Data mining and decision support techniques for patients’ treatment management: A case of Parkinson’s disease

Computer-assisted management of patients' treatment is a challenge, especially for chronic diseases where patients are exposed to treatments with multiple medications. While the exposure to prescribed medications aims at reducing the severity of symptoms, it can have adverse effects on patients' quality of life due to the potential side effects of prolonged medication treatments. This tutorial aims to present a selection of methods and tools to empower the tutorial participants in using appropriate data mining and decision support tools for managing treatment and drug management of patients with chronic diseases. We will present the case of utilizing the power of data mining and decision support models for the treatment management of Parkinson's disease (PD) - a chronic neurodegenerative disease affecting people worldwide. The methodologies presented in the tutorial are based on short time-series data of patients' clinical symptoms and their medications treatments. The tutorial will contain a presentation of the data mining methodologies developed for determining patterns of disease progression and patterns and justification of antiparkinson medication therapy modifications. We will conclude the tutorial with the demonstration of a decision support model for antiparkinson medication modifications developed in coordination with clinicians specializing in PD. The presented methods are tested on the data from the Parkinson's Progression Markers Initiative (PPMI) (https://www.ppmi-info.org/). The tutorial is intended for both the health-care and informatics community researchers and practitioners. The audience will be familiarized with some of the challenges of PD, the potential of the PPMI study, the developed short time series methodologies and their possible application to other health problems. The tutorial will be presented by Anita Valmarska, a researcher at the Faculty for Computer and Information Science in Ljubljana, Slovenia. She developed methodologies for the analysis of short time series for determining patterns of PD progression and therapy modifications. Currently, she is working on the development of multi-view based methods for determining subtypes of PD patients.