Big Data, Machine Learning and Contraceptive Use: A Scoping Review

Abstract

Abstract The use of big data sources, like Twitter, and big data analytical techniques, like machine learning, have increased in popularity in almost every area of scientific inquiry. However, recent reviews have not focused on contraceptive use to prevent pregnancy, which is surprising considering that over one-third of unmet need for contraception in low- and middle-income countries is made up of women who have discontinued a method. This manuscript details the results of a scoping review of peer-reviewed literature at the intersection of big data and contraceptive use to prevent pregnancy. We developed the Metrics of Reliability and Quality (MARQ) to provide guidance to assess studies using big data to understand contraceptive use and beyond. We found 31 articles that matched our inclusion criteria. The oldest article was published in 1971, and 61.3% (N = 19) of articles were published after 2016. Many articles using big data sources applied traditional analytical methods rather than big data methods. The overall quality of articles on the MARQ rubric was high; however, many articles employing big data sources did not discuss specific limitations, such as population representativeness or bias, and articles using big data methods seldom demonstrated whether big data methods outperform traditional analytical methods.