Use of machine learning techniques in chronic obstructive pulmonary disease: A case study in Baja California, Mexico

Chronic Obstructive Pulmonary Disease (COPD) is a chronic inflammatory disease of the lungs that obstructs airflow from the lungs. Symptoms include difficulty breathing, coughing, mucus production, and wheezing. The study was conducted with 769 patients. A total of 48.70% were women, and 51.29% were men. The average age of the patients enrolled was 60 years. The research included 67 variables considering the medical history and biochemical data. The objective is to evaluate chronic obstructive disease. We used automatic learning techniques to assess and identify the patient's determinant variables. The following classifiers were used: Vector Support Machines (SVM), K-Nearest Neighbors (kNN), Decision Tree, Random Forest, Neural Network, AdaBoost, and Logistic Regression. The model suggests that the determining variables for COPD in treated patients are the following: TA_dist, Cholesterol level, LDL levels, Dyslipidemia, Bradycardia, Venous Insufficiency, Systolic Dysfunction, Cardiac Arrhythmia, Vasomotor Headache, Smoking, and Esophageal Achalasia. Therefore, they are considered relevant in the decision-making process for choosing treatment or prevention. The analysis of the relationship between the presence of the variables and the classifiers used to measure COPD revealed that the Logistic Regression classifier, with the variables TA_dist, Cholesterol level, LDL levels, Dyslipidemia Bradycardia, Venous Insufficiency, Systolic Dysfunction, Cardiac Arrhythmia, Vasomotor Headache, Smoking, and Esophageal Achalasia, showed an accuracy of 0.90, precision 0.87 and an F1 score of 0.89. Therefore, we can conclude that the Logistic Regression classifier gives the best results for evaluating the determining variables for COPD assessment.