LEVERAGING MACHINE LEARNING METHODS IN PREDICTING AND ANALYZING THE ASSOCIATION BETWEEN DIETARY INFLAMMATORY INDEX AND ALOPECIA

Alopecia areata (AA) is considered a chronic inflammatory disorder and represents a worldwide public health problem. Diet was hypothesized to play a role in AA development, but little is known about the association between the dietary inflammatory index (DII) and AA. This study aimed to analyze the correlation between AA and DII using machine learning at aj(ML) models. DII scores were ascertained using a food frequency questionnaire (FFQ), and the Severity of Alopecia Tool (SALT) score was used to classify the severity of AA. Three machine learning models were developed: K-Nearest Neighbors (KNN) with dimensionality reduction to prevent overfitting, Logistic Regression with L2 regularization, and Random Forest enhanced through grid search for hyperparameter tuning. Additionally, to further understand the association between DII and AA, partial dependence plots (PDPs), correlation analysis, and multiple evaluation indicators, including accuracy, F1 score, recall, precision, and area under the receiver operating characteristic curve (ROC AUC) were used. Surprisingly, higher DII scores are significantly associated with an increase in AA. In addition, a higher inflammatory diet was associated with increased severity of the disease. The highest accuracy was achieved by the Random Forest Classifier 98.77%, whereas 98.64% and 98.10% were achieved by Logistic Regression and KNN models, respectively. This study presents evidence about the association between inflammatory food patterns and AA, which may provide important implications for future treatment and dietary interventions. A high score on the DII indicated an increased proinflammatory potential of food intake and was associated with an increase in AA