Talkspace gets acquired, Grow raises $150M and LOTS of fascinating research insights...

Hi friend,

Wow. What a week in mental health…

Let’s get straight into it…

⏺️ UHS is buying Talkspace. In a huge move for the space, the US healthcare giant will acquire Talkspace for $835M, a 10% premium to their current share price and a >500% increase from the lows of 2023. The deal still requires approval and is expected to close in Q3. I guess Talkspace’s Turnaround is finally complete… (link)

⏺️ Grow raised $150M at a $3B valuation. Grow shared that they have crossed $1B in annual revenue and will use the funding to continue their expansion into the employer market, expand primary care partnership and double down on their AI initiatives. The investment is another signal of the renewed investor appetite for insurance-based outpatient mental health infrastructure which we are seeing across public and private markets in Q1. (link)

⏺️ Ease Health raised $41M from a16z. The company just came out of stealth and raised this significant Serise A. Ease are buliding an AI-native behavioural health operating system (combining all the acronyms of EHR, RCM, CRM etc. into one solution). It’s good to see there is still interest in mental health solutions from top-tier, generalist VCs like a16z. (link)

⏺️ Brightn closed a $1M+ pre-seed. The AI-powered preventive wellness app closed a round in February but we didn’t pick it up originally. They also just announced a partnership with Vive18 to add Brightn as an Ai-mental health layer to Vive18’s school programs. (link)

⏺️ SAMHSA offered $69.1M in behavioural health grants. The package includes $43M specifically for children's mental health services. SAMHSA funding has been highly unstable this year - as Michael Scott would say, “SNIP-SNAP-SNIP-SNAP”. (link)

⏺️ Children's Mercy raised $150M for paediatric mental health. The Kansas City health system's "Illuminate initiative has already introduced nearly 14 new programmes since 2023 and their fundraising has come from more than 5,000 donors (link)

⏺️ California invested $65M in a new behavioural health campus. The state is converting six vacant Los Angeles facilities into a 162-bed campus, reflecting continued pressure on states to build residential capacity as the nationwide shortage of inpatient psychiatric beds persists. (link)

⏺️ Octave launched WiseMatch to improve insurance directory booking. The platform integrates real-time availability and direct booking into payer directories. Companies like Grow, Rula, and SonderMind are already using it. According to Octave, 40% of patients begin their mental health search inside health plan directories. Provider orgs want to be able to tap into that referral source. (link)

⏺️ The Being Initiative opened a CAD $300K–$1.5M funding call for youth mental health. Grand Challenges Canada is seeking proven interventions ready to scale across twelve countries. The deadline is March 31st, 2026. (link)

⏺️ Headstream Accelerator applications close this week. The program offers $30K in non-dilutive funding and a 16-week programme for digital youth mental wellbeing startups. (link)

⏺️ Headspace detailed its clinical governance model for AI companion Ebb. Chief Clinical Officer Jenna Glover wrote an interesting post on Headspace’s approach to clinical governance of Ebb. It shared details on their Clinical Safety & Experience Council and noted that they have introduced a 30-minute session cap to mitigate AI drift risk. (link)

Writing these round-ups takes a lot of time and effort each week. If you find these updates valuable and would like to support my work, consider becoming a Hemingway Pro member.

There’s lots of great benefits to being a member too. You will join a community of over 370 mental health leaders and get access to exclusive content, events and resources. You can learn more here.

⏺️ 41% of 18–34 year olds face a clinically significant mind health crisis. The Global Mind Health in 2025 Report from Sapien Labs surveyed approximately one million people across 84 countries and identified four primary drivers of poor mind health in young people: early smartphone adoption, ultra-processed food consumption, weakened family bonds, and reduced spirituality. There are many more interesting findings in the report, including that mind health tends to be worse in wealthier countries. This is a very important report that is worth a read. (link)

⏺️ Oxford's PETRUSHKA tool cut antidepressant discontinuation by ~40%. The RCT across 47 sites in the UK, Brazil, and Canada (n=493) found that a 3-minute AI tool matching patients to antidepressants based on clinical profile and side-effect preferences made participants significantly less likely to discontinue at 8 weeks. It also found better depression and anxiety outcomes at 24 weeks versus usual care. This is the first precision psychiatry tool of this kind to demonstrate this result in a large international RCT. (link)

⏺️ Individuals at elevated psychosis risk use generative AI more intensively. A survey of 1,003 young adults found more delusion-related chatbot interactions among those at high psychosis risk, raising important safety questions for AI mental health tools deployed to vulnerable populations. (link)

