ABSTRACT
BACKGROUND: Systematic Reviews (SR), studies of studies, use a formal process to evaluate the quality of scientific literature and determine ensuing effectiveness from qualifying articles to establish consensus findings around a hypothesis. Their value is increasing as the conduct and publication of research and evaluation has expanded and the process of identifying key insights becomes more time consuming. Text analytics and machine learning (ML) techniques may help overcome this problem of scale while still maintaining the level of rigor expected of SRs. METHODS: In this article, we discuss an approach that uses existing examples of SRs to build and test a method for assisting the SR title and abstract pre-screening by reducing the initial pool of potential articles down to articles that meet inclusion criteria. Our approach differs from previous approaches to using ML as a SR tool in that it incorporates ML configurations guided by previously conducted SRs, and human confirmation on ML predictions of relevant articles during multiple iterative reviews on smaller tranches of citations. We applied the tailored method to a new SR review effort to validate performance. RESULTS: The case study test of the approach proved a sensitivity (recall) in finding relevant articles during down selection that may rival many traditional processes and show ability to overcome most type II errors. The study achieved a sensitivity of 99.5% (213 out of 214) of total relevant articles while only conducting a human review of 31% of total articles available for review. CONCLUSIONS: We believe this iterative method can help overcome bias in initial ML model training by having humans reinforce ML models with new and relevant information, and is an applied step towards transfer learning for ML in SR.
Subject(s)
Diabetes Mellitus , Machine Learning , Humans , Mass Screening , Research DesignABSTRACT
In 2016, Centers for Disease Control and Prevention (CDC) established surveillance of pregnant women with Zika virus infection and their infants in the U.S. states, territories, and freely associated states. To identify cases of Zika-associated birth defects, subject matter experts review data reported from medical records of completed pregnancies to identify findings that meet surveillance case criteria (manual review). The volume of reported data increased over the course of the Zika virus outbreak in the Americas, challenging the resources of the surveillance system to conduct manual review. Machine learning was explored as a possible method for predicting case status. Ensemble models (using machine learning algorithms including support vector machines, logistic regression, random forests, k-nearest neighbors, gradient boosted trees, and decision trees) were developed and trained using data collected from January 2016-October 2017. Models were developed separately, on data from the U.S. states, non-Puerto Rico territories, and freely associated states (referred to as the U.S. Zika Pregnancy and Infant Registry [USZPIR]) and data from Puerto Rico (referred to as the Zika Active Pregnancy Surveillance System [ZAPSS]) due to differences in data collection and storage methods. The machine learning models demonstrated high sensitivity for identifying cases while potentially reducing volume of data for manual review (USZPIR: 96% sensitivity, 25% reduction in review volume; ZAPSS: 97% sensitivity, 50% reduction in review volume). Machine learning models show potential for identifying cases of Zika-associated birth defects and for reducing volume of data for manual review, a potential benefit in other public health emergency response settings.