Development of a distributed intelligent decision support system

点击翻译按钮获取中文摘要

Today, no technique does well in predicting the occurrence of a premature birth. We intend to integrate various data mining techniques to increase the accuracy of the prediction of the following outcomes: premature birth, delivery type, and status of the infant at birth. We plan to study three clinical databases to assess which factors best predict these outcomes. The proposed work includes the development of an electronic patient record (EPR) capable of handling the multiple types of information (data, images, etc..) and define a multimedia model adapted to current medical practice in obstetrics and perinatal care. The EPR will be integrated to an improved and expanded decision-support system (DSS). We will implement standards-driven, open and extensible plug-in system architecture; this will seamlessly analyse patient-specific data and information to deliver personalised healthcare. Results stored in the EPR, along with response to treatments and patient prognosis will offer a unique clinical practice tool not available in current research or commercial systems. The final stage is to develop a web services-based infrastructure to enable easy, but secure access to the DSS and to the databases from local and remote locations and by several users at points in time when the information is needed. The prototype will undergo a short pilot test in a clinical setting at the Ottawa Hospital and at the Isaac Walton Killam Hospital in Halifax. Physicians will be able to intervene earlier and improve patient outcomes. Using our system for screening, the costly invasive test (fibronectin) could now be done mainly for patients whose risk estimation is high. Several students will be trained in an area of high demand, in the multidisciplinary environment of engineering, computer science, and medicine.