Building Complex Disease Models using Ontologies and Data Repositories

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): This application addresses broad Challenge Area (10) Information Technologies for Processing Healthcare Data and specific Challenge Topic, 10-LM-102: Advanced Decision Support for Complex Clinical Decisions. A key direction for our research has been the development of information technologies that can focus and extend the ability of the clinician to make informed medical decisions at the patient bedside. The aim of the current project is to explore a novel modeling technology which may help organize a patient's clinical information and assist in its interpretation. This technology combines a disease ontology with computational tools borrowed from data mining. The ontology is designed to capture the terminology of medicine and to represent key relationships among medical concepts. In addition, it will contain links to data in an active Electronic Health Record (EHR). Using the ontology's semantic infrastructure as a framework, we will overlay computational tools that will support medical pattern recognition and prediction. This hybrid model should effectively recognize patterns that 1) define and diagnose human disease, 2) identify comorbid and complicating factors, 3) identified disease and patient specific therapeutic interventions, and 4) predict outcomes in the context of relevant patient characteristics. If the model proves sufficiently accurate, it can be embedded in applications that assist with diagnosis, documentation, therapeutic planning, and prognosis at the patient bedside. Our approach will be to develop efficient strategies to combine the data models used in Electronic Health Records (EHRs) with published ontologies and other semantic representations. The goal is an ontology that both represents the relationships described above and effectively links to the data models native to an active EHR. Data extracted from this EHR and collected and maintained in an Enterprise Data Warehouse (EDW) will be used to train the computable component of the hybrid model. The rationale for the development of this hybrid technology is to supplant the labor-intensive and time- consuming process used to develop evidence-based guidelines for use in standardizing clinical care. In these efforts, the availability of medical expertise is the rate limiting feature. We seek to develop an automated method that will substantially replace the need for medical experts in the development of guidelines. We will test the success of this approach by implementing this hybrid model for a group of diseases which have been studied extensively in our healthcare system. In this setting we will develop and test a prototype of this computable clinical ontology. The goal of this project is to bring together two technologies to create a mechanism for generating useful medical knowledge. The technologies involved are special electronic dictionaries (ontologies) that describe the way that medical concepts are related and tools that can be trained with information from previous episodes of care to detect diseases, suggest treatments, and predict disease outcomes. We will conduct tests to determine whether the combination of these technologies can be used by medical computing systems to aid in the management of disease by advising caregivers at the bedside.