Predictive AI Modeling of Biopsychosocial Determinants of Coagulation Risks Post-AstraZeneca Vaccination: A Retrospective Psychosomatic Cohort Study in Tripoli, Libya

Abstract

The ChAdOx1 nCoV-19 (AstraZeneca) vaccine has been globally linked to rare, severe thromboembolic complications like Vaccine-Induced Immune Thrombotic Thrombocytopenia (VITT). While physiological mechanisms are heavily investigated, a substantial research gap exists regarding how post-vaccination psychological distress (e.g., vaccine-related anxiety) interacts with hematological parameters in North African populations. This study bridges this gap using Machine Learning (ML) to process complex biopsychosocial datasets. Methods: A retrospective cohort study was conducted on 100 vaccinated individuals in Tripoli, Libya. Longitudinal laboratory data (D-Dimer, White Blood Cell count [WBC], and Platelet count [PLT]) were paired with psychometric scores evaluating post-vaccination health anxiety. Artificial Intelligence (AI) architectures, specifically Random Forest classifiers, were deployed to discover non-linear correlations and predict sub-clinical clotting risks. Results: Elevated D-Dimer levels (> 500 ng/mL) were identified in 6% (n=6/100) of the global cohort, showing an incidence rate of 11% among women. An explicit age-dependent escalation in D-Dimer concentration was recorded among the affected female patients. Crucially, all six hypercoagulable individuals presented with completely normal WBC and PLT counts, demonstrating a clear clinical dissociation from typical acute thrombocytopenia. Conclusion: This investigation establishes that AstraZeneca vaccination is associated with a sub-clinical hypercoagulable risk within a 6% subset of the cohort in Tripoli, Libya, driven by clear gender and age factors. Standard complete blood counts are insufficient on their own to rule out localized clotting risks, making quantitative D-Dimer testing a vital independent diagnostic tool.