<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"><channel>
  <title>Textbook of Digital Health · News</title>
  <link>https://www.textbookofdigitalhealth.com/news/</link>
  <description>Regular round-up of digital health stories.</description>
  <language>en-GB</language>
    <item>
      <title>Mayo Clinic and Microsoft to build a 'frontier' AI model for healthcare, owned by Mayo</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-06-02-mayo-microsoft-frontier-model.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-06-02-mayo-microsoft-frontier-model.html</guid>
      <pubDate>Tue, 02 Jun 2026 00:00:00 GMT</pubDate>
      <description>Mayo Clinic and Microsoft have announced a collaboration to build a 'frontier' AI model purpose-built for healthcare. The model, which will be owned by Mayo Clinic and draws on its de-identified clinical data, is designed to support clinical reasoning across a broad range of uses; Microsoft plans to offer it to other organisations through its Azure AI Foundry.</description>
    </item>
    <item>
      <title>Coalition for Health AI publishes governance playbooks to help health systems oversee their use of AI</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-29-chai-governance-playbooks.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-29-chai-governance-playbooks.html</guid>
      <pubDate>Fri, 29 May 2026 00:00:00 GMT</pubDate>
      <description>The Coalition for Health AI, a non-profit bringing together health systems, technology firms, patient groups, and regulators, has released governance playbooks to help health systems oversee their use of AI. Developed with more than 150 industry contributors and over 100 healthcare organisations, the guidance is organised around eight dimensions of organisational risk and is meant to scale from large academic centres to small community clinics.</description>
    </item>
    <item>
      <title>HHS restructures its Office for Civil Rights, adding a health information privacy and cybersecurity division</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-26-hhs-ocr-restructure.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-26-hhs-ocr-restructure.html</guid>
      <pubDate>Tue, 26 May 2026 00:00:00 GMT</pubDate>
      <description>The HHS Office for Civil Rights, the federal enforcer of HIPAA, will split into three subject-matter divisions: Conscience and Religious Freedom, Civil Rights, and Health Information Privacy, Data, and Cybersecurity. Breach reviews and complaint intake remain handled by a separate Enforcement Division.</description>
    </item>
    <item>
      <title>Innovaccer acquires CaduceusHealth in $66M deal to extend AI platform into revenue cycle</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-26-innovaccer-caduceushealth.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-26-innovaccer-caduceushealth.html</guid>
      <pubDate>Tue, 26 May 2026 00:00:00 GMT</pubDate>
      <description>Innovaccer's fifth acquisition adds outsourced revenue cycle management for approximately 4,000 ambulatory providers, with CaduceusHealth's coding and insurance-policy data becoming training material for Innovaccer's RCM agents.</description>
    </item>
    <item>
      <title>Mayo Clinic and Bayesian Health publish RCT-validated AI for identifying palliative-care candidates</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-26-mayo-bayesian-palliative-care.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-26-mayo-bayesian-palliative-care.html</guid>
      <pubDate>Tue, 26 May 2026 00:00:00 GMT</pubDate>
      <description>A randomised clinical trial at Mayo Clinic reported a 44% increase in timely palliative-care referrals, a 25% reduction in 60-day readmissions, and a 28% reduction in 90-day readmissions when an AI tool flagged hospitalised patients with serious illness for clinician review.</description>
    </item>
    <item>
      <title>Verizon 2026 Data Breach Investigations Report: 1,492 healthcare incidents, third parties involved in 32%</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-26-verizon-dbir-healthcare.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-26-verizon-dbir-healthcare.html</guid>
      <pubDate>Tue, 26 May 2026 00:00:00 GMT</pubDate>
      <description>Verizon's 2026 Data Breach Investigations Report records 1,492 healthcare incidents and 1,438 confirmed data disclosures. Vulnerability exploitation is the leading initial-access vector at 20%, followed by phishing, stolen credentials, and employee error. A third of healthcare breaches now involve a third-party vendor.</description>
    </item>
    <item>
      <title>CMS Medicare App Library opens with first vetted digital-health apps for 69 million beneficiaries</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-24-cms-medicare-app-library.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-24-cms-medicare-app-library.html</guid>
      <pubDate>Sun, 24 May 2026 00:00:00 GMT</pubDate>
      <description>The US Centers for Medicare and Medicaid Services (CMS) have launched the Medicare App Library, a directory of patient-facing digital-health applications available to Medicare beneficiaries. Four applications are listed at launch, with more than fifty under evaluation. Listing requires identity verification (ID.me or CLEAR), FHIR-based connection to a CMS-aligned data-exchange network, and external review by the Digital Medicine Society or the CARIN Alliance.</description>
    </item>
    <item>
