Abstract: Objective To establish prediction model of acute mountain sickness (AMS) based on radial basis function (RBF) neural network. Methods The hemoglobin, P₅₀, body mass index (BMI), single nucleotide polymorphism (SNP) of the EPAS1 and EGLN1 genes were detected in 98 people who urgent need to enter the plateau. According to the diagnostic criteria, AMS was judged after rapid ascent to high altitude. The diagnostic prediction model was established using RBF neural network. Results The correct percentage of the neural network learning training was 88.0%. The constructed neural network was used for prediction, and the coincidence rate between the results and the actual diagnosis results was 88.9%. The diagnostic ability of the prediction model was tested, and the area under the receiver operating characteristic (ROC) curve was 0.917, suggesting a good diagnostic ability. Conclusion The prediction model based on RBF neural network can provide an effective method for early diagnosis of AMS.

Key words: Acute mountain sickness, Neural network, Prediction model