Patient injury resulting from medication use, also known as adverse drug events (ADEs), is the second most frequent complication experienced by hospitalized patients. Aged patients who cumulate multiple chronic diseases and complex treatment regimen are particularly exposed to ADEs. More than one-third of hospitalized older patients experience at least one adverse drug event during their hospital stay and up to 70% of these events are deemed preventable. Among these, antithrombotic-related thromboembolic and hemorrhagic events are particularly frequent and have serious consequences in older inpatients. The increasing use of antithrombotics in this population and their substantial risk of toxicity and ineffectiveness have therefore become a major concern for patient safety and public health worldwide. Numerous approaches have been adopted to improve the quality and safety of drug prescription in this population, but existing strategies are only moderately capable of predicting clinical events or having a significant impact on clinical outcomes. The rise of computational technology has raised much hope for improving the quality of care in recent years.

The purpose of this thesis was to investigate several computational strategies to improve drug safety through two main projects: 1-Therapeutic drug optimization using physiology- and population-based pharmacokinetic modelling and simulation and 2-Automatic detection of ADEs in geriatric management. In both these approaches, we focused on antithrombotic therapies, with a particular emphasis on older hospitalized patients.