OpenAI model surfaces rare disease leads

AI gives unsolved cases another pass

A child’s old genetic file can still produce a new answer.

OpenAI says researchers used its o3 Deep Research model to reanalyze 376 previously unsolved rare disease cases. After expert review, testing, and clinical confirmation, physicians established 18 diagnoses, a 4.8% additional yield.

That matters. The most useful part here is not that AI “diagnosed” anyone, because the article is careful to say it did not. I think the real signal is that AI may help clinicians keep old cases active when papers, variant databases, and lab classifications keep changing.

The work is limited. The study did not measure time saved, cost, false-positive burden, or whether care changed after diagnosis.

Still, there is a human detail that sticks: a genetic counselor called Kyra about a week before her 28th birthday after nearly 20 years without an answer.

How many old files deserve another look?

Copilot usage gets easier to audit

GitHub is putting a cost counter next to each Copilot user.

The Copilot usage metrics API now includes an ai_credits_used field in user-level reports, covering single-day and 28-day views at enterprise and organization levels. It shows each user’s total daily AI credit consumption across Copilot activity, using the same source data as the usage-based billing API.

This is practical. I think the useful part is the move toward treating AI use like any other managed enterprise resource. Admins can now compare credit use with adoption patterns without opening a separate billing tool.

The limit matters too: GitHub says the number is not split by feature, model, or surface, and it is a metrics signal rather than a billed total. On a gray admin screen, this will feel like a new column with real money behind it.

Who used the credits?

Bots are entering the doctor call queue

A UnitedHealth bot is now another voice on the doctor’s phone.

Bloomberg reports that UnitedHealth is spending $3 billion on AI, including bots that call doctors through UnitedHealthcare. The detail landed in Modern Healthcare on June 19, 2026, alongside other health tech and insurance stories.

That is a lot. The money matters, but the more telling detail is where the AI shows up first: in the dull, high-friction space between insurers and clinicians. I think this is less about futuristic medicine and more about automating the administrative grind that doctors already complain about.

The risk is obvious in a quiet exam-room hallway: more machine-led contact could save time, or it could add another layer of irritation. For UnitedHealth, the test will be whether doctors feel less burdened after the calls start.

Who gets the time back?

Prohuman team

Free Guides

Explore our free guides and products to get into AI and master it.

All of them are free to access and would stay free for you.

Feeling generous?

You know someone who loves breakthroughs as much as you do.

Share The Prohuman it’s how smart people stay one update ahead.