⏺️ AI diagnostic models systematically underdiagnose female adolescents for anxiety. A Communications Medicine study from UCL and Cincinnati Children's Hospital (~20,000 paediatric anxiety cases) found female adolescents had 4% lower diagnostic accuracy and 9% higher false-negative rates than males. The root cause was documentation bias - clinical notes for male patients averaged 500 words longer with distinct word distributions. A de-biasing framework reduced diagnostic bias by up to 27%, but the finding reminds us that AI clinical tools can inherit the inequities baked into their training data. Happy International Women’s Day everybody… (link)

⏺️ Clinician Telehealth adoption has not increased access for rural or underserved patients. A Harvard/Brown cohort study published in JAMA Network Open (n=17,742 Medicare mental health specialists, 2018–2023) found that clinicians with high telemedicine adoption showed only marginal increases in rural visits (+0.88 percentage points) and near-zero increases in shortage-area visits compared to low adopters. High telemedicine adopters also saw 3.55 percentage points fewer new patients, suggesting Telehealth adoption by clinicians may not be expanding access for underserved populations. This is counterintuitive and the paper offers some explanations and solutions which I won’t go into here. (link)

⏺️ CBT and resilience training improved cancer surgery survival rates. A Chinese RCT (n=280) found that adding CBT, resilience training, and sleep hygiene education to standard surgical care produced 2-year overall survival of 84.3% versus 70.7% in controls. I am super interested in businesses using psyhological interventions to improve other health outcomes. Oncoloy seems to be a area of focus with a couple businesses now doing well in this space. (link)

⏺️ Loneliness mediates the path from anxiety and depression to suicidal ideation. A JAMA Network Open cross-sectional study using the NIH All of Us dataset (n=62,685 US adults) found loneliness accounted for 25% of the anxiety-to-suicidal ideation pathway and 10% of the depression-to-suicidal ideation pathway. The finding positions reducing loneliness as a potentially high-leverage, transdiagnostic intervention point that could interrupt escalation toward suicidal ideation across diagnostic categories. (link)

⏺️ Mixed results for single-session digital interventions. A Nature Human Behaviour study tested 11 single-session digital interventions in 7,505 adults and found that while nearly all improved outcomes immediately, only two produced meaningful depression reductions at 4-week follow-up. Completing an SSI was also associated with reduced confidence and motivation to change at follow-up on average which needs to be understood more if we SSIs are to become a scaled intervention. (link)

⏺️ GLP-1 drugs reduced substance use disorder risk across all substances. A Washington University cohort study published in The BMJ (n=606,434 US veterans with type 2 diabetes) found GLP-1 use associated with a 14% reduced risk of developing any SUD - with reductions of 18% for alcohol, 20% for cocaine and nicotine, and 25% for opioids. Among those with pre-existing SUD, GLP-1 use was associated with a 40% reduction in overdose and a 50% reduction in drug-related deaths. (link)

⏺️ TMS reduced PTSD symptoms in veterans but didn't beat placebo in trials. A meta-analysis of 7 studies (n=963) found large within-group improvements in PTSD symptoms following TMS treatment, but the controlled trials comparing active TMS to sham stimulation did not reach statistical significance. (link)

⏺️ A new framework clarified how digital therapeutics actually work. A npj Digital Medicine Perspective proposed the first systematic framework for DTx mechanisms of action, introducing a critical distinction between intended dose (what the product offers), consumed dose (what the patient uses), and realised dose (the cognitive or behavioural change that actually drives outcomes). The core argument was that engagement metrics like logins and time-in-app are a necessary but not sufficient condition for clinical benefit (unsurprising) and tying them to reimbursement risks misaligning incentives and producing worse products (agreed). Considering the relative underperformance of DTx, suggestions like this are much needed. (link)

⏺️ DTx regulatory fragmentation remains a core commercial obstacle. Staying on the topic of digital therapeutics, a review in Therapeutic Innovation & Regulatory Science mapped the global DTx regulatory landscape across the US, Europe, and Korea. It identified the mismatch between pharmaceutical-level clinical validation requirements and underdeveloped reimbursement infrastructure as the central barrier to international scaling of DTx. The authors call for harmonised guidelines and structured approval pathways. (link)

That’s all for this week.

Keep fighting the good fight!

Steve Duke

Founder of Hemingway