Detecting diseases in medical prescriptions using data mining methods.

Area of Science:

• Medical Informatics
• Data Mining
• Machine Learning

Background:

• Misdiagnosis affects millions globally, leading to adverse health outcomes and significant financial burdens.
• Diagnostic errors and medication prescription mistakes negatively impact physicians' careers.

Purpose of the Study:

• To leverage data mining techniques for improved diagnostic processes using medical prescription data.
• To develop and evaluate an Ensemble Learning model for enhanced disease and disease category prediction.

Main Methods:

• Utilized four single classification algorithms: decision tree, random forest, simple Bayes, and K-nearest neighbors.
• Implemented an Ensemble Learning methodology to improve the performance of individual classification algorithms.
• Conducted experiments to compare the performance of various data mining techniques.

Main Results:

• The proposed Ensemble Learning model achieved 62.86% accuracy and 0.620 kappa score for disease prediction.
• The model demonstrated 74.39% accuracy and 0.720 kappa score for disease category prediction.
• The developed model outperformed existing studies in the field.

Conclusions:

• The study's findings can improve patient health outcomes and reduce financial waste in healthcare.
• The results offer timely information to aid physicians in preventing diagnostic errors and supporting their careers.
• This research provides a foundation for future investigations in data-driven medical diagnosis.