Machine learning models based on routine blood and biochemical test data for diagnosis of neurological diseases.

Globally, nervous system diseases are the leading cause of disability-adjusted life-years and the second leading cause of mortality in the world. Traditional diagnostic methods for nervous system diseases are expensive. So this study aimed to construct machine learning models using the convenient blood routine and biochemical detection data for diagnosis of nervous system diseases. After the data preprocessing, 25,794 healthy people and 7518 nervous system disease patients with the blood routine and biochemical detection data were utilized for our study. We selected logistic regression, random forest, support vector machine, eXtreme Gradient Boosting (XGBoost), and deep neural network to construct models. Finally, the SHAP algorithm was used to interpret models. The nervous system disease prediction model constructed by XGBoost possessed the best performance (AUC: 0.9782). And the most models of distinguishing various nervous system diseases also had good performance, the model performance of distinguishing neuromyelitis optica from other nervous system diseases was the best (AUC: 0.9095). The model interpretation by SHAP algorithm indicated features from biochemical detection made major contributions to predicting nervous system disease. The present study constructed multiple models using 52 features from the blood routine and biochemical detection data for diagnosis of various nervous system diseases. Meanwhile, distinct hematologic features of various nervous system diseases also were explored. This cost-effective work will benefit more people and assist in diagnosis and prevention of nervous system diseases.