An Automated Decision Support System to Predict the Health Status of Eye Tissues

Abstract

The success of a corneal transplant depends
highly on the health condition of the donated cornea.
Currently the health condition of an eye tissue is determined
by a medical expert after the slit lamp test procedure for
bulbous evaluation, which is both expensive and time
consuming. If the health status of an eye can be predicted
accurately without running expensive and time consuming
laboratory tests the valuable tissue samples can be distributed
more effectively with the available limited resources. This
paper introduced a methodology to predict health status of eye
tissues without laboratory test procedures. A novel approach
for developing a medical decision support system combining
Data mining and expert system technologies were attempted in
this study. Flex Expert System shell was designed for the
inference engine and SQL Server analytics were employed for
data mining. The implemented system was tested with actual
eye tissue data and the agreement between the human expert’s
decisions and the system’s predictions are promising. These
predictions were generated as “Not Approved”, “Approved”,
“Good”, “Better” and “Best” based on the prediction
probability values generated by the Inference Engine.