Detection of Motor Impairment in Parkinson's Disease Via Mobile Touchscreen Typing

Abstract

Mobile technology is opening a wide range of opportunities for transforming the standard of care for chronic disorders. Using smartphones as tools for longitudinally tracking symptoms could enable personalization of drug regimens and improve patient monitoring. Parkinson's disease (PD) is an ideal candidate for these tools. At present, evaluation of PD signs requires trained experts to quantify motor impairment in the clinic, limiting the frequency and quality of the information available for understanding the status and progression of the disease. Mobile technology can help clinical decision making by completing the information of motor status between hospital visits. This paper presents an algorithm to detect PD by analyzing the typing activity on smartphones independently of the content of the typed text. We propose a set of touchscreen typing features based on a covariance, skewness, and kurtosis analysis of the timing information of the data to capture PD motor signs. We tested these features, both independently and in a multivariate framework, in a population of 21 PD and 23 control subjects, achieving a sensitivity/specificity of 0.81/0.81 for the best performing feature and 0.73/0.84 for the best multivariate method. The results of the alternating finger-tapping, an established motor test, measured in our cohort are 0.75/0.78. This paper contributes to the development of a home-based, high-compliance, and high-frequency PD motor test by analysis of routine typing on touchscreens.

Motor Impairment Estimates via Touchscreen Typing Dynamics Toward Parkinson's Disease Detection From Data Harvested In-the-Wild

“…The diagnostic properties of the produced indexes achieved up to 0.8/0.77 sensitivity/specificity on classifying PD and healthy subjects in the wild setting, when aggregating for the whole time period of data contribution, which matches the satisfactory performance seen in-the-clinic setting analyses, i.e., 0.81/0.82 in Iakovakis et al (2018) and 0.81/0.81 in Arroyo-Gallego et al (2017). The results are also compliant with the findings of Arroyo-Gallego et al (2018), who suggests that keystroke dynamics on physical keyboard can be used for remote PD screening with sensitivity/specificity of 0.73/0.69.…”

“…Fine-motor skills decline can also be detected from typing patterns on mobile touchscreen as derived from recent similar works on typing pattern analysis on mobile touchscreen (Arroyo-Gallego et al, 2017;Iakovakis et al, 2018), during in-the-clinic experiments. In our latest study, we proposed a feature vector representation of enriched keystroke information and a two-stage machine learning-based pipeline to process multiple typing sessions as captured from mobile touchscreen, performing 0.92 AUC with 0.82/0.81 sensitivity/specificity on early PD and healthy subject's classification.…”

Evaluation of smartphone‐based testing to generate exploratory outcome measures in a phase 1 Parkinson's disease clinical trial

“…Several pilot research studies have successfully developed smartphone applications for PD . In these pilot studies, proof of concept was typically established in a clinical setting by demonstrating significant differences between individuals with PD and healthy controls, and/or significant relationships between the sensor‐based measures and the International Parkinson and Movement Disorder Society–Sponsored Revision of the Unified Parkinson's Disease Rating Scale (MDS‐UPDRS) clinical gold standard . For example, Kassavetis and colleagues tested 14 PD participants (mean disease duration = 3.7 years) with the MDS‐UPDRS and a custom Android application with the following active tests: resting, postural and kinetic tremor, pronation‐supination, leg agility, and finger‐tapping.…”

“…1,[9][10][11][12][13][14][15][16][17] In these pilot studies, proof of concept was typically established in a clinical setting by demonstrating significant differences between individuals with PD and healthy controls, and/or significant relationships between the sensor-based measures and the International Parkinson and Movement Disorder Society-Sponsored Revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS) clinical gold standard. 14,[18][19][20][21][22][23][24] For example, Kassavetis and colleagues 18 tested 14 PD participants (mean disease duration 5 3.7 years) with the MDS-UPDRS 25 and a custom Android application with the following active tests: resting, postural and kinetic tremor, pronation-supination, leg agility, and fingertapping. For all tasks, the extracted sensor feature data significantly correlated with corresponding MDS-UPDRS 25 item scores (e.g., item 3.17, rest tremor amplitude).…”

Remote Monitoring of Treatment Response in Parkinson's Disease: The Habit of Typing on a Computer

“…14 Recently, we demonstrated similar performance with data acquired during typing on a touch-screen smartphone. 15 In the present study, our aim is to detect the response to medication in PD by using remotely gathered, unsupervised typing data in an at-home, everydaylife setting as an additional step to this new digital care model. Thus, we designed a prospective naturalistic study enrolling early PD patients who were going to start dopaminergic medication and followed them for 6 months.…”