Crossing heterogeneous information sources for better analysis in Long Term

Abstract. In this paper we describe a methodology that emerged during an im-plementation of a healthcare-oriented data repository for LTC, which consists in grouping information from heterogeneous and distributed information sources. We developed this methodology by first constructing a concrete data repository, con-taining information about elderly patients flows in the UK’s long-term care system (LTC). In our specific case, the role of the data repository is to allow knowledge ex-traction about consumption and behavioural tendencies in the elderly people pop-ulation within the LTC system. These tendencies can be depicted in terms of sur-vival behaviour (modelling), cost evolution, and bed use. Other types of knowledge that can be extracted are typical patient profiles, placement policy in term of rules and criteria effectively applied and, specific features of the business process behind the long-term care provision. A well-constructed data repository can support the discovery (analysis) of hidden aspects about the way the patients are placed and accepted in the LTC, and also how the allocated resources are consumed. We ar-gue that the use of this methodology could save time in similar undertakings or in other fields than LTC or even healthcare. We present some examples where the methodology could be applied with good results