Abstract
Purpose :
Diagnostic challenge represents a significant source of delay in treatment and morbidity in medicine. While artificial intelligence electronic differential diagnosis support systems have been developed and found to be helpful in many branches of medicine, they have not been studied in orbital diseases. We propose a probability-weighted scoring system based differential diagnosis spreadsheet to offer potential diagnoses in cases of orbital pathology.
Methods :
Using a probability-weighted scoring system algorithm, features on history, physical exam, lab testing and imaging were gathered and deemed to be important in diagnosis of orbital pathology. These features were then given weighting based on probability and prevalence for many orbital pathologies and integrated into a spreadsheet to create the orbital differential diagnosis spreadsheet (ODDS). 16 cases from an orbital textbook were then entered into ODDS and a list of 10 differential diagnoses were compiled and analyzed.
Results :
The ODDS was able to determine the correct diagnosis in the top ten differentials in 9/16 cases (56.25%). Among these cases, the median position of the diagnosis was second. Cases without a correct diagnosis were pathologies that were quite rare and signaled a need to update the database of the spreadsheet.
Conclusions :
The ODDS shows promise in aiding diagnoses in standard orbital pathology. Optimization for more rare pathology is required. Further work will look into comparing results with artificial intelligence electronic differential diagnosis support systems such as Isabel.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.