RESEARCHERS DEVELOP AI-DRIVEN PLATFORM FOR PRECISION THROMBOSIS RISK STRATIFICATION Scientists at the Sant Pau Research Institute and the Biomedical Research Networking Center for Rare Diseases have engineered an artificial intelligence system capable of mapping complex biological profiles associated with thrombosis risk. Recently published in the Journal of Thrombosis and Haemostasis, the platform addresses a persistent clinical gap: predicting venous thromboembolism in patients who lack obvious clinical triggers or hereditary markers. The research team leveraged data from the GAIT2 family cohort, which includes 790 individuals with familial clotting disorders, including 70 patients experiencing idiopathic venous thrombosis. By deploying machine-learning algorithms, the platform fused standard clinical metrics, genomic variants, and transcriptomic expression data spanning nearly thirteen thousand genes. This multi-omic integration moves beyond legacy risk assessment models that rely heavily on age, obesity, hormone therapy, and isolated genetic mutations. The AI architecture successfully isolated 494 genes tightly correlated with thrombotic events, highlighting long noncoding RNAs as previously overlooked regulatory biomarkers. The team subsequently constructed a composite molecular signature that generates a similarity score, quantifying how closely an individual's biological profile mirrors historical thrombosis cases. Transcriptomic integration proved decisive for accuracy. Initial models relying solely on clinical and genetic data misclassified 43 percent of healthy participants as high-risk. Adding gene expression data cut that false-positive rate to 23 percent while raising the true-positive detection rate from 70 to 74 percent. Dr. Jose Manuel Soria, director of the Complex Disease Genomics Unit at IR Sant Pau, stated that synthesizing thousands of biological variables yields a substantially more precise risk stratification than evaluating traditional factors in isolation. The computational model also identified molecular pathways tied to cardiomyopathy and renal proximal tubule activity, reinforcing the physiological relevance of the findings and expanding the known biological network governing clotting disorders. First author Dr. Pol Ezquerra emphasized that coupling artificial intelligence with transcriptomics reveals disease-associated patterns invisible to conventional diagnostic workflows. The resulting signature provides a scalable framework for refining patient categorization and engineering targeted preventive interventions. Although the platform marks a significant advancement in computational medicine, the investigators note that independent cohort validation is mandatory before clinical adoption. Future development will focus on standardizing the algorithm for routine healthcare deployment, potentially shifting thrombosis management toward data-driven personalization. By decoding the dynamic interplay between genetic predisposition and active gene expression, the system establishes a new benchmark for precision cardiovascular risk prediction.