A Survival Metadata Analysis Responsive Tool (SMART) for web-based analysis of patient survival and risk

Abstract

Health information systems contain extensive amounts of patient data. Information relevant to public health and individuals’ medical histories are both available. In clinical research, the prediction of patient survival rates and identification of prognosis factors are major challenges. To alleviate the difficulties related to these factors, Metadata Utilities was developed to help researchers manage column definitions and information such as import/query/generator Metadata files. These utilities also include an automatic update mechanism to ensure consistency between the data and parameters of the batch produced in the conversion procedure. Survival Metadata Analysis Responsive Tool (SMART) provides a comprehensive set of statistical tests that are easy to understand, including support for analyzing nominal variables, ordinal variables, interval variables or ratio variables as means, standard deviations, maximum values, minimum values, and percentages. In this article, the development of a raw data source and transfer mechanism, Extract-Transform-Load (ETL), is described for data cleansing, extraction, transformation and loading. We also built a handy method for data presentation, which can be customized to the trial design. As demonstrated here, SMART is useful for risk-adjusted baseline cohort and randomized controlled trials.