Diabetes is a serious, long-term condition that occurs when the body cannot produce any or enough insulin or cannot effectively use the insulin it produces. Diabetes type I (T1D) – a chronic, life-long, non-curable disease with high percentage of complications. Blood glucose control and insulin dose calculations are pivotal components in diabetes self-monitoring. This study represents an attempt to use machine learning techniques to personalize and optimize the bolus insulin dosage calculations for patients with T1D. The proposed model has been trained using pharmacodynamics profiles and a virtual dataset. The study lasted for 2 months.