Hi friend,

AI research continues to flow. Flourish Science shared RCT results on depression and anxiety, Slingshot showed that PHQ-9 depression scores can be predicted from AI chat text, and researchers published a new benchmark measuring how LLMs impact cognitive atrophy.

On the business side, WPS acquired women's mental health platform Mavida Health, HHS opened over $700M in funding applications, and Sonder received investment from IAG. Wellcome, one of the world’s largest mental health funders, also set out its priorities for the decade ahead.

Finally, new studies on esketamine and the SAINT TMS protocol add to a growing body of evidence on alternatives for treatment-resistant depression.

Let’s get into it.

⏺️ Klinic raised a $24M Series A to expand its provider-enablement platform. The company, founded in 2021 with a focus on behavioural health, has since expanded to support independent specialty practices across oncology, cardiology, and rare diseases. The platform combines patient acquisition, EHR, intake, care coordination, and revenue cycle tools into a single system. (link)

⏺️ IAG made a strategic investment in Sonder to extend mental health support to customers. IAG, Australia's largest general insurer, invested in Sonder, a 24/7 mental health, medical, and safety support platform, through its corporate venture arm Firemark Ventures. IAG said it planned to extend Sonder's early intervention capabilities to customers during high-risk moments. Terms were not disclosed. (link)

⏺️ WPS Health Solutions acquired women's mental health platform Mavida Health. Mavida, founded in 2023, provides virtual therapy, group therapy, and medication management for women across California, New Jersey, New York, and Texas. WPS, a Wisconsin-based nonprofit health insurer, pursued the deal as part of a new revenue diversification strategy. Terms were not disclosed. (link)

⏺️ HHS announced over $700M in newly opened funding applications for mental health and addiction services. Secretary Kennedy announced the funding across SAMHSA initiatives, including $238.6M for the 988 Suicide Crisis Lifeline, $223.1M to expand Certified Community Behavioral Health Clinics, and $96M for the new STREETS Program. STREETS will award eight communities up to $3M a year for four years to develop care systems for people who are homeless with substance use disorders or serious mental illness. (link)

⏺️ FamilyWell Health launched a Menopause Hub with clinical guidelines and a new certification. The Boston-based women's mental health company released a free clinical white paper for providers and opened enrollment for a Menopause Behavioral Health Certification for coaches and clinicians. If you are interested in FamilyWell, you can learn more about them in a story I wrote on them earlier this year. (link)

⏺️ Menthra launched their emotional well-being platform in India with 30,000 beta users and 500 practitioners. The Texas-headquartered emotional well-being platform launched offers a continuum of AI companions, coaches, and clinical care and last week announced their launch in India. (link)

⏺️ Neumora discontinued its lead depression drug after it failed to beat placebo in three trials. Navacaprant, a drug designed to treat depression through a different biological mechanism than standard antidepressants, failed to produce meaningful improvements over placebo in two Phase 3 trials. In the most recent trial, patients on the drug saw depression scores fall by 10.1 points compared with 10.8 points for those on the placebo. The company is discontinuing the drug and plans to cut roughly 35% of staff. (link)

⏺️ Ketim Technologies won A$250,000 for a postpartum depression diagnostic test. The Brisbane-based company received the grant from Australia's Industry Growth Program Early Commercialisation Fund to support prototype assay development toward a CE-IVD approved diagnostic device. (link)

⏺️ Sword Health published an interesting blog post on MindGuard, its open-source safety classifiers for mental health chatbots. The post provides transparency into how MindGuard works to make mental health AI safer and highlights some of the limitations, e.g., that it is a turn-level classifier and does not evaluate risk across longer-term use. (link)

⏺️ Wellcome set out its priorities as it approaches its 90th anniversary. The foundation, one of the world's largest funders of mental health research, will invest more than £16 billion in science between 2022 and 2032, with mental health as one of three global focus areas alongside infectious disease and climate change. CEO John-Arne Røttingen said at least 25% of active research investments will be in Africa, South Asia, and South-East Asia by 2030, reflecting a push to direct funding toward the geographies where mental health need is most acute and scientific capacity most underdeveloped. (link)

Hey, I wanted to give you a quick reminder of all the resources and benefits available to you as a Hemingway Pro member. We’ve got lots happening in the next month, so make sure to register for the events and join the discussions in the Slack group.

