AUTOMATIC EVALUATION OF MOBILE HEALTH APPLICATIONS ACCORDING TO PERSUASIVE SYSTEM DESIGN PRINCIPLES AND MOBILE APPLICATION RATING SCALE

Date

2023-6

Author

AFŞİN, YASİN

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

246
views

259
downloads

Cite This

Mobile applications have become an integral part of our daily lives and they have increasingly been used by the healthcare industry. However, many mobile health applications lack proper regulations or preliminary assessments, which could compromise users' health and safety. Existing literature has relied on manual assessment methods using Persuasive System Design (PSD) principles and Mobile Application Rating Scale (MARS) to evaluate user engagement and the quality of these applications. This thesis proposes a novel automatic evaluation technique to extend the assessment of applications in the market, with a focus on employing large language models to filter user reviews and generate sentence embeddings to classify the applications' use of PSD principles. Results show that it is possible to predict an application's employed PSD principles based on user reviews, while application descriptions do not provide enough information. Additionally, predicted classification probabilities of PSD principles are enriched with additional descriptive data, such as install counts and user ratings to predict MARS scores. Regression models trained using these methods outperform baseline models, with feature importance scores and SHAP values demonstrating the contribution of predicted classification probabilities of PSD principles to the models. Overall, this study suggests that automatic evaluation techniques can be effective in assessing the quality and user engagement of mobile health applications, providing an alternative to manual assessment.

Subject Keywords

Data Mining, Mobile Health Applications, Persuasive System Design, Mobile Application Quality

URI

https://hdl.handle.net/11511/104449

Collections

Graduate School of Informatics, Thesis