With the increasing prevalence of type 1 and type 2 diabetes, management,
particularly for those receiving insulin therapy, becomes a global health concern. Since
the advancement of in silico models, diabetes treatment has greatly improved.
Computer simulations forecast the insulin dosages needed for bolus and basal infusion
based on real-time glucose dynamics. By replicating glucose-insulin dynamics and
taking into account factors like age, activity, food intake, insulin resistance, and more,
these models—like the UVA/PADOVA simulator—have completely changed the way
people with diabetes are treated. These algorithms not only forecast insulin dosages but
also help with medication customization, thereby improving patient outcomes. When
factors like stress and physical activity are taken into account, in silico models offer
vital information on the ideal basal and bolus insulin dosages. Additionally, even in the
absence of exact insulin-to-carbohydrate ratios, reinforcement learning models have
demonstrated potential in predicting bolus insulin doses, increasing the accuracy of
insulin delivery. Despite improvements, data validation, device integration, and the
requirement for individualized care continue to pose challenges to these models' ability
to accurately forecast insulin dosages. Several in silico modeling techniques, their uses,
and the significance of tailored care in maximizing diabetes management are covered
in this research.
Keywords: Diabetes, In silico, Insulin dose, Insulin therapy, Personalized treatment, Simulation.
