Integration of Clinical and Molecular Data in Precision Risk Stratification: Vascular Instability and Reproductive Health in the Context of COVID-19

Abstract

The greater availability and accessibility of molecular methods and the digitalization of healthcare data have enabled obtaining large amounts of clinical, molecular, and population-level data. This, as a result, transforms healthcare worldwide. Conventional “one-size-fits-all” approaches are restricted in addressing the increasing burden of complex diseases. Thus, the overall aim of this thesis was to develop and evaluate integrated clinical–molecular models for risk stratification of adverse outcomes related to SARS-CoV-2 infection and brain arteriovenous malformations rupture risk. The research consists of four interconnected empirical studies conducted within a precision healthcare framework. The studies focus on (1) host genetic determinants of asymptomatic SARS-CoV-2 infection; (2) integrated risk prediction models for adverse maternal and perinatal outcomes among pregnant women with COVID-19; (3) prediction of COVID-19-associated pneumonia in pregnancy using combined clinical, biochemical, and genetic markers; and (4) integration of clinical and inflammatory genetic determinants for predicting rupture of brain arteriovenous malformations. Across studies, genetic variants were selected based on biological plausibility within key pathways involved in immune regulation, inflammation, vitamin D metabolism, and the renin–angiotensin system. Multivariable logistic regression models were developed and evaluated using measures of discrimination and explanatory capacity, including area under the receiver operating characteristic curve, pseudo-R², log-likelihood, sensitivity and specificity balance of models (Youden index). This thesis identified that although the individual studies address distinct clinical conditions, they share common biological and methodological patterns. The first if the inflammation as a central factor and its role in vascular and perinatal complications. The second is modulation of risk patterns by host genetic variations. Lastly, improvements in explanatory capacity of traditional clinical, sociodemographic factors-based risk prediction models upon inclusion of genetic factors. Overall, this thesis demonstrates that integrating genetic markers with routinely collected clinical data can improve risk stratification for diverse inflammatory and vascular conditions. The findings support a translational precision healthcare approach that combines molecular and clinical information to better understand disease heterogeneity and improve population-specific risk assessment strategies. These results are particularly relevant for healthcare systems in middle-income countries, where accessible genotyping technologies combined with clinical data may provide feasible tools for advancing precision medicine in routine practice.