A Kinematic Sensor and Algorithm to Detect Motor Fluctuations in Parkinson Disease: Validation Study Under Real Conditions of Use

Abstract

BackgroundA new algorithm has been developed, which combines information on gait bradykinesia and dyskinesia provided by a single kinematic sensor located on the waist of Parkinson disease (PD) patients to detect motor fluctuations (On- and Off-periods).ObjectiveThe goal of this study was to analyze the accuracy of this algorithm under real conditions of use.MethodsThis validation study of a motor-fluctuation detection algorithm was conducted on a sample of 23 patients with advanced PD. Patients were asked to wear the kinematic sensor for 1 to 3 days at home, while simultaneously keeping a diary of their On- and Off-periods. During this testing, researchers were not present, and patients continued to carry on their usual daily activities in their natural environment. The algorithm’s outputs were compared with the patients’ records, which were used as the gold standard.ResultsThe algorithm produced 37% more results than the patients’ records (671 vs 489). The positive predictive value of the algorithm to detect Off-periods, as compared with the patients’ records, was 92% (95% CI 87.33%-97.3%) and the negative predictive value was 94% (95% CI 90.71%-97.1%); the overall classification accuracy was 92.20%.ConclusionsThe kinematic sensor and the algorithm for detection of motor-fluctuations validated in this study are an accurate and useful tool for monitoring PD patients with difficult-to-control motor fluctuations in the outpatient setting.

Systematic Review Looking at the Use of Technology to Measure Free-Living Symptom and Activity Outcomes in Parkinson’s Disease in the Home or a Home-like Environment

“…Often combined with other motor outcomes such as dyskinesia or gait, eight studies investigated the ability of technology to measure bradykinesia/akinesia in their participants [42,52,[54][55][56][80][81][82]. All the research groups used wearable devices to measure bradykinesia/akinesia: four used single devices whereas four used multiple wearable devices.…”

“…All of the devices employed to measure bradykinesia/akinesia contained accelerometers, with gyroscopes also in four devices [52,54,55,80] and a gyroscope/magnetometer within one [56]. Validation was attempted through the use of videotape [55,56], comparison with the UPDRS [42,52], the use of telephone calls from researchers and comparison with participant study diaries [82]. Clinimetric properties were evaluated in six studies, and included measurements of accuracy [52,55,81,82] and agreement (Pastorino et al [80] documented good correspondence of 88.2% between technological outcomes and participant diaries).…”

“…Validation was attempted through the use of videotape [55,56], comparison with the UPDRS [42,52], the use of telephone calls from researchers and comparison with participant study diaries [82]. Clinimetric properties were evaluated in six studies, and included measurements of accuracy [52,55,81,82] and agreement (Pastorino et al [80] documented good correspondence of 88.2% between technological outcomes and participant diaries). Responsiveness to dose adjustments of levodopa/carbidopa was assessed in a study by Johansson et al [42], with the UPDRS motor scores changing significantly (P < 0.05) but the wearable device's summary scores (bradykinesia & dyskinesia) showing no significant change following dose adjustments of levodopa/carbidopa.…”

“…Motor fluctuations specifically were evaluated as a stand-alone outcome measure by two studies [83,84], and they were evaluated alongside the dyskinesia outcome measure in three papers [81,82,85]. Dyskinesia was measured by technologies in 11 further papers included in this review [37-40, 42, 45, 52, 54-56, 68].…”

“…Wearable devices were used in all studies which evaluated dyskinesia and/or motor fluctuations, with seven using single wearable devices. Sama et al [56], Rodriguez-Molinero et al [82,84] and Perez-Lopez et al [81] used devices located at the waist or on the belt, Johansson et al [42] and Bayes et al [83] used a device at the wrist and Ramsperger et al [45] used an ankle-worn device. Accelerometers were ubiquitously employed; they were used in some studies without other sensing technology and in other papers alongside gyroscopes [45,52,[54][55][56], magnetometers [56] and surface electromyographs [37][38][39][40].…”