AI tools to predict hospital stays are hampered by a string of shortcomings – STAT

Youre reading the web edition of STAT Health Tech, our guide to how tech is transforming the life sciences.Sign up to get this newsletterdelivered in your inbox every Tuesday and Thursday.

Biofourmis senses an opportunity

Remote patient monitoring companyBiofourmisannounced a $300 million Series D round this morning, a critical investment as the firm looks to products that not only identify when patients are likely to deteriorate, but can help manage their care. The company says that 20 hospitals have signed on to use its care at home services which includes monitoring devices and an app to help prevent costly ER care and readmissions.

advertisement

Next up: using patient data to drive treatment decisions, like titratingdrug doses. Biofourmis has a pivotal trial under way for its BiovitalsHF software but even if results are positive, it faces hurdles in patient and provider acceptance. Thats going to be the linchpin, is if the patient trusts it enough to say, Im not seeing my doctor every week, but I have to increase my dose every week, said cardiologist Ritu Thamman.Read more in Marios latest.

Getting wearables on the right hands

advertisement

When the federalAll of Usresearch programlaunched, its goal was to build a truly representative pool of participants, evening out the imbalances seen in medical research. But the effort hit a roadblock when it tried to fold in data from participants own wearable devices. When our team compared the demographics of all program participants to Fitbit data contributors, we saw the representation of historically underrepresented communities shrink, wrote All of Us team members Yashoda Sharma and Chris Lunt in a newSTATFirst Opinion.

In a newsurveyof federally qualifiedhealthcenters, they confirmed the primary barrier was cost. Thats why last year All of Us launched a program to distribute 10,000 devices to participants; today, they report about 3,000 program-provided devices are sharing data with the researchers.

Metas tool to fight Covid-19 misinformation

WhatsApp, theMeta-owned instant messaging company, has launched a chatbot for theCalifornia Department of PublicHealthto help combat Covid-19 misinformation, particularly in the states Latino community. The free tool, available in English and Spanish, offers users a menu of options, including local vaccination sites, transportation options, and FAQs to help provide factual information about the vaccines. Meta, which also ownsFacebook, has been under pressure throughout the pandemicto fight misinformation, as vaccine skeptics have used its platforms to circulate false information on Covid-19 and the vaccines.

AI tools to predict hospital stays are all over the map

Hospitals across the globe are using AI algorithms to help predictlength of stayto target care to the neediest patients, cut costs, and improve operations. But a newsystematic reviewpublished inPLOS DigitalHealthfound they are hampered by a variety of shortcomings, such as data cleaning approaches that limit their generalizability beyond individual institutions. It calls for the development of a unified framework that would make it easier to assess the quality and durability of these tools and help identify populations with higher risks of adverse events such as hospital acquired infections.

Making the rounds

Other new business

New jobs all around

What were reading

Read the original:

AI tools to predict hospital stays are hampered by a string of shortcomings - STAT

Related Posts

Comments are closed.