Characterizing machine learning studies on ADRD caregivers' social media posts: A systematic review. (December 2021)

• Record Type: Journal Article
• Authors: Zou, Ning; Wang, Zhendong; luo, Zhimeng; Xie, Bo; He, Daqing; Hilsabeck, Robin C; Aguirre, Alyssa B
• Is Part Of: Alzheimer's & dementia. Volume 17(2021)Supplement 7
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-12
• Subjects: Alzheimer's disease -- Periodicals; Alzheimer Disease -- Periodicals; Dementia -- Periodicals; Démence; Maladie d'Alzheimer; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.83
• DOI: 10.1002/alz.056528
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: Social media have increasingly been used by ADRD caregivers to share and seek information. An emerging trend in the literature is using machine learning (ML) methods to understand caregivers' information behaviors on social media. To date no systematic review exists on what ML methods have been used, what the findings are, and what needs to be improved in order to better understand ADRD caregivers' information behaviors on social media. This systematic review addresses these gaps in the literature. Method: Following the PRISMA guidelines, during October and November 2020, we performed three rounds of screening. First, we searched predetermined keywords in ACM Digital Library, IEEE Xplore Digital Library, and PubMed. This step generated 324 results. After removing 92 articles that were not peer‐reviewed journal/conference proceedings and 2 duplicate articles, 230 results remained. Next, we screened the titles and abstracts of the 230 articles using predetermined inclusion/exclusion criteria; 200 articles were excluded, and 30 remained. Finally, we screened the full text of the 30 articles to ensure that they met the inclusion/exclusion criteria; 12 articles were excluded, leaving a final sample of 18 studies for analysis. Result: All 18 studies covered at least two of the three focal areas: ADRD, ML, and social media; only 6 covered all three areas. Study aims of these 18 studies were in two groups: 1) 4 papers focused on characterizing ADRD caregivers' behaviors on social media; and 2) 14 papers aimed to build ML models to predict ADRD related activities. Twitter was the data source for 14 studies; other sources included Reddit, Youtube, Weibo, and AskaPatient. The key features for modeling ADRD related activities included semantic representations of post content, positive/negative sentiments, and network structures representing posting/answering activities. Conclusion: ML provides promising tools for extracting useful information from large amounts of social media posts about ADRD caregivers' information behaviors. Research in this area is still at its infancy, and much more attention is needed to understand how ML methods can be best utilized. Due to the interdisciplinary nature of the task, effective collaborations leveraging expertise from health sciences, information science and social sciences will be critical.