Researcher profile

Castro-de-Araujo, LFS

I am a psychiatrist and epidemiologist trained in Brazil and Australia who has spent more than a decade developing causal inference methods for complex health research questions. My most significant methodological contribution is MR-DoC2, a structural equation model that integrates Mendelian Randomization (MR) with twin data to allow bidirectional causal inference while relaxing the horizontal pleiotropy assumption that constrains standard MR. Published in Behavior Genetics (2023), MR-DoC2 enables the detection of causal feedback loops - a common feature of psychiatric and developmental systems that is unidentifiable with classical MR. I subsequently examined the consequences of measurement error for power and bias in family-based MR estimators, providing practical guidance for researchers designing or interpreting such studies (Behavior Genetics, 2025). Extending causal inference into longitudinal designs, I contributed to the development of instrumental variable estimators within cross-lagged panel models (IV-CLPM; Multivariate Behavioral Research, 2024), and have further extended this framework to the random-intercept specification (RI-CLPM+IV), currently under review, which separates within-person causal effects from stable between-person confounders over time. I am also an active developer of umx, a widely used R package for structural equation modeling in behavioral genetics. My applied research uses these and related methods across psychiatric epidemiology, neurodevelopment, and public health. Key recent contributions include: causal analyses of associations between brain morphometry and suicidal behavior in childhood and adulthood, using ABCD Study data; investigation of the bidirectional relationship between alcohol use and suicidal thoughts and behaviors (Alcohol, Clinical and Experimental Research, 2025); a causal analysis of bullying perpetration and depressive symptoms using twin data from the TwinLife study (Clinical Psychology & Psychotherapy, 2025); I also contributed to methodological work on longitudinal analysis in the ABCD Study (Developmental Cognitive Neuroscience, 2025) and to the Global Burden of Disease study. My Google Scholar h-index is 16 and my iCite weighted Relative Citation Ratio stands at 461, reflecting both productivity and strong alignment with NIH-funded research priorities. Throughout my career I have prioritized leading projects and attracting external funding. I served as co-investigator and project manager on an MRC UK grant (MR/T03355X/1; £200,000; 2019–2021) and currently serve as co-investigator on an active NIMH R01 (MH128911; $1,994,000; 2021–2026). At VCU I have contributed to graduate-level instruction in Mendelian Randomization, ordered regression, and interaction modeling, and received the Lindon J. Eaves Award for Excellence in Post-doctoral Research in 2023. My ongoing research program focuses on expanding the toolkit of longitudinal causal inference and applying these methods to neurodevelopmental, psychiatric, and aging outcomes in large observational and administrative datasets.