AI‑startup funding map
A study of 3,807 AI healthcare startups from 2010–2024 found heavy investor interest in clinical decision support and diagnostics, while acute‑care areas such as mental health saw gaps in investment. The analysis highlighted concentration of funding around diagnostics/CDS rather than a uniform distribution across clinical domains (x.com).
Artificial intelligence healthcare startups did not attract money evenly from 2010 to 2024; investors clustered around diagnostics, drug discovery, and clinical decision support. (nature.com) The study, published April 14, 2026 in *npj Digital Medicine*, analyzed 3,807 startups founded between 2010 and 2024. The authors were Ahmed Zahlan of University College London, Pek Hooi Soh of Simon Fraser University, and Bart Clarysse of ETH Zurich. (nature.com) Clinical decision support means software that helps doctors make choices, like a navigation app suggesting a route from patient data. The paper found nearly two-thirds of AI investment went to clinical decision support, drug discovery, and diagnostics. (nature.com) Diagnostics and clinical decision support often fit deep-learning systems that can scale across scans, records, and lab results. The same paper said mental health, public health, and rehabilitation drew less venture capital, which the authors linked to data and scaling limits rather than lower clinical need. (pubmed.ncbi.nlm.nih.gov) The researchers sorted companies with a five-tier framework for AI system complexity, then compared medical domain, funding, geography, and team makeup. They also posted sample data and code in a public GitHub repository tied to the paper. (nature.com) (github.com) The map of funding also skewed by place and by who founded the companies. The paper said startups were concentrated in high-income countries and founding teams were mostly technical and business-oriented, with limited clinical representation and gender diversity. (nature.com) That pattern sits against a mental-health field where AI tools are already being studied for diagnosis, monitoring, and treatment support. A 2025 systematic review identified 85 studies and found AI applications across all three areas, with common uses including risk prediction, symptom tracking, and chatbots. (pmc.ncbi.nlm.nih.gov) The funding map does not say underfunded areas lack research activity; it says venture capital has favored business models that look easier to scale. The paper’s authors said the dataset offers a base for a more equitable and evidence-driven digital-medicine ecosystem. (nature.com) The result is a startup market that put the most money where data is abundant and products can spread fastest. In healthcare AI, the busiest investment lanes were not the same as the areas of greatest unmet need. (nature.com)
