Specialised Tools for Automating Data Mining for Hospital Management

This paper presents a research project which is directed to the partial automation of Data Mining (DM) in hospital information systems (HIS). We concentrate on hospital management applications and information systems such as emergencies and ward management, human resources (services, night duties, etc.), physical resources (beds, intervention theatres, etc.), etc. We have realised how the business objectives are usually the same across several hospitals and so is the information which is gathered in several HIS (even using different DBMS). This means that although the models extracted highly differ between hospitals, data mining processes are highly similar across different hospitals. We argue how a tool can be constructed in such a way that it automates many DM processes and that can be ported to other hospitals which could benefit more quickly of a first DM experience. Our work plan covers all the stages in the process of Knowledge Discovery from Databases (KDD): data cleansing, extraction and integration from the HIS and external data, construction of tasks and minable views, model generation, and finally a module to carry out and interpret their predictions. We also consider a module to perform simulations and to integrate the models extracted by the previous modules with other decision support systems as well as model monitoring.