Hi friend,

We saw lots of interesting developments in mental health last week..

FullBloom completed a strategic investment in Daybreak Health. BetterHelp revenues fell again in Q3, despite growth in its insurance business. Kipu Health announced a new set of agentic AI tools for behavioural health organisations, and NRx was granted FDA approval for its preservative-free Ketamine.

New research from Fault Line AI highlighted significant gaps in how AI safety is tested in health care. Data from Iceland found that mobile-phone bans showed no consistent mental-health benefits for teenagers, and a randomised trial found interesting differences in the efficacy of telephone calls and text messages for suicide prevention among adults and teenagers.

Let’s get into it.

⏺️ FullBloom completed a strategic investment in Daybreak Health to expand school-based mental healthcare. FullBloom, a US provider of educational and behavioural services, made an undisclosed investment in Daybreak Health, which provides teletherapy to schoolchildren. The companies plan to combine Daybreak's therapist network with FullBloom's CharacterStrong division, which helps schools teach social and emotional skills and identify pupils who may need additional support. The aim is to make it easier for schools to connect children with therapy when prevention and early support are not enough. (link)

⏺️ Kaiser Permanente invested $15M to expand psychiatric facilities at UCSF Benioff Children’s Hospital in Oakland. The US healthcare organisation committed the funding for a new 20-bed inpatient psychiatric unit. The facility will provide hospital care for children and teenagers with serious mental-health needs and is expected to treat around 700 patients a year when it opens in 2031. (link)

⏺️ New York received $10M to expand mental healthcare in primary-care clinics. Governor Kathy Hochul announced a five-year federal grant to support ACCESS-NY, a programme that will bring mental-health and addiction services into eight primary-care clinics across the state. The programme will use a collaborative-care approach, in which primary-care doctors work with mental-health specialists to identify and treat conditions such as depression and anxiety, rather than requiring patients to seek separate specialist care. It aims to reach approximately 1,750 patients aged six and older, with funding also supporting staff training and technology to coordinate treatment. (link)

⏺️ BetterHelp's revenue fell 13% YoY, despite continued growth in insurance-covered therapy. Teladoc reported that BetterHelp generated $206.6M in third-quarter revenue, down 13% YoY, while average paying users fell 12% to 336,000. Revenue from insurance-covered therapy rose to $32.1M, compared with $3.7M a year earlier. Adjusted EBITDA for BetterHelp increased 17% to $4.5M, a 2.2% Adjusted EBITDA Margin. Teladoc, BetterHelp’s parent company, saw a 3% decline in revenue for the quarter to $605.1M. LifeStance and Acadia will announce their Q3 results this week, with Brainsway announcing theirs on November 10th. (link)

⏺️ FDA approved NRx's preservative-free Ketamine. NRx Pharmaceuticals received US regulatory approval for an injectable generic form of ketamine that does not contain benzethonium chloride, a preservative used in some existing formulations. The approval covers anaesthesia, the original label under which Ketamine was first approved back in the seventies. NRx is separately developing intravenous ketamine for depression and suicidal thoughts. They have also asked the FDA to withdraw approval for competing ketamine products containing benzethonium chloride, arguing that the preservative raises safety concerns. The FDA has not yet ruled on that request. (link)

⏺️ OpenEvidence partnered with the American Society of Addiction Medicine (ASAM) to expand its clinical AI platform. OpenEvidence, a US company whose AI tool helps clinicians find and interpret medical research, partnered with ASAM to include guidance for treating patients with substance use disorders and addiction in its medical research aggregate platform for doctors. The partnership will give clinicians access to specialist addiction-treatment recommendations alongside research evidence, with links to the original sources. (link)

⏺️ Kipu Health launched Agentic AI tools to automate behavioural-health administration. The US healthcare technology company introduced AI agents designed to review patient records, identify missing documentation and prepare requests to insurers for continued treatment. (link)

⏺️ New study found gaps in how AI safety is tested in healthcare. An interesting preprint from researchers at Fault Line AI reviewed a number of scientific studies and identified 139 ways AI systems could fail when used in healthcare. They then examined how these failures overlapped across 14,232 health-related examples, using statistical analysis to group them into 22 broader risk categories. Next, they analysed more than 132,000 real-world and simulated AI conversations to assess how often the different risks appeared and how well existing safety tests covered them. Their main findings were that (a) no single test covered all 22 risk categories and (b) existing tests did not reflect the types and frequency of risks seen in real-world use. The researchers from Fault Line also shared a framework to help make future testing more complete and better reflect real-world use. (link)

⏺️ AI chatbots gave less complete mental-health responses to teenagers with severe symptoms. In a study published in Nature Mental Health, researchers developed a tool to assess whether AI chatbots followed clinical guidance when responding to teenagers experiencing anxiety or depression. They tested four models—GPT-4, GPT-5, Claude 4.5 and Gemini 2.5—using simulated conversations with teenagers whose symptoms ranged from moderate to severe. Responses were assessed against clinical criteria across seven areas of mental-health support. The chatbots covered more of the relevant criteria when responding to anxiety than depression, and to moderate symptoms than severe ones. The findings suggest that AI responses may be less complete in more serious mental-health situations. However, the study used simulated conversations and did not assess whether chatbot responses helped or harmed real teenagers. (link)

