Computer model for gait assessments in Parkinson's patients using a fuzzy inference model and inertial sensors.

Abstract

Patients with Parkinson's disease (PD) in the moderate and severe stages can present several walk alterations. They can show slow movements and difficulty initiating, varying, or interrupting their gait; freezing; short steps; speed changes; shuffling; little arm swing; and festinating gait. The Movement Disorder Society-Unified Parkinson's Disease Rating Scale (MDS-UPDRS) has a good reputation for uniformly evaluating motor and non-motor aspects of PD. However, the motor clinical assessment depends on visual observations, the results are qualitative, and subtle differences are not identified. This study presents a fuzzy inference model for gait assessments in PD patients with detailed descriptions of signal processing and eight biomechanical indicators computations; as such, other authors can replicate the presented methods. The computer model uses 334 bilateral measurements of 58 Parkinson's patients and 15 healthy control subjects performed over one year. The computer model validations are based on physician evaluations in real-time and post-analysis using an extensive database of videos and signals. The assessment results are explainable, quantitative, and qualitative, increasing their acceptance and use in clinical environments. The computer system design considers three expert motor evaluations, including the PD patients' evolutions; this facilitates correlation with medication doses and appropriate intervals for follow-up medical consultations. The assessments include three qualitative gait conditions of MDS-UPDRS-normal, slight, and mild-as well as a numerical evaluation of up to two decimal places.