The crisis in youth mental health has intensified, especially after the COVID-19 pandemic. Traditional assessment tools like the Highly Sensitive Person – HSP index (1) and the Perceived Stress Scale – PSS (2) provide valuable insights but may overlook the multifaceted nature of mental health. Thus, incorporating molecular biomarkers becomes crucial to unravel the interplay of genetic, environmental, and psychological factors. In this work, we combined neuropsychological and epigenetic data in a machine learning algorithm to predict HSP scores in university students and to provide individualised multivariate fingerprints to assess personal mental health status.In this study, 104 participants were recruited: ten behavioural variables were assessed, including the HSP index, PSS test, Internet Addiction Test, EAT-26, and Barratt Impulsiveness Scale (1-5). Ten molecular variables were obtained from saliva samples and included DNA methylation analysis at DAT1, SERT, and OXTR gene promoters, as these genes have been identified to significantly influence personality traits (6,7). By conducting exhaustive feature selection, a data-driven binary classification model (SVM with polynomial kernel) was trained to provide individual multivariate fingerprints.Despite a limited sample size, the model achieved remarkable accuracy (84.6%), sensitivity (90.4%), and precision (88.0%), paving the way for future clinical applications. The integration of genetic features appears crucial, indicating the importance of balancing neuropsychological and genetic influence for accurate modelling. Future clinical implications include the potential for streamlined data collection using questionnaires and saliva samples, offering cost-effective and accessible avenues for mental health assessment and personalised healthcare approaches.