Hi friend,
Let’s get straight into it. What’s been going on?
What happened in mental health last week?
Business News:
⏺️ LearnWell announced an investment from Goldman Sachs. Learnwell is a student behavioural health service provider that has been around since 1995 and serves over 51,000 students each year. (link)
⏺️ Hazel cut staff. Hazel Health cut 135 employees in its second layoff announced in 4 months. (link)
⏺️ OneMind announced their 2026 accelerator cohort. The successful companies include Biomia, Birches Health, Eratos Therapeutics, Flow Neuroscience, Forecast Bio, Limbic, Manatee, Monument Therapeutics, Neurofundus, Nosis Health, Orbit Neuro, Radicle Science, Social RX and Vitalic Health Science, If I could own an index of mental health startups, it would be the OneMind Accelerator graduates. I’m very excited to see what these companies do and will be sharing more on them soon!
⏺️ UK-based Self Space launched a crowdfunding campaign. I’m a fan of Self Space and am interested to see how this campaign performs. (link)
⏺️ Congress extended medicare telehealth coverage for two-years. A permanent extension - which many are desperate for - was not included. (link)
⏺️ The US government funded nine pilot programs with $20 million. The program will test new data standards that would make it easier for mental health providers, addiction treatment centres, and doctors to share patient information. The insights from the pilots will shape industry-wide requirements and a new Behavioral Health Information Resource launching in 2027. (link)
⏺️ Behavioural health conditions are more common than diabetes among commercially insured. This report from FAIR Health showed just how prevalent mental health conditions. For any healthcare data nerds amongst us, this report is a fun read, lots of fascinating insights on prevalence and costs of chronic conditions (link)
⏺️ Researchers (including from Microsoft AI) introduced SIM-VAIL, an automated testing framework that simulates realistic multi-turn conversations between vulnerable users and AI chatbots. This is an interesting read for anyone looking at AI safety. (link)
⏺️ Researchers from the University of Louisiana launched MHDash, an open-source platform for benchmarking mental health-aware AI assistants. The platform revealed that simple baselines and advanced LLMs exhibited comparable overall accuracy yet diverged significantly on high-risk cases. (link)
⏺️ KORA, the first public benchmark for AI child safety, was launched. I’m glad to see more people building safety benchmarks for this space. Now they just need to be used to hold builders accountable.(link)
Research News:
⏺️ A new study found that 40% of physicians enrolled in Medicaid didn't treat patients covered by the program in 2021. (link)
⏺️ Study on postpartum mental health screening. An Irish study of 2,572 first-time mothers found that at three months postpartum only 50-70% reported being asked about mental health, with screening dropping sharply after that. The study has some more interesting findings and I’d recommend reading it for anyone interested in postnatal mental health. (link)
⏺️ AI to predict Substance Use Disorder. University of Cincinnati researchers built an AI that predicts substance use disorder with 83% accuracy using a simple smartphone taskThe patient looks at 48 pictures and rates how much they like or dislike each one. The AI uses those ratings to calculate behavioral economics variables—how the person weighs gains vs losses, how much risk they'll tolerate, how consistent their preferences are. Those variables create a unique behavioral profile that correlates with substance use disorder, allowing the AI to predict addiction with 83% accuracy. (link)
⏺️ Nema Health released a study on the effectiveness of Virtual massed Cognitive Processing Therapy for PTSD. Nema released its first peer-reviewed publication showing virtual massed Cognitive Processing Therapy for PTSD produced large effect size improvements in PTSD symptoms (d=2.94), anxiety (d=1.79), and depression (d=1.49) from intake to discharge, with improvements maintained at 90 days. (link)
⏺️ Smart emotion detection using body sensors. Researchers built a system that recognises six emotions (happy, sad, angry, fearful, surprised, neutral) by monitoring internal body signals like heart rate, blood pressure, temperature, blood sugar, oxygen levels, and muscle activity through wearable IoT sensors. I mean, I’m not sure I need an electromyography sensor to tell me I’m sad, but I guess this could have some interesting applications? Right?? (link)
That’s all for this week. I hope you found it valuable.
If you’d like to discuss this news with like-minded peers, consider becoming a Hemingway Pro member. You will join our community of over 330 mental health leaders, and get access to exclusive content. You can learn more here.
Keep fighting the good fight!
Steve Duke
Founder of Hemingway
