Proceedings of LOUHI '08. The First Conference on Text and Data Mining of Clinical Documents

Abstract

The First Louhi Conference on Text and Data Mining of Clinical Documents (Louhi'08) seeks to bring together researchers and practitioners into a multi- disciplinary conference on new application areas for intelligent systems. The often repeated challenges to health care include also the vast amounts of documentation produced during care and as outcomes of health and medical research. For the practitioners to manage with these amounts, many intermediaries – such as summaries, code lists and rulebooks - are in use. To turn health and medical research usable also to practitioners, medical journals present concise reports, medical databases provide search engines and classifications for finding relevant information, and evidence based guidelines form the rules to follow. The methods and algorithms presented in this conference seek to give computational tools for improving the automated processes and for supporting the manual work in forming these intermediaries. What make the Louhi Conference and the Louhi project1 unique is that we seek to constantly reflect our work onto the actual work carried out in hospitals and clinics (Suominen et al 2005). Thus, daily and discharge summaries in a hospital ward are more than just summaries; they are carefully constructed assessments of care given so far (Berg 2004). Code lists such as ICD-10 not only enumerate illnesses, they also tell of the culture of care giving (Bowker & Star 1999). Clinical guidelines not only provide state of the art advice on certain situations and conditions, but following them may also have unintended consequences (Timmermans & Berg 2003). We are only beginning to find out the ethical quagmires of using intelligent systems in actual care (Suominen et al 2007). We have also learned that the work practices in the busy hospitals are quite fragile, especially when technologies around them break down (Forsell, Karsten & Vuokko 2007; Karsten & Vuokko 2008). The Proceedings of this very first international Louhi conference show some of the multitude and complexity of the issues we are facing. In the five full research papers we can read many of the challenges outlined. In the two first papers, evidence-based guidelines are to be constructed with the help of research databases and analyses of patient records. The third paper seeks to build a syntactic parse to enable consequent semantic analysis of patient records. The fourth paper deals with improving the quality of information in diagnosis-based registries. The fifth paper moves the discussion to a hospital by presenting four actual cases of technology use in wards and by outlining lessons for future technology implementations, including intelligent systems.