therapeutics (US): Raised $50M from Boehringer Ingelheim to commercialize its prescription digital therapeutic CT-155 for schizophrenia. [Link]

Joyful Health (US): Raised $17M to build AI-powered financial infrastructure that helps healthcare providers recover lost revenue. [Link]

Keebler Health (US): Raised $16M to process unstructured clinical documentation for risk adjustment and missed HCC coding. [Link]

Ethermed (US): Raised $8.5M to automate prior authorization and medical-necessity workflows. [Link]

Worki (US): Raised $2.75M to build AI infrastructure for healthcare workforce and HR workflows. [Link]

Gyde (US): Acquired Benavest to expand its AI-powered health insurance brokerage platform, using tools such as GydeOS and its assistant Gia to automate broker workflows and support growth across ACA and Medicare. [Link]

Sectra (Sweden): Completed the acquisition of Oxipit to strengthen its position in regulatory-approved radiology AI, adding autonomous chest X-ray analysis through Oxipit’s CE Class IIb-certified ChestLink solution. [Link]