Home // ALLDATA 2019, The Fifth International Conference on Big Data, Small Data, Linked Data and Open Data // View article

Efficient Qualitative Method for Matching Subjects with Multiple Controls

Authors:
Hung-Jui Chang
Yu-Hsuan Hsu
Chih-Wen Hsueh
Tsan-sheng Hsu

Keywords: matching; observational study; relative entropy

Abstract:
In the era of learning healthcare systems and big data, observational studies play a vital role to discover hidden (causal) associations in the dataset. To control bias, a matching step is usually employed to match case subjects to control candidates in observational studies randomly. The matching ratio refers to the number of control candidates matched with one case subject, and the successful matching rate is the percentage a matching is found given a matching ratio. A good matching algorithm should be not only efficient but also have high successful matching rate and high quality of randomness which means that a control candidate has a roughly equal chance of being matched with any of the matchable study cases. In this paper, we propose a matching algorithm, which is efficient with above mentioned good properties, RandFlow, a high-quality matching algorithm, is proposed and compared with commonly used ones – Simple_Match, Matchit, and Optmatch. The benchmark testing shows the effectiveness of the new algorithm. In our experimental studies, we noticed that the variation of the estimated Relative Risk (RR) value is minimized at the maximum matching ratio. Thus, we propose a two-phase matching method to obtain more reliable study results. The first phase is to identify the maximum matching ratio, and followed by matching multiple times and then take an average.

Pages: 46 to 51

Copyright: Copyright (c) IARIA, 2019

Publication date: March 24, 2019

Published in: conference

ISSN: 2519-8386

ISBN: 978-1-61208-700-9

Location: Valencia, Spain

Dates: from March 24, 2019 to March 28, 2019