Healthcare AI News 2/18/26
News
CNN reports that AI training firm Mercor is paying subject-matter experts, including physicians, hundreds of dollars per hour to validate and refine the commercial AI models of other companies. Mercor, which was founded in 2023 by three 19-year-old college dropouts who are now billionaires, generates $500 million in annual revenue and carries a $10 billion valuation. The company says that the most in-demand expertise is in software engineering, finance, medicine, and law.
In India, the chief minister of Andhra Pradesh proposes that every resident be assigned an AI-powered personal doctor to track health data and offer advice on diet, lifestyle, and preventive care. He met with Bill Gates this week to discuss integrating the state’s digital health records with other platforms.
A man says that he used Anthropic’s Claude AI to negotiate a $163,000 reduction of a hospital’s $195,000 bill for four hours of ED care that his uninsured brother-in-law received before dying of a heart attack. He asked Claude to create a worksheet that listed each billed CPT code and the corresponding Medicare payment, which totaled $29,000 for the same services. He offered that amount to the hospital, arguing that its original bill included improper unbundling, mutually exclusive services, and an inpatient procedure billed as outpatient. The hospital countered at $37,000, the parties agreed to split the difference, and they executed a settlement agreement that Claude drafted.
The United Arab Emirates launches Amal, an AI-powered physician assistant that conducts pre-visit interviews with patients, creates medical summaries for physicians, and answers post-visit patient questions. The system was developed by Boston Health AI, which offers a live, interactive online demo.
Business
Elsevier expands ClinicalKey AI clinical decision support tool with a full-text, copyright-cleared knowledge base of 1,000 journals and major guidelines, adding real-time citation traceability, daily updates, workflow APIs, and enhanced security to strengthen trust and point-of-care usability.
Mental health AI startup Kintsugi shuts down after raising $30 million and places its technology in the public domain. Its product analyzes the journal recordings of users to detect signs of anxiety or depression. CEO Grace Chang says that it is not financially viable for startups to build healthcare AI applications because investors expect rapid annual recurring revenue growth while FDA regulatory clearance takes significant time.
Research
A survey of 50 health system executives by Guidehouse and HIMSS finds that while 78% have AI projects in progress, only half feel ready to implement them. Major concerns include cybersecurity, competing financial priorities, data quality, and lack of internal expertise.
Other
Cleveland Clinic pilots an AI tool that it developed with startup Piramidal Inc. to detect seizures from 24-hour ICU EEGs, monitoring 120 patients daily whose studies would otherwise require about two hours of technician review each.
AI systems are increasingly embedded in clinical workflows to generate alerts and automate tasks, yet their unreliable outputs still require nurses to interpret and manage them in real time. Nurses recognize AI’s potential to improve documentation and monitoring, but they call for stronger testing, greater transparency, and meaningful frontline involvement in development.
Contacts
Mr. H, Lorre, Jenn, Dr. Jayne.
Get HIStalk updates.
Send news or rumors.
Follow on X, Bluesky, and LinkedIn.
Sponsorship information.
Contact us.
