Anthropic introduces Claude for Healthcare
The generative artificial intelligence company Anthropic has released Claude for Healthcare, a complementary set of tools and resources that allow health care providers, payers, and consumers to further the use of the Claude AI assistant and chatbot for medical purposes.
Users of Claude for Healthcare can give the platform direct access to the Centers for Medicare and Medicaid Services Coverage Database; International Classification of Diseases, 10th Revision; and National Provider Identifier Registry. This reduces the time that users spend finding data and generating reports, according to Anthropic.
Those U.S. clients who use the Claude Pro and Claude Max subscription plans can give Claude for Healthcare secure access to their laboratory results and health records. “When connected, Claude can summarize users’ medical history, explain test results in plain language, detect patterns across fitness and health metrics, and prepare questions for appointments,” Anthropic reported.
The tool can also be used to coordinate care and triage patient portal messages, referrals, and handoffs, as well as speed up review of prior authorization requests and support claims appeals.
Stanford system assesses accuracy of AI-generated clinical documentation
Researchers at Stanford University have developed VeriFact, an artificial intelligence-based system for verifying the accuracy of statements in a clinical document about a patient that were generated via large language models by cross-checking the documentation against facts in a patient’s electronic health record.
The researchers developed VeriFact as part of a study (Chung P, et al. NEJM AI. Published online Dec. 24, 2025. doi:10.1056/Aldbp2500418) to address growing concerns about LLM hallucinations and other errors that can limit use of the technology.
The system, which is based on an LLM-as-a-judge framework, in which LLMs assess the quality of text outputs from other LLMs, performs reference-based, patient-specific fact verification of EHR data to assess whether statements in AI-generated documents are accurate, instead of relying on general medical knowledge to make such determinations.
The authors concluded that “VeriFact can help clinicians verify facts in documents drafted by LLMs prior to committing them to the patient’s EHR and can automate tasks requiring chart review. VeriFact-BHC [the clinician-annotated data set VeriFact–Brief Hospital Course] can be used to develop and benchmark new methodologies for verifying facts in patient care documents.”
The authors acknowledged, however, that the study has limitations. Among them, that it includes summaries from a single LLM and documentation creation pipeline, uses a fixed set of prompts for each task, and focuses on discharge summaries from a specific data set.
OpenAI purchases health care startup Torch
The artificial intelligence research and deployment company OpenAI has acquired the health care startup Torch to build out its ChatGPT Health offering.
OpenAI recently launched ChatGPT Health, which connects the company’s AI chatbot with users’ medical records and various wellness apps for more personalized answers to medical questions.
Torch developed an app to consolidate users’ medical data from such sources as laboratories, hospitals, wearable devices, and consumer health testing companies to provide a single comprehensive view of a patient’s health information. The product unifies lab test results, medications, and recordings of patients’ consultations with physicians.
Animal disease diagnostic lab selects Gestalt solution
The Indiana Animal Disease Diagnostic Laboratory at Purdue University has selected Gestalt Diagnostics’ PathFlow platform to support its digital pathology initiatives.
“The deployment will begin with PathFlow’s AP (clinical) module, enabling pathologists to streamline their workflows [and] have faster collaboration and access to their whole slide images for diagnosis, reporting, and image analysis,” according to a press statement from Gestalt. “Through collaboration across the university, the ADDL hopes to expand its use of PathFlow to include educational and research workflows—empowering students, faculty, and scientists with leading-edge tools for digital pathology training and discovery.”
Google upgrades open next-gen AI model
Google has released MedGemma 1.5 4B, the latest update in its MedGemma collection of open medical generative artificial intelligence models, released last year.
This updated open-source version expands on MedGemma 1’s capabilities for two-dimensional medical imaging by offering support for three-dimensional volume representations of computed tomography and magnetic resonance imaging, as well as whole slide histopathology imaging.
Release 1.5 4B also has significantly better baseline performance than MedGemma 1 4B for such tasks as medical lab report data extraction; medical image interpretation for dermatology, histopathology, and other areas; longitudinal medical imaging; and localization of anatomical features in chest x-rays, according to a Google press announcement.
MedGemma 1.5 4B is free for commercial and research use. It can be downloaded from Hugging Face or trained and adapted into scalable applications in the cloud on Vertex AI.