Random Forest Analysis Based on Fingertip Pulse for Changes in an Individual Before and After Discontinuation of Medication

Depression is a common mental disorder, and medication is one of the common treatment methods. However, there is currently a lack of a simple and non-invasive method to assess the changes in depression patients before and after medication. In this study, the researchers combined depression with fingertip pulse and machine recognition. They designed a method that uses signal processing to extract peaks from fingertip pulse and calculate heart rate variability features. The random forest algorithm was used to identify the changes in depression patients before and after discontinuation of medication. The experimental results indicate that by utilizing the random forest algorithm in combination with features of heart rate variability, a classification accuracy of 72%, sensitivity of 68%, specificity of 78%, and an area under the ROC curve (AUC) of 80% were achieved for identifying differences before and after medication, demonstrating the best classification performance.