HealthPro: Designing Search Strategies for Retrieving Data from Clinical Data Warehouse

Abstract

Reusing the data collected in electronic medical record system (EMRs) is essential to improve clinical research efficiency. But, it is difficult to identify patients who meet research suitable criteria and collect the necessary information from EMRs because the data collection process must include various type of information stored in EMRs and integrate various techniques, including the development of a data warehouse. We designed a data model optimized for patient identification and for the collection of necessary information from EMRs, and provide the way to search a criteria in the data model. This research aimed to demonstrate an retrieval system(HealthPro) and an example of a hospital-based data that used the HealthPro to identify suitable patients.

The searching system uses data from 550,852 patients, which were imported into clinical data warehouse. The search condition was structured in tree for clinical research, and the tree structure consist of patient demography, diagnosis, laboratory test, clinical document. It enables to search detailed extraction in conditions of various clinical terms. Semantic search with questionaries of clinical documents using MDRS(MetaData Repository System) and drug treatment pattern for patients in particular period are well performed. The algorithm is developed for converting EAV format to table format when extracting data in data warehouse. Query performance averagely recorded 0.9945 seconds of response time with one to ten queries.

In this work, HealthPro performed to design the data model and develop it for retrieving optimized clinical data. We provide the way to search a criteria using hierarchical tree and standardization terminology in the system. Also meta search is available by using controlled vocabulary for identifying patient who meet expanded criteria. HealthPro may be useful by providing a system package supported from data collection to analysis. Additionally, HealthPro plan to include analysis module with R statistics tool and can provide simple calculation about the result of retrieving clinical data.