Software components integration in medical data warehouses: a proposal

Abstract

Data warehousing is an attractive solution for centralizing and analyzing high-quality data. In the medical research field, this technology can be used to validate assumptions and to discover trends in large amounts of patient data. However, in many medical studies based on knowledge extraction, medical data, and especially raw sensor data, need to be pre-processed before turning to valuable material for data-mining features. Scientists, researchers and physicians emphasize the necessity to be able to select raw sensor data, processed information, and also the appropriate transforming processes. We propose a solution to ease medical data mining by integrating data and processes in a single warehouse. Our solution provides features for loading, modelling and querying a medical data warehouse. The stored data are multi-dimensional data (patient identity, therapeutic data, etc.), raw sensor data (ECG, X-rays, etc.) and software components. A prototype has been implemented and is illustrated in the cardiology domain.