Downloads & Free Reading Options - Results

Objective: This scoping review aims to identify artificial intelligence methods applied to patient-reported data for prediction of patient outcomes in healthcare. Introduction: Artificial intelligence methods have been used to predict outcomes for patients from many types of data, including diagnostic images, genetic data, and time-series data from wearable sensors or electronic healthcare records. The lack of sufficient evidence of AI methodology applied to potentially equally informative self-reported symptom data is the motivation for this scoping review. Inclusion criteria: Primary research studies reported in English will be included in this review. Considered population will be patients, who self-reported their symptoms in the context of healthcare. The concept analysed in the review will be the methods of Artificial Intelligence used for prediction of patient outcomes. The outcomes can be related to health status, mortality, treatment, healthcare utilisation and self-management of health issues. Methods: The databases used in the proposed scoping review will be Web of Science, IEEE Xplore, ACM Digital Library, Cochrane Central Register of Controlled Trials, Medline and Embase. The search terms will be adapted to each database, addressing the same research question. Following exporting of references, duplicates will be removed, and identified papers will be screened against the inclusion and exclusion criteria. The comparison of methods, evaluation styles, data types and healthcare contexts reported in selected articles will be presented in a narrative and a tabular form.