Trister: AI can help us understand mental health

The ideas and innovators shaping health care

Apr 08, 2024 View in browser   By Ruth Reader, Erin Schumaker and Daniel Payne THE LAB Trister | Rod Lamkey for POLITICO Alphabet health company Verily and pharmaceutical brand Otsuka are embarking on a multiyear study using artificial intelligence to better understand and support people with mental health conditions. Verily is enrolling 500 subjects with symptoms of depression in its digital clinical trial. The study’s main goal is to understand the real-world effects of treatment at the onset of symptoms. Participants will regularly share how they’re feeling via self-assessments. But Verily’s chief science officer, Andrew Trister, said he’d eventually like to include more tangible data points like digital biomarkers. “Leveraging phones and, say, wearables that have different sensors on them, can we make some sort of inference about what’s happening with the person in this remote model because we have a continuous measure?” he asked. The study will also look at the effects of prescribed drug therapy. Why it matters: A fifth of American adults experience mental illness, according to a 2023 survey from the nonprofit Mental Health America. About 28 percent don’t get the care they need. What’s next: Trister believes the study could provide insights that lead to more precise diagnosis and treatment for patients, which could help Otsuka develop more targeted drugs for mental health conditions.   YOUR GUIDE TO EMPIRE STATE POLITICS: From the newsroom that doesn’t sleep, POLITICO's New York Playbook is the ultimate guide for power players navigating the intricate landscape of Empire State politics. Stay ahead of the curve with the latest and most important stories from Albany, New York City and around the state, with in-depth, original reporting to stay ahead of policy trends and political developments. Subscribe now to keep up with the daily hustle and bustle of NY politics.      WELCOME TO FUTURE PULSE San Diego, Calif. | Daniel Payne This is where we explore the ideas and innovators shaping health care. We want to hear about — and see photos from — your eclipse experience. Send your best shots our way! Send tips securely through SecureDrop, Signal, Telegram or WhatsApp.   PROBLEM SOLVERS Bertagnolli | AP The NIH Council of Councils, which advises the director, voted unanimously Friday to pilot a clinical research network in primary care settings. “We are very worried about the overall health of the United States’ population,” said NIH Director Monica Bertagnolli, citing “deaths of despair” from drug overdoses, alcohol, suicide and obesity. Why it matters: Individuals with low incomes, people of color and those in rural areas are underrepresented in medical research and disproportionately likely to experience poor health. “They’re the same communities that we need to do more to engage,” Bertagnolli said. “We are losing ground instead of making progress in certain very key areas, and we must reverse this trend.” The plan: Pilot a cross-NIH research network in a small number of primary care sites for two years, with a goal of engaging underserved communities, such as minority groups, older adults, children and rural Americans. A three-year implementation phase would follow. The role of primary care providers is vital because they are connected to underserved communities and provide long-term care to their patients. The bottom line: While the agency anticipates $5 million in funding for the program in fiscal year 2024 and $25 million in fiscal year 2025, Bertagnolli acknowledged the larger reality of NIH’s tight budget. “Everyone needs to realize we have a negative cash balance at NIH this year and probably next year,” she said, explaining the rationale behind initially focusing on a modest set of already funded sites. “If we’re an inch deep and a mile wide, we will fail.”   Easily connect with the right N.Y. State influencers and foster the right relationships to champion your policy priorities. POLITICO Pro. Inside New York. Learn more.     EXAM ROOM AI systems can now sometimes pass medical exams. | Matt Rourke/AP Even when large language AI models aren’t specially trained with medical information, they could be close to passing board exams. That ability was highlighted today in JAMA Neurology, where researchers tested the abilities of Chat GPT-4, Google’s Bard and Anthropic’s Claude 2 to perform on practice board exams from the American Epilepsy Society. How did they do? Across eight practice exams, each with 100 questions, the models’ scores almost always hovered between 60 and 80 percent. Though there’s no set score to pass the board exam, anecdotal reports suggest passing scores are usually 70 percent or above, the authors write. The authors noted that the models often arrived at the right answer with correct logic to explain how an answer was determined — though sometimes made up the logic that led to the wrong answer. Other promising signs: Advanced AI models have done well in other medical exams, such as GPT-4 passing a practice version of the United States Medical Licensing Examination. Even so: The skills needed for the exam are far different than those needed in the exam room, the authors warned — and the latter could be even more challenging for advanced AI systems.   Follow us on Twitter Carmen Paun @carmenpaun Daniel Payne @_daniel_payne Ruth Reader @RuthReader Erin Schumaker @erinlschumaker   Follow us

To change your alert settings, please log in at https://login.politico.com/?redirect=https%3A%2F%2Fwww.politico.com/settings This email was sent to rouf@idiot.cloudns.cc by: POLITICO, LLC 1000 Wilson Blvd. Arlington, VA, 22209, USA Unsubscribe | Privacy Policy | Terms of Service