⏺️ Teenagers using AI for emotional support reported more mental-health difficulties. The Jed Foundation surveyed 5,560 students across 21 US middle and high schools. 57% reported using AI when they felt sad, stressed, lonely or overwhelmed. Those using AI for companionship or emotional support reported more symptoms of anxiety, depression and loneliness, and were more likely to report a history of suicidal thoughts. Of course, the study cannot determine whether AI use contributed to these difficulties or whether teenagers already struggling with their mental health were more likely to seek support from chatbots. (link)

⏺️ One in three young US adults found to use AI for information about drugs and alcohol. Rutgers researchers surveyed nearly 1,000 adults aged 18–29 and found that 32.3% had used generative AI tools to ask questions about substances or substance use. Among those users, 92.8% had sought information about potential risks and consequences, while 34.9% had asked about substance-use disorders. Almost half used AI for substance-related questions at least once a month. Users generally viewed the tools as trustworthy, reliable and helpful. (link)

⏺️ Brief digital mental-health lessons reduced depression and anxiety symptoms in teenagers. In a randomised trial involving 1,080 ninth-grade students in a US school district, researchers compared an 80-minute digital mental-health education programme with another digital programme of similar length. Ten weeks later, students who received the mental-health lessons reported significantly fewer symptoms of depression and anxiety. The researchers also found evidence that greater willingness to seek help partly explained the reduction in depression symptoms. (link)

⏺️ School mobile-phone bans showed no consistent mental-health benefits for teenagers. Researchers analysed nearly 24,000 survey observations from Icelandic teenagers between 2016 and 2023, comparing changes in schools that introduced phone restrictions with those that did not. They examined depression symptoms, well-being, screen use and bullying before and after the policies were introduced. The results showed no consistent improvements in mental health following the bans. However, schools were not randomly assigned to introduce restrictions, so the study cannot establish whether the policies themselves caused the observed outcomes. The findings also do not rule out benefits from different types of phone restrictions. (link)

⏺️ Psychedelic nasal spray found to reduce depression symptoms in a phase 2 trial. Researchers tested mebufotenin, also known as 5-MeO-DMT, in 193 adults with treatment-resistant depression. Participants received different doses of the drug alongside psychological support. After 29 days, depression scores fell by an average of 11.1 points in the 12mg group, compared with 5.5 points in the low-dose comparison group, a statistically significant difference. Some 34% of participants receiving 12mg showed a clinically meaningful improvement, compared with 12% in the comparison group. (link)

⏺️ Long-term oral esketamine treatment research showed that maintenance home treatment may be associated with sustained symptom stability in patients with TRD. In an observational study, researchers examined treatment-resistant depression patients who first received oral esketamine in a clinic over 6 weeks, before moving maintenance treatment at home. Over an average of 20.8 months, average depression scores remained broadly stable. 23% of patients who had initially responded to treatment maintained that response throughout the period. 12% successfully tapered off esketamine maintenance treatment and around 13% stopped treatment partly because of side effects. As the study had no comparison group, it cannot establish whether continued esketamine treatment prevented symptoms from returning. (link)

⏺️ Global eating-disorder cases rose by 71% since 1990 and are expected to keep increasing through 2050. Researchers estimated that 18.1 million people worldwide were living with anorexia nervosa or bulimia nervosa in 2023, including 5.5 million males. Although total cases increased by 71% since 1990, the rise was around 21% after adjusting for population growth and ageing. Around 41% of those affected were aged 10–24. Rates were highest in wealthy regions, but increased fastest in parts of Asia. Researchers forecast that the total number of cases will continue to rise through 2050. (link)

⏺️ Telephone calls and text messages showed different benefits for suicide prevention among adults and teenagers. In a randomised trial involving 1,520 people at risk of suicide in US emergency departments and primary-care clinics, researchers compared supportive telephone calls with personalised, two-way text messages. Among adults, telephone calls were more effective at reducing suicidal thoughts and behaviours after six months. Among teenagers, text messages showed better results after 12 months, including fewer reports of suicidal behaviour during the final six months. (link)

⏺️ Large brain-imaging study challenged assumptions about emotional control. An international research team analysed 40 brain-imaging datasets involving 2,175 people. They examined cognitive reappraisal, a strategy for managing emotions by changing how we think about a situation. The researchers found no consistent pattern of brain activity that could reliably identify people who were better than others at using this strategy. The findings challenge the idea that activity in certain brain regions can provide a simple measure of someone's ability to regulate their emotions. (link)

That’s all for this week.

Keep fighting the good fight!

Steve Duke

Founder of Hemingway