Telemonitoring is the use of technology (e.g., smart phones, wearable devices, and other electronics) that enables patient monitoring outside of the clinical setting. It reduces the inconvenience of frequent clinical visits and allows for more frequent monitoring of the disease of interest. Parkinson's Disease (PD) is the second-most common neurodegenerative disease and affects 7-10 million individuals world-wide. In order to assess the severity of someone's PD, the Movement Disorder Society Unified PD Rating Scale (MDS-UPDRS) was devised [1]. To acquire the MDS-UPDRS score of a PD patient, one must use an in-clinic questionnaire that requires a patient's physical presence. Unfortunately the assessment is subjective and the frequency of the assessment is low, which results in a lack of disease progress information and leads to sub-optimal treatment. Advances of sensing and computer technologies have produced immense amounts of data that can be used to improve PD diagnosis and treatment.
A novel statistical model for simultaneous feature and instance selection in semi-supervised regression (S2SSL) is developed. S2SSL is based on manifold learning in the Reproducing Kernel Hilbert Space (RKHS). Both feature and instance selection are integrated within the same non-linear model framework that provides flexibility for modeling the complicated relationship between the activity data and MDS-UPDRS PD severity score.
S2SSL is applied to smartphone-based telemonitoring of PD patients to predict PD severity by using activity data collected by the mPower smartphone app [3]. The selected features by the algorithm also shed some light on which aspects of the movement and speech functions of PD patients are mostly impaired by the disease. S2SSL can also help improve healthcare automation in several aspects such as; enabling frequent, remote health monitoring for each patient and timely medical decision; improving patient access to advanced care; and facilitating the design of efficient and effective patient triage systems.
Goetz, C. G., Tilley, B. C., Shaftman, S. R., Stebbins, G. T., Fahn, S., Martinez‐Martin, P., ... & Dubois, B. (2008). Movement Disorder Society‐sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS‐UPDRS): scale presentation and clinimetric testing results. Movement disorders: official journal of the Movement Disorder Society, 23(15), 2129-2170.
Gaw, N. (2019). Novel Semi-Supervised Learning Models to Balance Data Inclusivity and Usability In Healthcare Applications (Doctoral dissertation, Arizona State University). https://repository.asu.edu/attachments/221561/content/Gaw_asu_0010E_19135.pdf
Bot, B. M., Suver, C., Neto, E. C., Kellen, M., Klein, A., Bare, C., ... & Friend, S. H. (2016). The mPower study, Parkinson disease mobile data collected using ResearchKit. Scientific data, 3(1), 1-9.
