Abstract
Global mental health needs are escalating, yet few people with mental disorders receive effective care, underscoring the need for robust real-world evidence (RWE) to guide system transformation. Real-world data (RWD)-such as electronic records, claims, patient-reported outcomes, and digital sources-can capture the complexity of mental health care delivery beyond trials but remain underused. In this mini-review, we discuss methodological and infrastructural priorities for leveraging RWD to improve mental health services research and care. We describe recent peer-reviewed studies that have used RWD to examine complex mental health care and associated outcomes, focusing on applications of artificial intelligence and machine learning (AI/ML) and on approaches that enhance validity and reproducibility. Many recent studies report the use of AI/ML to identify study populations, extract unstructured clinical information, or predict treatment patterns, while others report the use of RWD to characterize trajectories, service use, and costs. Building on these examples, we propose two urgent actions: (1) adopt relevant, reliable, and routine RWD curation, transformation, and analysis strategies-including target trial emulation for causal inference-and (2) strengthen mental health care data systems through standardization, harmonization, and interoperability. To promote transparency, we highlight protocol and reporting tools (e.g., HARPER, ATRAcTR, TARGET) and recommend registration of RWE studies. Collectively, these advances can enable high-quality, patient-centered RWE that better reflects real-world mental health care and informs more equitable, effective services.