      <title>AMA 2026 survey: 81% of US physicians now use AI professionally, more than double the 2023 rate</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-22-ama-physician-ai-survey.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-22-ama-physician-ai-survey.html</guid>
      <pubDate>Fri, 22 May 2026 00:00:00 GMT</pubDate>
      <description>The American Medical Association's third annual physician AI survey finds 81% of US physicians using AI in practice (up from 38% in 2023), with an average of 2.3 use cases per physician; documentation and research summarisation are the most common. Patient privacy and physician-patient relationship remain the leading concerns.</description>
    </item>
    <item>
      <title>FDA rejects Harrison.ai proposal to partially exempt AI imaging diagnostics from 510(k) review</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-22-fda-rejects-harrisonai.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-22-fda-rejects-harrisonai.html</guid>
      <pubDate>Fri, 22 May 2026 00:00:00 GMT</pubDate>
      <description>Australian health-AI company Harrison.ai had asked the agency to exempt categories of AI-enabled diagnostic and detection software from premarket review under defined conditions. The FDA closed the docket in May, citing that prior authorisation of one device does not predict future-product safety from the same developer.</description>
    </item>
    <item>
      <title>NYC Health + Hospitals breach exposes medical data and fingerprints of 1.8 million people</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-22-nychhc-breach.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-22-nychhc-breach.html</guid>
      <pubDate>Fri, 22 May 2026 00:00:00 GMT</pubDate>
      <description>The largest US public-hospital system says attackers were inside its network for three months via a third-party vendor, exfiltrating medical records and biometric scans before being detected in early February. Notifications to affected individuals began in mid-May.</description>
    </item>
    <item>
      <title>FDA opens AI-enabled clinical trials pilot, starting with AstraZeneca and Amgen</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-19-fda-ai-trials-pilot.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-19-fda-ai-trials-pilot.html</guid>
      <pubDate>Tue, 19 May 2026 00:00:00 GMT</pubDate>
      <description>The FDA's new pilot reviews trial data in real time using AI, beginning with an AstraZeneca lymphoma Phase 2 trial and an Amgen small-cell lung cancer Phase 1b trial. A formal Request for Information on the broader programme ran until late May 2026.</description>
    </item>
    <item>
      <title>RXNT EHR breach exposes patient data, including records held by Congress's attending physician</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-19-rxnt-breach.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-19-rxnt-breach.html</guid>
      <pubDate>Tue, 19 May 2026 00:00:00 GMT</pubDate>
      <description>A March 2026 intrusion into the RXNT EHR/practice-management platform exposed patient names, dates of birth, contact details, patient IDs, and prescription information. The Office of the Attending Physician serving members of Congress is among the affected customers.</description>
    </item>
    <item>
      <title>VA restarts Oracle EHR deployment with four Michigan hospitals</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-19-va-oracle-ehr-michigan.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-19-va-oracle-ehr-michigan.html</guid>
      <pubDate>Tue, 19 May 2026 00:00:00 GMT</pubDate>
      <description>The Department of Veterans Affairs went live with the Oracle (formerly Cerner) EHR at four Michigan medical centres on 11 April 2026, the first of 13 planned 2026 sites. It is the largest single-day VA deployment to date.</description>
    </item>
    <item>
      <title>Four US healthcare data breaches disclosed this week</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-18-breach-roundup.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-18-breach-roundup.html</guid>
      <pubDate>Mon, 18 May 2026 00:00:00 GMT</pubDate>
      <description>Western Orthopaedics, Community Health Systems California, Tri-Cities Gastroenterology, and Integrated Pain Associates have all disclosed breaches. The PEAR ransomware group leaked the Western Orthopaedics dataset after a refused ransom.</description>
    </item>
    <item>
      <title>NHS Federated Data Platform uses AI to predict A&amp;E demand</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-18-fdp-ae-forecast.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-18-fdp-ae-forecast.html</guid>
      <pubDate>Mon, 18 May 2026 00:00:00 GMT</pubDate>
      <description>An AI tool on the NHS FDP combines weather forecasts, admission histories, and day-of-week trends to predict surges in emergency department demand so trusts can pre-plan staffing, beds, and capacity.</description>
    </item>
    <item>
      <title>JAMA study: ambient scribes save 16 minutes per 8-hour day; separate study links use to lower burnout</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-18-jama-scribe-study.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-18-jama-scribe-study.html</guid>
      <pubDate>Mon, 18 May 2026 00:00:00 GMT</pubDate>
      <description>Across 5 academic medical centres, AI scribes saved 16 minutes of documentation per 8 patient-hours on average. Heavy users saw 27 minutes saved, and burnout dropped from 51.9% to 38.8% in a separate 263-clinician quality-improvement study.</description>
    </item>
    <item>
