Arterial hypertension is a chronic disease with multifactorial origin, which has a variety of developments in different patients and can progress with multiple comorbidities. This study aims to develop approach for data-driven identifying hypertensive patient profiles. A digital twin of the hypertensive patient profile will allow to lead various virtual clinical trials, that are necessary to reduce uncertainty within the conditions of clinical decision making. This study represents two approaches for probabilistic modelling of the annual average blood pressure variability based on diverse features of the patient profiles.