Diagnostic Value and Effectiveness of an Artificial Neural Network in Biliary Atresia

Abstract

Purpose: Biliary Atresia (BA) is a devastating pediatric liver disease. Therefore, early diagnosis is important for timely surgical intervention and better prognosis. Using clinical parameters for non-invasive and efficient diagnosis of BA, we aimed to apply and evaluate the effectiveness of an artificial neural network (ANN). Methods: A total of 2,384 obstructive jaundice patients who were suspected of BA from 2012 to 2017 were screened for eligibility. Furthermore, 137 clinical parameters from all patients were screened for data integrity. A standard binary classification feed-forward ANN, consisting of an input layer, 3 hidden layers, and an output layer, was employed. To avoid overfitting, a dropout method was included for the hidden layer. The network was trained and validated for accuracy. As an independent predictor, gamma-glutamyl transpeptidase (GGT) level was used as a comparison to assess the effectiveness of the network by ROC curves. Results: A total of 46 qualified parameters and 1,452 patients were included for ANN modeling, including 1274 patients (87.7%) in the BA group and 178 patients (12.3%) in the non-BA group. Total bilirubin, direct bilirubin, and GGT were the most significant indicators in BA patients. The network consisted of an input layer, 3 hidden layers with 12 neurons each, and an output layer with one neuron. The network showed good predictive property with a high Area under curve (AUC) (0.967), a sensitivity of 97.2% and a specificity of 91.0%. Five-fold cross validation for the ANN model showed a mean accuracy (95% CI) for the training data was 93.2% (92.5%-93.9%), while the mean accuracy to the validation data was 88.6% (88.2%-88.9%). Conclusions: The high accuracy and efficiency demonstrated by the ANN model is promising in the non-invasive diagnosis of BA patients. As a result, the application of ANN as a low-cost and independent expert diagnosis system for medical institutions at different levels of healthcare should be considered for the future. Funding Statement: This study received financial support from Shanghai Key Disciplines (no.2017ZZ02022), National Natural Science Foundation of China (no. 81770519, no. 81771633, no. 81873545 and no. 81974059), The Science Foundation of Shanghai (no. 14411969800, no. 16411952200, no. 18411969100 and 19ZR1406600), and Children's National Medical Center (no. EK1125180104, no. EKYY20180204, EK112520180211 and no. EK112520180310). Declaration of Interests: The authors stated: None. Ethics Approval Statement: This study was approved by the Ethics Committee of the Children’s Hospital of Fudan University. Informed consent was obtained from the legal guardians of all patients before enrollment in the study.