Hi friend,

Last week saw a major boost for the metabolic psychiatry movement with the establishment of MetPsy and a $20M investment from The Baszucki Group. AI companion company Tolan raised $30M, and Talkiatry announced the expansion of its therapy services into all 50 states.

In the UK, the government announced an interesting £9M pilot to integrate mental health and social care.

We also saw some interesting new research on use cases for AI (outside of conversational agents). This came in the form of personalised rTMS treatments and improved detection of early psychosis risk. Finally, there is fascinating research on the relationship between the legalisation of sports gambling and rates of mental health treatment across US states, as well as insights on how young men engage with masculinity content.

Let’s get into it.

⏺️ The Baszucki Group announced the launch of the Global Network for the Advancement of Metabolic Psychiatry, alongside $20M of investment. Known as MetPsy, the initiative aims to advance research, clinical care and education focused on the relationship between metabolism and mental illness. The $20M investment will establish a new centre at McLean Hospital to develop and evaluate metabolic treatments for psychiatric conditions, including ketogenic diets. Early studies have suggested potential benefits for conditions such as bipolar disorder and schizophrenia however, larger controlled trials are needed to establish their effectiveness. (link)

⏺️ Thrive Therapies raised $7.7M to expand school-based healthcare. The US company provides speech, occupational and physical therapy, alongside psychological assessments and behavioural-health services for children with disabilities. It currently operates across more than 200 schools and districts in six states. The funding will support expansion into additional schools, recruitment of clinicians and development of technology to manage service delivery and Medicaid billing. (link)

⏺️ AI companion Tolan raised $30M. Let me be clear: this is NOT a mental health company. But I did think the news was relevant given the importance of AI and companions in the mental health discourse. Last month, I reported that the AI companion company had surpassed $1M in monthly revenue from over 100,000 paying users. (link)

⏺️ PsychPlus acquired California psychiatric provider, Precision Psychiatric Services. PsychPlus, a US provider of in-person and digital mental-healthcare, acquired the California-based psychiatric services company, which supplies clinicians to hospitals, emergency departments, outpatient clinics and addiction-treatment facilities. The deal expands PsychPlus's presence in California and follows its earlier acquisition of digital mental-health company Koa Health. Financial terms were not disclosed. (link)

⏺️ US prosecutors charged addiction-treatment operators over alleged $1.6B fraud. The Department of Justice charged a father and son who operated more than 20 addiction treatment facilities in Southern California. Prosecutors allege that the network submitted $1.62B in false insurance claims between 2020 and 2026, receiving more than $210M in payments. The alleged scheme involved paying recruiters to bring insured patients into treatment, billing for services not delivered and, in some cases, preventing patients from leaving. (link)

⏺️ Talkiatry announced the expansion of its online therapy services to all 50 US states. The US virtual mental-health provider now offers psychotherapy alongside psychiatric care and medication management nationwide. According to the company, it now employs more than 800 psychiatrists and 400 therapists, works with over 100 health insurers and has delivered more than four million patient visits. The expansion follows a $210M Series D funding round earlier this year.(link)

⏺️ Headway paused new clinician contracts following objections from therapists. Reporting from BHB noted that the US mental-health platform circulated agreements containing broad rights to use providers' names, images and voices, restrictions on soliciting patients referred through Headway, and requirements concerning clinical documentation and billing. Following complaints from therapists, CEO Andrew Adams acknowledged that provisions had been interpreted in ways the company had not intended and paused the rollout the following day. (link)

⏺️ Study exposed widespread ghost networks in US mental-health insurance directories. Researchers made 5,822 calls to psychiatrists and psychologists listed in commercial insurance directories across five states. Just 12.4% of calls resulted in an appointment. Among calls with a definitive outcome, 56% encountered inaccurate directory information, while 22% reached providers unable to accept new patients. The findings highlight how insurers' published provider networks can substantially overstate patients' access to mental healthcare. (link)

⏺️ UK government announced £9.4M pilot integrating mental health and social care. The Care Together programme will place NHS mental-health professionals within children's social care teams in Stockport, Tameside and Rochdale. It aims to support up to 7,000 children and young people, alongside their families and carers. The three-year pilot is expected to begin in early 2027, with an evaluation to determine whether the model should be expanded. (link)

⏺️ MHRA launched initiative to help digital mental-health developers meet regulatory requirements. The UK's medicines and medical devices regulator announced a community of practice and a survey aimed at developers of digital mental-health technologies. The initiative will help companies navigate existing medical-device rules and identify areas where further regulatory guidance may be needed. (link)

⏺️ Australia reviewed whether mental-health apps should face stricter regulation. Under Australia's current rules, certain digital mental-health tools are excluded from medical-device regulation. The Therapeutic Goods Administration (TGA) reviewed whether these exemptions remain appropriate, particularly for apps that assess mental-health symptoms or provide therapeutic interventions. The review found that existing rules could allow higher-risk software, including some AI-enabled products, to fall outside medical-device oversight if they reference accepted clinical guidelines. This raised concerns that the current criteria do not adequately account for a product's intended use, outputs or potential consequences of errors. The regulator plans to consult on possible changes. No new requirements have yet been introduced. (link)

⏺️ International researchers proposed a public-health framework for brain health. A group of 26 researchers led by Timothy Daly at the University of Bordeaux proposed extending brain-health policy beyond individual diagnosis and treatment to population-wide prevention. Building on existing WHO initiatives and an international public-health charter, the framework calls for governments, healthcare systems and communities to address factors such as poverty, education, environmental risks and unequal access to care. The authors want brain health to become a broader public-health priority, with governments addressing factors such as poverty, education and environmental risks alongside clinical care. The Brain Economy movement continues to gain traction, and I anticipate this will continue. (link)

