Quantitative and Qualitative assessment of human movement is a key challenge for clinicians, physiotherapists and researchers supporting clinical practice. This is because the congruence of limb, head and trunk movements with expected (existing or newly described by computational models) patterns is widely used for screening. This is used to assess the need to refer patients for further, more accurate diagnosis and to assess functional status (e.g. in physiotherapy) for the purpose of planning therapy and evaluating its effectiveness and the need to modify the therapy plan. Such movement analysis reflects the sensitivity of the patient's musculoskeletal system to disorders resulting from injury or disease (including at the level of motor control - of neurological origin) and requires precise and refined methods of analysis. The aim of this study was to use a model of multiple adaptive neural-fuzzy inference system with a hybrid learning algorithm to classify data collected in patients undergoing post-stroke rehabilitation of gait (including haemiplegic gait) and hand movement (including spastic hand). The approach described in this study, when practically developed, may lead to a breakthrough in mass automatic or semi-automatic movement assessment and thus in the provision of physiotherapy services through the use of artificial intelligence, in particular neuro-fuzzy systems.