Automated Knowledge Structuring of Medical Charts Data

Some major military health problems are caused or exacerbated by environmental factors. Gulf War Syndrome, reactions to Agent Orange, and other such problems tend to have diverse symptoms and delayed (and variable) latencies to onset. A major reason they are so difficult to identify, diagnose, and manage is capability shortfalls in military (and civilian) medical information systems. Linkages among systems are sub-optimal (although being vigorously pursued). Critically, tools do not exist that support the kinds of information analysis, extraction, and exploration required in current and future data sources. Aptima and its partner, the University of Colorado Computational Bioscience Program, propose to design (in Phase I) and develop (in Phase II) a toolset called Automated Structuring of Knowledge for Medical Extraction of Data (ASK-MED). ASK-MED will enable military medical professionals to find clinically-relevant patterns in data across various information systems. Our strategy avoids the technical tarpit of ontology alignment and instead integrates a powerful top-down, ontology-neutral knowledge-based method with bottom-up, domain-independent statistical language analysis technique. The solution will add value to existing informatics efforts, without requiring that all the surrounding institutional and technical problems be solved.