JMIR Publications

ABSTRACT

Background:

Diabetic retinopathy (DR) is one of the most important causes of blindness worldwide, especially in developed countries. In diabetic patients, examining periodically the back of the eye using a non-mydriatic camera has been widely demonstrated to be an effective system to control and prevent the onset of RD. Convolutional neural networks have been used to detect RD, achieving very high sensitivities and specificities.

Objective:

● To develop an artificial intelligence logarithm for the detection of signs of DR in diabetic patients. ● To scientifically validate the logarithm to be used as a screening tool in primary care.

Methods:

This project will consist in the conduction of two studies in a concomitant way: 1. Development of an algorithm with artificial intelligence to detect signs of DR in patients with diabetes. 2. Prospective study comparing the diagnostic capacity of the algorithm with respect to the actual system of family physicians evaluating the images. The standard reference to compare with will be a double blind reading done by ophthalmologists specialists in retina. For the development of the artificial intelligence algorithm, different iterations and workouts will be performed on the same set of data. Before starting each new workout, the strategy of dividing set date in two groups will be used randomly. A group with 80% of the images, which will be used during the training (training data set) and the other, of the remaining 20% (validation data set) it will be used to validate the results of each cycle (epoch). During the prospective study, the values of correct positive (TP), correct negative (TN), positive (FP), false negative (FN) will be calculated again. From here we will obtain the resulting confusion matrix and other indicators to measure the performance of the algorithm.

Results:

It is hoped to develop and validate an artificial intelligence system for the detection of DR in diabetic patients. If the results are satisfactory, the developed algorithm can be used as a support tool for family doctors.

Conclusions:

The study will allow the development of an algorithm based on artificial intelligence that can demonstrate an equal or superior performance, and that constitutes an alternative, to the current screening of RD in diabetic patients.