This biweekly video recap in AI in Healthcare and Digital Health covers regulatory initiatives, clinical workflow advances, platform capabilities, research integrity, and strategic partnerships shaping care delivery and evidence generation.
🎯 Watch Our Video Summary Capturing AI in Healthcare and Digital Health News from the Last Two Weeks
Dive deeper
🗓️ Explore more weekly details and sources
- Week 12– 18 January 2026
- Week 19–25 January 2026
📚 Find your one-stop page for the full AI and Digital Health archive.
Top Stories Covered In This Video
Chapters
0:00 Introduction
0:10 NeurIPS 2025 papers flagged for fabricated citations and invalid references
0:47 ARPA-H ADVOCATE program to develop agentic AI for cardiovascular care with a 39 month FDA path
1:24 Bristol Myers Squibb and Microsoft to deploy FDA cleared radiology AI for earlier lung cancer detection
2:12 OpenAI’s compute and revenue scaling strategy toward practical 2026 adoption and agents
2:49 OpenAI invests in Merge Labs, a noninvasive brain computer interface startup using ultrasound and molecules
3:18 Google Gemini Personal Intelligence beta connects Gmail, Photos, YouTube, and Search with opt in privacy controls
3:50 Anthropic expands Claude for Healthcare and Life Sciences with HIPAA ready connectors and new agent skills
4:38 How to reach us
Transcript
Welcome to the latest edition of AI in Healthcare and Life Sciences Updates, covering breakthroughs in the past two weeks. Brought to you by LucidQuest.
First, research integrity made headlines. An analysis by GPTZero of 4,841 NeurIPS 2025 accepted papers found at least 100 fabricated citations across 51 papers, including fictitious authors, invalid DOIs, and incomplete arXiv IDs. These papers passed multi-reviewer peer review in a year with a 24.52 percent acceptance rate. The finding underscores pressure on conference workflows and the need for stronger citation checks.
Next, ARPA-H is launching ADVOCATE to build two agentic AI systems for cardiovascular care. One is a patient-facing agent for always-on support. The other is a disease-agnostic supervisory agent for ongoing monitoring of learning systems. The aim is an FDA authorization pathway over 39 months. Teams are slated for selection around June 2026, with capabilities that connect to EHRs and wearables, help schedule visits, and suggest medication adjustments under clinical oversight.
In imaging, Bristol Myers Squibb and Microsoft announced a collaboration to accelerate early detection of lung cancer. FDA-cleared radiology AI will be deployed via Microsoft’s Precision Imaging Network, which Microsoft says is used by more than 80 percent of U.S. hospitals to share imaging and access third-party AI. The tools can surface hard-to-see lung nodules on X-ray and CT, and track follow-up to reduce loss to care. The effort targets earlier diagnosis and access in underserved settings. Lung cancer remains the leading cause of cancer death in the U.S., with about 125,000 deaths and 227,000 new cases each year.
OpenAI detailed how its business scales with compute. Available capacity rose from 0.2 GW in 2023 to 0.6 GW in 2024 and about 1.9 GW in 2025, while annualized revenue rose from 2 billion dollars to 6 billion and then to more than 20 billion. The model mixes consumer and workplace subscriptions, a free tier with ads and commerce, and usage-based APIs, with a 2026 focus on practical adoption and agentic workflows.
OpenAI also invested in Merge Labs, Sam Altman’s brain-computer interface startup. Merge raised about 250 million dollars at an 850 million valuation, with OpenAI writing the largest check. Merge is pursuing noninvasive interfaces that use molecules and modalities like ultrasound to communicate with neurons. This contrasts with Neuralink’s implanted electrodes, which require surgery.
Google introduced Gemini Personal Intelligence, a U.S. beta that lets users opt in to connect Gmail, Photos, YouTube, and Search for more personalized help. Google states connected data is referenced to answer requests, with controls to choose what to link, and that it does not train models directly on inboxes or photo libraries. The feature starts with eligible AI Pro and AI Ultra subscribers and will expand over time.
Finally, Anthropic advanced Claude for Healthcare and Life Sciences. HIPAA-ready tools now include connectors to the CMS Coverage Database, ICD-10, and the National Provider Identifier registry, supporting tasks such as prior authorization checks, coding, and provider verification. For life sciences, new connectors include Medidata, ClinicalTrials.gov, bioRxiv, medRxiv, Open Targets, ChEMBL, ToolUniverse, and Owkin. Anthropic highlights improvements with Opus 4.5 on medical agent benchmarks, plus skills for FHIR development and a sample workflow for prior authorization and clinical trial protocol drafts.
Stay ahead in AI in Healthcare and Life Sciences research! Like, share, and subscribe for our updates. Visit www.lqventures.com or email us at info@lqventures.com for expert healthcare consulting. See you next time!
Why it matters
- Hospitals and payers are piloting AI that touches high-stakes clinical workflows, from imaging to authorization.
- Regulators are being tested by agentic systems, prompting new oversight patterns.
- Platform players are racing to personalize assistants while stressing privacy controls.
- Investment continues in neurotech and foundation models, expanding modality reach.
- Research integrity concerns in AI conferences could slow translation into practice.
🗓️ Explore more weekly details and sources
- Week 12– 18 January 2026
- Week 19–25 January 2026
📚 Find your one-stop page for the full AI and Digital Health archive.
FAQ
What is ADVOCATE, and when could systems be authorized?
ARPA-H’s ADVOCATE funds a CVD patient agent and a supervisory agent, aiming for a 39-month path to FDA authorization. Team selections are targeted for June 2026 [2].
How will the BMS–Microsoft collaboration work in clinics?
FDA-cleared radiology AI will run on Microsoft’s Precision Imaging Network to analyze X-ray and CT images, surface nodules, and track follow-up, with an equity focus on underserved sites [3].
Did NeurIPS 2025 actually accept papers with fake citations?
Yes. GPTZero reported at least 100 hallucinated citations across 51 accepted papers among 4,841 analyzed, including fictitious authors and invalid DOIs [1].
How is OpenAI funding its growth?
OpenAI ties monetization to usage across subscriptions, ads/commerce, and APIs, supported by scaled compute from 0.2 GW to ~1.9 GW over 2023–2025 and ARR rising to $20B+ in 2025E [4].
What is Merge Labs building vs. Neuralink?
Merge targets noninvasive, molecule-based BCIs using modalities like ultrasound, while Neuralink uses implanted electrodes. OpenAI invested in Merge’s ~$250M seed round [5].
What personal data does Gemini’s Personal Intelligence use?
It references connected app data to answer prompts, with opt-in controls, and Google states it does not train directly on Gmail or Photos content for model training [6].
How can Claude help with payer/provider workflows?
Claude for Healthcare adds connectors to CMS, ICD-10, and NPI, plus skills for FHIR and prior authorization reviews, supporting claims, appeals, and care coordination in HIPAA-ready setups [7].
Entities / Keywords
NeurIPS 2025; GPTZero; ARPA-H ADVOCATE; FDA; heart failure; Bristol Myers Squibb; Microsoft Precision Imaging Network; OpenAI revenue and compute; Merge Labs BCI; Google Gemini Personal Intelligence; Anthropic Claude for Healthcare; Medidata; CMS Coverage Database; ICD-10; NPI; ClinicalTrials.gov.