An artificial neural network can select patients at high risk of developing progressive IgA nephropathy more accurately than experienced nephrologists

Background. The object of the study was to develop an artificial neural network (ANN) to identify patients with IgA nephropathy (IgAN) with a poor prognosis and to compare the predictions of the ANN with the predictions of six experienced nephrologists. Methods. The following data from the time of renal biopsy were retrieved from the records of 54 patients with IgAN: age, sex, systolic and diastolic blood pressure, number of prescribed antihypertensive drugs, 24-h urine protein excretion, and serum creatinine. Patients aged less than 14 years, or who had serum creatinine >350 mu mol/l at presentation, liver disease or concomitant kidney disease were excluded. Outcome was assigned as 'stable' if serum creatinine was <150 mu mol/l after 7 years and 'non-stable' if serum creatinine was greater than or equal to 150 mu mol/l. The ANN was trained and tested using a 'jack-knife' sampling technique and performance evaluated in terms of number of correct predictions, sensitivity and specificity. The six nephrologists were asked to predict outcome at 7 years for each patient using the same data as the ANN and their performance was assessed in the same manner. Results. The ANN assigned the correct outcome to 47/54 (87.0%) patients: sensitivity 19/22 (86.4%), specificity 28/32 (87.5%). The mean score for nephrologists was 37.5/54 (69.4%, range 35-40), mean sensitivity 72% and mean specificity 66%. Conclusions. An ANN trained using routine clinical information obtained at the time of diagnosis can potentially predict 7-year outcome for renal function in IgAN more accurately than experienced nephrologists, and can therefore identify a group of high-risk patients requiring close follow-up.