Objective: To provide the neurologist with a decision support system based on inertial sensors and learning algorithms for an objective early diagnosis of Parkinson’s disease (PD) analysing motion.

Background: Disabling motor and non-motor symptoms characterize PD since its development in prodromal phase. Even if PD diagnosis is still mainly based on motor assessment, researches are focusing also on non-motor deficits. Among them, idiopathic hyposmia (IH), a reduced olfactory sensitivity, is a PD preclinical marker and affects >95% of PD patients.

Method: 15 healthy controls (HC), 15 IH people, and 15 PD patients were instrumented with two wearable inertial devices, i.e. SensHand [1] and SensFoot [2][Figure 1], for acquiring motor data from upper and lower limbs. Their motion was recorded while they performed ten tasks of the MDS-UPDRS III. Inertial signal processing in spatiotemporal and frequency domains allowed to extract 142 kinematic parameters from analysis of both right and left sides. The statistically significant and uncorrelated features were selected to classify the different groups. Performances of a Random Forest (RF) classifier were evaluated on two-group (HC vs. PD) and three-group (HC vs. IH vs. PD) classification, and considering upper and lower limbs separately or the four limbs together as a full system.

Results: The best performances were achieved using the full system configuration. Excellent results were obtained for HC vs. PD classification (1.00 accuracy), and also including IH subjects as a third group (0.91 accuracy) [Figure 2].

Conclusion: The system results suitable to support an objective PD diagnosis. Furthermore, combining motion analysis with a validated olfactory screening test [3], people at risk for PD can be analyzed, and subtle changes in motor performance that characterize the prodromal phase and the early PD onset can be identified.

References: [1] F. Cavallo, et al. (2019) Upper limb motor pre-clinical assessment in Parkinson’s disease using machine learning. Parkinsonism Relat Disord https://doi.org/10.1016/j.parkreldis.2019.02.028 [2] E. Rovini, et al. (2018) Comparative motor pre-clinical assessment in Parkinson’s disease using supervised machine learning approaches Ann Biomed Eng 46(12):2057-68 [3] C. Maremmani, et al. (2018) Combining olfactory test and motion analysis sensors in Parkinson’s disease preclinical diagnosis: A pilot study Acta Neurol Scand 137:204–11

« Back to 2019 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/motor-based-assessment-of-prodromal-parkinsons-disease-combining-wearable-sensors-and-machine-learning/