Assessing Parkinson's Disease Apps Using Thematic Analysis And Machine Learning Techniques

Abstract

Parkinson's disease (PD) is a complicated neurological condition that needs to be managed and watched over constantly. People with Parkinson's disease (PD) now have access to a range of tools to help with their care because to the widespread use of mobile health applications, or apps. This study offers a novel method for qualitative and quantitative assessing PD apps by utilizing machine learning methods combined with a sentient analysis and thematic examination of user feedback. The research processes a sizable dataset of user reviews from mobile apps linked to Parkinson's disease (PD) using sentiment analysis and natural language processing (NLP). The main topics and user-expressed demands are found using thematic analysis, which also highlights the applications' advantages and disadvantages in dealing with PD-related issues. Prior studies have frequently concentrated on technical elements, but in order to improve the selection of applications catered to the specific requirements of PD patients. In our research, we evaluated 105 Parkinson's disease apps that were offered on Google Play and the App Store. We used a two-step method that involved thematic analysis of these reviews after doing sentiment analysis on 65,972 user reviews using machine learning (ML) techniques. Five supervised ML classifiers that are well-known for classification tasks were implemented for sentiment analysis, and their performance was compared. We were able to forecast the emotion polarity thanks to the best classifier, which received a remarkable F1 score of 96.97% and show accuracy of negative and positive reviews in well quantitative structure. The negative and positive themes encompassed various aspects, including low navigation problems, goal setting, app stability, simplicity, customized, alert and notify, in-app support, desired effect and Parkinson's disease, enjoy ability high-quality content, logging, encouragement of data. In conclusion, addressing these negative factors, we aim to enhance the overall effectiveness quality and user experience of these vital healthcare applications.