AI Tool Personalizes Antidepressant Treatment
A new AI tool from the UK's NIHR, called PETRUSHKA, is being used to personalize antidepressant prescriptions. By integrating patient characteristics with deep learning, the platform helps clinicians and patients select the most effective medication with the fewest side effects, marking a practical advance in precision medicine.
The PETRUSHKA AI model, developed at the University of Oxford and backed by the UK's NIHR, was trained on a massive dataset combining real-world data from over a million primary care patients with clinical trial data from roughly 40,000 individuals. Its algorithm uses machine learning to generate a ranked list of the top three antidepressant recommendations for a patient based on their unique characteristics. A 2024 international trial with over 500 adults found that patients using the tool were about 40% less likely to stop their medication within the first eight weeks. For consumer health startups, the path to user acquisition and retention, exemplified by apps like Noom and Headspace, is rooted in a deep focus on behavior change rather than just data tracking. Noom, with over 45 million downloads and $400 million in revenue in 2020, is expanding its model to be an employer-offered healthcare benefit. Headspace has also broadened its reach by partnering with insurers and public health organizations, moving beyond a direct-to-consumer approach. Integrating with wearables is key for personalization and engagement, with Apple Watch, Fitbit, Garmin, Oura Ring, and Whoop being the most commonly used due to their mature APIs and large user bases. While a multi-device integration can take 3-6 months, leveraging unified APIs can slash that development time to just a few weeks. These integrations are pivotal for apps in the longevity and coaching space, which often rely on the detailed sleep and recovery data provided by devices like the Oura Ring and Whoop. Navigating health data privacy is a critical hurdle. Many consumer health apps are not covered by HIPAA, but the FTC's Health Breach Notification Rule can apply to vendors of personal health records. In California, the California Consumer Privacy Act (CCPA) grants residents control over their personal data, though it includes a broad exemption for health information already protected under HIPAA and the Confidentiality of Medical Information Act (CMIA). The fundraising landscape for digital health is robust, with global funding hitting $28.8 billion in 2025, a 9% increase from the previous year. The United States represents 70% of this funding. Investors are showing a strong preference for early-stage startups, especially those in the AI and TechBio sectors. In 2024, AI-focused companies attracted a significant 58% of all digital health venture funding.
