Applicability of Data Mining and Predictive Analysis for Tobacco Cessation: An Exploratory Study

Area of Science:

• Healthcare data analysis
• Artificial intelligence in medicine
• Predictive modeling in public health

Background:

• Healthcare generates vast data, requiring advanced analysis for insights.
• Predictive analytics, powered by AI, uncovers hidden relationships in health data.
• Tobacco cessation is complex, with varied patient responses influenced by multiple factors.

Purpose of the Study:

• Assess data mining techniques with the WEKA tool.
• Evaluate WEKA's role in predictive analysis for healthcare.
• Predict patient quit status in tobacco cessation programs.

Main Methods:

• Utilized WEKA, a data mining software, for data classification.
• Employed 10-fold cross-validation for model evaluation.
• Applied algorithms including Naïve Bayes, SMO, Random Forest, J-48, and Decision Stump.
• Analyzed secondary data from 655 tobacco cessation clinic patients with 20 attributes.

Main Results:

• Decision Stump and SMO algorithms demonstrated superior prediction accuracy for quit status.
• Identified key predictors of early quitting: previous cessation attempts, intervention type, and years of tobacco use initiation.

Conclusions:

• Data mining and predictive models like WEKA enhance patient outcomes.
• WEKA facilitates the identification of critical variables for effective tobacco cessation interventions.