      <title>King's College Hospital rolls out cardiac AI for angioplasty</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-18-kings-cardiac-ai.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-18-kings-cardiac-ai.html</guid>
      <pubDate>Mon, 18 May 2026 00:00:00 GMT</pubDate>
      <description>AI-powered intracoronary imaging helps cardiologists characterise plaque and assess flow in real time during PCI. Planned to support around 500 patients a year at King's.</description>
    </item>
    <item>
      <title>Digital health round-up: 17 May 2026</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-17-roundup.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-17-roundup.html</guid>
      <pubDate>Sun, 17 May 2026 00:00:00 GMT</pubDate>
      <description>Epic launches ambient AI Charting, altering the market for third-party scribes. NHS England moves source code to private repos. FDA loosens oversight of low-risk AI and wearables. Q1 funding hits $4B but concentrates in mega-deals.</description>
    </item>
    <item>
      <title>Patients give AI less detail than they give a doctor, 500-participant experiment finds</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-05-01-patients-ai-symptom-detail.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-05-01-patients-ai-symptom-detail.html</guid>
      <pubDate>Fri, 01 May 2026 00:00:00 GMT</pubDate>
      <description>In a 500-participant experiment published in Nature Health, people wrote shorter symptom descriptions when they believed an AI rather than a doctor would read them, about 28 fewer characters on average. The authors argue this information gap could degrade AI triage and diagnosis.</description>
    </item>
    <item>
      <title>21 large language models reach the right diagnosis but fail the reasoning steps, study finds</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-04-13-llm-clinical-reasoning.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-04-13-llm-clinical-reasoning.html</guid>
      <pubDate>Mon, 13 Apr 2026 00:00:00 GMT</pubDate>
      <description>A Mass General Brigham study in JAMA Network Open evaluated 21 large language models on a new clinical-reasoning benchmark. The models reached the correct final diagnosis more than 90% of the time when given complete information, but failed to produce an appropriate differential diagnosis more than 80% of the time.</description>
    </item>
    <item>
      <title>Utah's mental-health AI rules offer a 'safe harbour' model for regulating therapy chatbots, npj analysis argues</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-03-27-utah-mental-health-ai.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-03-27-utah-mental-health-ai.html</guid>
      <pubDate>Fri, 27 Mar 2026 00:00:00 GMT</pubDate>
      <description>In an npj Digital Medicine analysis, researchers from the University of Utah describe the state's framework for governing mental-health AI: a 'safe harbour' for chatbots that meet defined safety standards, including pre-deployment testing, crisis-escalation protocols, and ongoing monitoring. The authors present adaptive regulation as an alternative to outright bans.</description>
    </item>
    <item>
      <title>When AI gives a second opinion it tends to defer to the doctor, 70-clinician diagnosis trial finds</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-03-18-clinician-ai-collaboration.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-03-18-clinician-ai-collaboration.html</guid>
      <pubDate>Wed, 18 Mar 2026 00:00:00 GMT</pubDate>
      <description>A randomised controlled trial of 70 clinicians, led by Stanford Medicine and published in npj Digital Medicine, compared ways of combining physician and AI judgement on five patient cases. When the model gave its opinion after the clinician, it tended to defer to them; the most effective design had clinician and model assess each case in parallel, then had the model synthesise the two.</description>
    </item>
    <item>
      <title>Nine leading AI models repeat false medical claims hidden in clinical notes, Lancet study finds</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-02-09-llm-medical-misinformation.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-02-09-llm-medical-misinformation.html</guid>
      <pubDate>Mon, 09 Feb 2026 00:00:00 GMT</pubDate>
      <description>Researchers at the Icahn School of Medicine at Mount Sinai tested nine large language models with more than a million prompts and 300 physician-validated scenarios, and found the models would repeat fabricated recommendations when these were embedded in realistic discharge summaries or social-media health myths. The authors call for routine stress-testing before clinical deployment.</description>
    </item>
    <item>
      <title>AI-supported mammography cuts interval cancers by 12% in first randomised trial of 100,000 women</title>
      <link>https://www.textbookofdigitalhealth.com/news/2026-01-29-masai-ai-mammography.html</link>
      <guid isPermaLink="true">https://www.textbookofdigitalhealth.com/news/2026-01-29-masai-ai-mammography.html</guid>
      <pubDate>Thu, 29 Jan 2026 00:00:00 GMT</pubDate>
      <description>Final results from the MASAI trial, the first randomised controlled trial of AI-supported breast screening, were published in The Lancet. Among more than 100,000 Swedish women, AI support was associated with 12% fewer interval cancers and 27% fewer aggressive-subtype cancers, with no rise in false positives.</description>
    </item>
</channel></rss>