Effective telemonitoring for the optimization of new deep brain stimulation therapies for parkinson’s disease patients

openIntroduction: Deep brain stimulation (DBS) outcomes could benefit from monitoring the fluctuation of Parkinson’s disease symptoms in order to set the best stimulation parameters. A correct and constant monitoring of this time-frame is essential to regulate their therapies. The scope of this work is to develop and validate a system architecture to continuously monitor the patient through personally collected data. In addition, it could support the development and implementation of new DBS approaches aimed to real-time adapt DBS parameters according to the evaluation of the patient’s clinical state (i.e., adaptive DBS, aDBS). Methods: The implementation followed a three-step approach. First, a dedicated Android App was developed to provide a clinical e-diary to be filled in by the patients at predefined times. The App was paired with a smartwatch that acquires tri-axial accelerometric data from the wrist. The data acquired were used to verify whether this combination of two simple technologies was enough to collect data relevant for tracking patient’s activity and symptoms. Second, accelerometric data and diary data were integrated to neurophysiological data, in order to obtain a comprehensive view on the patient’s state. Subthalamic nucleus local field potentials (STN LFPs) were recorded from the implanted DBS electrodes and integrated with the self-collected data. A web-based platform was developed to support data collection and analysis. The platform expanded the architecture of an already established technology in order to introduce a standards-based architecture aimed to implement a bidirectional exchange between patient-generated data and the clinical data repository. The validation study enrolled 13 Parkinson’s disease patients undergoing DBS electrode implant surgery. During an 8 hours session, the patients were asked to fill in the e-diary and to wear the smartwatch. A clinician assessed their condition compiling a Unified Parkinson’s Disease Rating Scale part III (UPDRSIII). After the 8 hour trial, a clinician asked the patients if the smartwatch was uncomfortable during the day. Results: In total, of the 13 patients, 2 were dropped due to technical issues. Two algorithms, the Bradykinesia Accelerometric Score (BAS) and Bradykinesia Index (BradIndex) were developed using the data collected through the smartwatches to estimate bradykinesia. Both provided significative inverse correlation with UPDRSIII evaluation (BAS: Pearson’s correlation coefficient, 0.541, p<0.004; BradIndex: Pearson’s correlation coefficient -0.500, p < 0.0005). The patient reported e-diary status provided significative correlation with the UPDRSIII assessment (Pearson’s correlation coefficient -0. 7416, p < 0.0005) and also with the BradIndex accelerometric index (Pearson’s correlation coefficient 0.6042, p < 0.05). LFP recordings were modulated during walking, with respect to talking and relaxing (beta power change from baseline during walking: -14%±4.212, talking:-11.2 %±2.724, and relaxing: -8.811%±2.418, one-way ANOVA p