Atif Khan, John Doucette, and Robin Cohen
"Holmes – Hybrid Ontological & Learning MEdical System Decision Support System"

Medical decision support systems have a great potential to enhance clinical 
care, and they have been shown to improve practitioner performance as well 
as other medical work-flows. Despite this significant advantage, the 
traditional MDSS have not been successful in gaining mainstream acceptance 
in the industry. Information and its various dimensions (such as access, 
representation, heterogeneity, availability, interoperability, etc.) are key 
contributors to this lack of success.  These challenges forces real world 
medical decisions to be often based on partial information. We present a 
framework that enables medical decision making in the presence of partial 
information.  Our proposed construction leverages ontological representation, 
machine learning and automated reasoning to create a 'hybrid' medical decision 
support system - Holmes.  Holmes combines the strengths of traditional 
knowledge-based and learning-based decision support systems, allowing for 
informed patient-centric evidence-based decision making.  It integrates 
distributed medical datasets in real-time, and provides a robust solution for 
missing information. It is suitable for deployment in many different healthcare 
settings for various users (such as physicians, nurses, and other medical 
staff). Furthermore, the framework facilitates a strong integration support using 
a query based interface, enabling multiple types of software application clients 
such as desktop applications, web-based browser applications, and mobile 
applications.  We demonstrate its effectiveness on real world data and sketch its 
use for a variety of medical domains. In short, we demonstrate the potential for 
artificial intelligence to support a task where there is a critical need from 
medical professionals.