Medical Expert Systems

Supporting Doctors Making Difficult Diagnoses

 

UPCL is developing a generic framework for intelligent, adaptive medical decision support systems.  Designed in accordance with medical communications protocols such as HL7, a Medical Expert System using our framework would use state of the art technology to reduce misdiagnoses.  In remote communities, where there is limited access to qualified specialists, these systems would be an invaluable tool for doctors and their associates.

Our Medical Expert Systems would calculate the risks of different diagnoses.  The variables for the calculation would be gathered in a number of ways.  A doctor or nurse would input the patient's current symptoms manually.  The patient’s medical history would be transferred into the system directly by accessing existing EMRs. Test results would also have to be factored in.  These could be transferred digitally from a remote lab facility.

Once the Medical System Expert has processed the data and conducted the necessary calculations, the findings are presented to the medical expert.  To illustrate these results, the system would generate a diagnostic space (see pdfs below).  These state of the art 3D graphics would provide medical experts with a clear, visual understanding of the risk values associated with potential diagnoses.

To learn more about this project please vew this presentation and the academic paper that explains the research in full.

Bayesian Decision Space for Medical Diagnosis

Probability Evaluation (using 2 disease hypotheses)

Probability Evaluation (derived from 3 sets of lab results)

Disease Liklihood Analysis (using 4 decision boundaries)