After patients die, they often remain alive in medical records

My Health Tech co-author Mohana Ravindranath has a wonderful story about a disturbing problem: When patients die, they often aren't marked as dead in medical records.

After discovering the problem while studying seriously ill primary care patients, health professor Neil Wenger set out to find how widespread it was. In a study of 12,000 seriously ill patients from UCLA Health he found that several hundred who were marked as alive in records but dead in state public health files.

Wenger has made it his mission to fix the problem, which not only leads to waste in the form of effort expended trying to reach dead patients, but could also skew the tsunami of predictive algorithms that are being trained on patient data.

Read more here

Medical Devices

Senators call for inquiry into device recalls

Sens. Dick Durbin (D-Ill) and Richard Blumenthal (D-Conn) this week sent a letter to the Government Accountability Office asking for a review of the Food and Drug Administration's recent track record on medical device recalls. The letter follows the FDA's questionable handling of the disastrous recall of millions of faulty Philips CPAP machines that have been linked to hundreds of deaths. More specifically, the senators are calling the GAO to update a 2011 report entitled, "Medical Devices: FDA Should Enhance Its Oversight of Recalls." The senators go on to list a series of questions GAO ought to ask about what exactly FDA is doing to ensure that necessary recalls happen in a timely fashion.

Artificial intelligence

Google announces new health care AI models

Google this week announced MedLM, a family of generative AI models tailor made for the health care industry and available through Google Cloud. The new general purpose models, which come in large- and medium-size flavors, allow health care customers to tackle a range of problems. Google's blog post highlights the ability to answer medical questions and drafting summaries as use cases where the company identified customer need. The company also promised that there will be forthcoming MedLM models based on Google's new Gemini AI models that are capable of working across images, text, audio, and other forms of data

regulation

New rules demand transparency into health AI

The Biden Administration is laying down a new law in health AI. Starting in 2025, electronic health record vendors who supply predictive models to health systems will have to disclose details about how they were developed, tested, and trained.

The new rules from the Office of the National Coordinator for Health Information Technology are designed to put guardrails around a new generation of machine learning tools gaining rapid adoption in hospitals nationwide. Specifically, EHR vendors such as Epic and Oracle-Cerner will have to disclose the contents and sources of training data, explain how their models were tested, and enumerate steps taken to identify and control risks. The rule, in all its glory, is 912 pages. Thankfully, STAT's Casey Ross has written a story to help you make sense of the most important aspects of it.

Read more here

digital therapeutics

Bankrupt Pear's addiction treatments sold again

Half a year after selling its assets at a fire sale bankruptcy auction, Pear Therapeutics' most significant digital treatments will find a new life as part of PursueCare, which provides online addiction care and mental health counseling. The company hopes to build a business model around the apps where Pear failed.

CEO Nick Mercadante told me that while PursueCare faces many of the same challenges as Pear, it's got more flexibility because it doesn't need to find a way to make buckets of money on the apps fast. PursueCare didn't spend hundreds of millions of dollars developing the products and has existing lines of care and pharmacy businesses it expects to be profitable soon. Plus, its current businesses can help grease the wheels of adoption and payment where Pear was always reliant on the goodwill of others.

Read more here

denied by ai

Humana sued over care denying algorithm

A group of Medicare Advantage patients are suing Humana alleging the insurer illegally used an algorithm to cut off rehabilitation care to seriously ill patients. It is the second class action lawsuit filed following a STAT investigation into the algorithm.

Read more here

More around STAT

Check out more exclusive coverage with a STAT+ subscription

Read premium in-depth biotech, pharma, policy, and life science coverage and analysis with all of our STAT+ articles.

What we're reading

With big ambition — and a family name to match — Reed Jobs charts his own course in biotech, STAT
Utility of smart watches for identifying arrhythmias in children, Communications Medicine
Twin Health gets $50M injection for metabolic care, Axios