International projects
Predicting patient’s future health state: Development and deployment of fast, effective, and interpretable algorithms for healthcare
Organisations (1)
Abstract
We are facing a huge shift in healthcare systems, as more and more data is being analyzed to improve quality of healthcare and reduce costs. Most of the data in hospitals is currently stored in Electronic Medical Records (EMRs) and has usually large volume, includes temporal sequences of events, and is often very sparse. A lot of efforts are put into analysis of EMR data, and there are several algorithms proposed for analysing EMR data and extracting useful patterns for healthcare. However, algorithms that can handle the complex nature of this data and provide interpretable results are currently in their infancy. Therefore, the goal of this project is to develop algorithms that will report interpretable, still reliable, patterns to the scientific and practitioner community through the Rapid Miner open-source platform. More specifically, with the newly developed algorithms in this project we aim at predicting readmission probability, predict future health states, and extract patient-related and clinical-related features that determine future health states. EMR data in the experiments will be taken from US and Slovenian hospitals.
