A Dynamic, Platform-Based, Serious Personalized Game Suite to Assist Rehabilitation of Cardiovascular and Parkinson Disease Patients

Abstract

Cardiovascular disease (CVD) is considered the major cause of premature death and disability in the UAE. Cardiac Rehabilitation (CR) is a well-known effective factor in significantly improving mortality and morbidity rates. However, the currently unsustainable provision of healthcare in CR Phase III results in patients failing to adhere to their rehabilitation program. Nonetheless, neurodegenerative Parkinson's disease (PD) is also one of the common incurable diseases amongst the elderly. Clinical assessments are characterized as standardized means for PD diagnosis. However, relying on medical evaluation of a patient's movements can be subjective to physicians' experience, making the diagnosis process susceptible to human errors, and hence possible misclassification. 'Care4MyHeart' (C4MH) project introduces a personalized home-based rehabilitation program, allowing lifestyle behavioral change for increasing quality of life for patients. A personalized and gamified Serious Games (SGs) environment that can monitor and track the behavioral change of CVD and PD patients is presented here. The work is extended by the development of a rigorous web platform to assist physicians in managing their patient's rehabilitation programs through analyzing their acquired data from the platform. The data produced from the gamified platform taken from the i PROGNOSIS project was used as an initial evaluation in this study, due to the delays encountered from the ongoing COVID-19 pandemic. These patient-specific data were used in a feature selection technique and Machine Learning frameworks for correlation analysis and prediction of PD patients based on the severity level of their disease. The experimental results show that the integrated Assessment Tests are highly correlated with the clinical rating scale, resulting in using the highly correlated features for building the classifier model. The classification problem was further expanded using LOOCV, achieving an overall accuracy of 88.89%. Through the integration of advanced machine learning models, the C4MH-PGS platform is able to generate gender and age-specific rehabilitation programs, along with an autonomous agent, to provide informative feedback to the patient and the physician. The findings of this work can assist SGs designers in developing better personalized, holistic, and home-based CR programs. At the same time, due to its holistic design, the C4MH-PGS main system and sub-components can easily be transferred to address several other diseases such as diabetes, obesity, and more, commercialization by that the C4MH-PGS in the market.

Date of Award	Dec 2021
Original language	American English