A fuzzy logic system for the home assessment of freezing of gait in subjects with Parkinsons disease

• Published in: Expert systems with applications Vol. 147; p. 113197
• Main Authors: Pepa, Lucia, Capecci, Marianna, Andrenelli, Elisa, Ciabattoni, Lucio, Spalazzi, Luca, Ceravolo, Maria Gabriella
• Format: Journal Article
• Language: English
• Published: New York Elsevier Ltd 01.06.2020; Elsevier BV
• Subjects: Algorithms; Diagnostic systems; Freezing of gait; Fuzzy logic; Gait; Gait recognition; Home monitoring; Laboratories; Parkinson's disease; Smartphone; Freezing of gait; Fuzzy logic; Home monitoring; Smartphone
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2020.113197

•A novel fuzzy logic algorithm for freezing of gait detection is presented.•A smartphone app was developed to enhance usability and acceptability.•High reliability in laboratory tests against clinicians observation.•Home monitoring correlates significantly with laboratory clinical evaluation.•A well known algorithm was applied on the same data for comparison. Gait dysfunctions are pathognomonic, progressive and, generally, continuous in Parkinson’s Disease (PD). The Freezing of Gait (FoG) is an episodic gait disorder involving up to 70% of people with PD, within 10 years of clinical onset, and associated with an increased risk for falls and immobility, which in turn, contributes to greater disability. Automatic and objective monitoring of FoG may help clinicians to understand and treat this phenomenon. In this work, a smartphone app for real-time FoG detection is presented and tested both in a laboratory setting and at patients’ home. The app implements a novel fuzzy logic algorithm that uses important spatio-temporal parameters of gait and is built according to clinical knowledge about FoG. The app includes a gait detection function and the evaluation of two important clinical statistics, i.e. FoG time and FoG number. The app FoG detection performance was assessed against clinicians evaluation and compared with the Moore–Bachlin FoG detection algorithm through ROC analysis, the calculation of confusion matrix, and FoG hit rate. The proposed algorithm achieved better results with respect to the Moore–Bachlin algorithm. Home reports were compared with respect to the FoG Questionnaire and laboratory reports; results indicated significant correlations for both FoG time and FoG number. The results confirm the reliability and accuracy of this app for FoG detection, supporting its wide use for diagnostic and therapeutic purposes.