⏺️ Slingshot study suggests AI chat text can predict depression severity with reasonable accuracy. Slingshot tested whether PHQ-9 scores could be estimated from conversation text alone. The model predicted depression severity within 2.6 points of a person's true score (r = 0.80) and achieved above 0.87 AUC at every severity level, though accuracy fell for people with few symptoms. The ability to make such predictions may prove very helpful in both screening and monitoring. (link)

⏺️ RCT from Flourish Science suggests that twice-weekly usage of AI app may reduce depression and anxiety in university students. Flourish Science tested its app against care as usual in 1,137 university students over six weeks. Students using the app twice weekly showed significantly greater improvement in depression and anxiety, with a number needed to treat of 4.9 for depression and 8.0 for anxiety. The app appeared to help students with elevated symptoms improve while preventing worsening in those with lower baseline symptoms. Results were presented at Stanford's AIMI Symposium. (link)

⏺️ An AI symptom checker integrated into a government maternal health app was found to improve care-seeking in South Africa. Researchers evaluated Ada Health's diagnostic decision support system embedded in MomConnect, South Africa's WhatsApp-based maternal health platform, across 968 pregnant women and mothers. After using the tool, 44.6% of users changed their care-seeking intentions, and care-seeking rose from 17% to 43% at one-week follow-up. The study was observational with no control group and relied largely on self-reported data. (link)

⏺️ Co-design study finds young people want transparency and human oversight in AI mental health tools. Poulsen, Hickie, and colleagues at the University of Sydney ran online workshops with 32 young people to explore their perceptions of Mia, a generative AI chatbot originally designed for professionals in Australian youth services. Four themes emerged: young people wanted AI to feel human without replacing human care, demanded clarity on how the system works, questioned whether AI was appropriate for all mental health contexts, and wanted personalisation within safe boundaries. (link)

⏺️ Study finds AI chatbots may foster cognitive dependence by defaulting to advice over reflection. The researchers were concerned about Cognitive Atrophy - a phrase they define as “whether LLM interactions help users keep reflecting, coping, and making decisions themselves”. To measure this, they developed COGNITIVE ATROPHY BENCH. They found that over 42,230 responses from 1,576 real counselling conversations, all major AI models defaulted to directive advice and reassurance rather than open questions that prompt reflection. The problem worsened as conversations went on, leading the authors to argue that current chatbot design may be doing the user's thinking for them. (link)

⏺️ Large primary care study found ADHD diagnosis rates in England far below global prevalence estimates. John, Stewart, and colleagues analysed primary care records from 42.2 million people and found recorded ADHD prevalence of 1.19% in mid-2025. When benchmarked against global population prevalence estimates, the gap between recorded and expected diagnosis rates ranged from around 44% in adults aged 18-24 to 93.5% in those aged 65 and over. (link)

⏺️ Global Burden of Disease analysis estimated 46.9 million youth ADHD cases worldwide. Researchers analysed global epidemiological data and reported ADHD prevalence of 1.78% in under-18s, with male prevalence (2.52%) more than double female and a peak of 5.62% in Australia. (link)

⏺️ RCT found pramipexole may ease anhedonia where standard antidepressants fail. Lund University researchers randomised 85 patients to pramipexole or placebo in a trial published in Nature Medicine. The generic dopamine agonist, already approved for Parkinson's disease, eased anhedonia with benefits holding through a six-month extension. (link)

⏺️ Retrospective study suggests that esketamine may produce faster response than rTMS in treatment-resistant depression. Researchers analysed outcomes for 372 patients across both treatments. Esketamine produced a response in a median 36 days versus 49 days for rTMS, though long-term effectiveness converged between the two treatments. (link)

⏺️ Open-label study found Magnus Medical's SAINT protocol improved most treatment-resistant patients within five days. Researchers enrolled 101 patients across seven private clinics in a study published in the Journal of Mood and Anxiety Disorders. MADRS depression scores fell 41.3%, and 68.3% of patients were classified as improved by day five, with no discontinuations for adverse events. The study was open-label and uncontrolled. (link)

That’s all for this week.

Keep fighting the good fight!

Steve Duke

Founder of Hemingway

P.S. If you find these updates valuable, consider becoming a Hemingway Pro Member. As a Pro Member, you will get access to even deeper industry insights and will be invited to join our community of over 450 mental health leaders. If you join today, you will get 50% off your first 3 months.

P.P.S. Can you guess the graph? Reply and let me know what data you think the graph below represents. Winner gets a Hemingway prize!