⏺️ Chatbots differed substantially in their responses to simulated psychosis. Researchers tested ChatGPT, DeepSeek and Replika over seven days of simulated conversations in which psychotic symptoms became progressively more severe. ChatGPT became less likely to affirm delusional beliefs, while DeepSeek's responses remained relatively stable and Replika became more affirming. DeepSeek and Replika gave potentially harmful advice, including suggestions to stop antipsychotic medication, in 21% and 37% of conversations, respectively. The findings concern simulated interactions rather than outcomes among people experiencing psychosis. (link)

⏺️ AI algorithm identified potentially better brain stimulation protocols for depression. Researchers developed an AI system to help personalise repetitive transcranial magnetic stimulation (rTMS). During rTMS, clinicians can adjust the type, strength and location of magnetic stimulation, but there is limited evidence on which settings work best for individual patients. Using treatment records from 1,609 patients across 19 US clinics, researchers trained an AI model to recommend settings for each session. When clinicians' actual choices matched those recommendations, patients experienced slightly greater improvements in depression symptoms by their next session. Most interestingly, the researchers estimated that following the recommendations throughout treatment could produce an additional 3.2-point improvement on a 27-point depression scale. This was a retrospective analysis, but interesting findings all the same. (link)

⏺️ AI system improved detection of early psychosis risk from everyday speech. Researchers in Singapore analysed 45-second speech samples from 429 people aged 14–29. Standard language models achieved 55% accuracy when asked to identify those at high risk of psychosis. A system combining AI analysis of emotional cues with machine learning improved accuracy to 67% and identified up to 93% of at-risk participants. The findings suggest that ordinary conversation may contain useful warning signs, although the approach would require further clinical validation before use in routine screening. (link)

⏺️ Four-day intensive therapy significantly outperformed self-guided online CBT for OCD. A German randomised trial assigned 119 adults with obsessive-compulsive disorder to either concentrated exposure therapy or an eight-week self-guided online programme. After three months, 61% of assessed participants receiving intensive treatment had responded, compared with 25% in the online group. Remission rates were 42% and 14%, respectively. (link)

⏺️ Insomnia therapy reduced suicidal thoughts across 19 randomised trials. A meta-analysis involving 4,974 adults found that cognitive behavioural therapy for insomnia reduced the severity of suicidal thoughts by 23% compared with control conditions immediately after treatment. Evidence of lasting benefits was less certain. The findings support treating insomnia as a potential addition to suicide-prevention care. (link)

⏺️ Long-acting aripiprazole was linked to fewer hospitalisations in bipolar disorder. Researchers compared 1,814 patients who switched to injectable aripiprazole with an equal number who continued taking the oral medication. Over an average of 1.6 years, those receiving injections had a 23% lower rate of hospitalisation and a 12% lower rate of emergency-department visits. However, they had higher rates of obesity and movement-related side effects. The study used observational health records rather than random assignment, so differences between patients may partly explain the results. (link)

⏺️ Genetic study linked serious mental illness to markers of biological ageing. A genetic study examined whether major depression, schizophrenia and bipolar disorder share biological characteristics associated with ageing. Researchers found genetic overlap between all three conditions and markers of biological ageing, suggesting that some genetic factors associated with psychiatric illness may also be linked to how the body ages. Major depression was additionally associated with shorter telomeres and greater epigenetic age. (link)

⏺️ Researchers identified a possible depression subtype linked to disrupted body clocks. Using data from the Australian Genetics of Depression Study, researchers identified a group of people whose depression was associated with disturbances in their circadian rhythms. Around 23% of participants fell into this proposed subtype, which was associated with earlier depression onset, more severe symptoms and poorer responses to antidepressants. The findings suggest that disrupted body clocks may help explain why depression differs between individuals and could eventually inform more personalised treatments. (link)

⏺️ US behavioural-health facilities showed substantial gaps in digital infrastructure. Researchers analysed a 2024 national survey of 27,957 mental-health and addiction-treatment facilities. Around half used electronic health records exclusively, but only 13.2% actively participated in organisations that allow providers to exchange patient information. Nearly one in five facilities did not know whether such information-sharing networks were available. Digital capabilities also varied considerably by facility type and state, suggesting that adopting electronic records alone has not ensured that behavioural-health services can share information effectively. (link)

⏺️ Online sports betting legalisation was linked to increased mental-health treatment. Researchers at Johns Hopkins University analysed US insurance records covering nearly 62 million person-years between 2016 and 2023, comparing changes in states that legalised online sports betting with those that had not. Legalisation was associated with increases in antidepressant and anti-anxiety prescriptions, alongside outpatient diagnoses of depression and anxiety. Among men, outpatient depression diagnoses rose by an additional 0.34 percentage points and anxiety diagnoses by 0.67 percentage points relative to comparison states. The study found no significant changes in substance-use disorder treatment or psychiatric hospital admissions.(link)

⏺️ Young men reported both benefits and concerns about masculinity content on TikTok. Researchers analysed responses from 142 men aged 16–25 in Australia, Britain and the US. Participants described finding motivation and self-improvement advice in masculinity-related content while also recognising harmful stereotypes and more extreme messages about gender and relationships. The findings suggest that young men do not simply accept or reject this content, but actively weigh its perceived benefits and risks. (link)

That’s all for this week.

Keep fighting the good fight!

Steve Duke

Founder of Hemingway