Using LASSO Regression to Predict Rheumatoid Arthritis Treatment Efficacy

Area of Science:

• * Biomedical informatics
• * Clinical informatics
• * Rheumatology

Background:

• * Rheumatoid arthritis (RA) is a leading cause of disability and mortality in the US.
• * Current RA treatment guidelines lack drug efficacy prediction.
• * Understanding autoimmune disorder mechanisms and patient variability poses treatment challenges.

Purpose of the Study:

• * To develop and demonstrate a method for classifying patient outcomes in rheumatoid arthritis.
• * To leverage Electronic Health Records (EHR) and Biomedical Linked Open Data (LOD) for improved treatment prediction.
• * To provide insights into differential drug treatment outcomes.

Main Methods:

• * Utilized LASSO penalized regression for patient outcome classification.
• * Integrated Electronic Health Records (EHR) with Biomedical Linked Open Data (LOD).
• * Developed a predictive model for treatment success.

Main Results:

• * Demonstrated that Linked Data enhances prediction accuracy for RA patient outcomes.
• * Identified insights into how different drug treatment regimens affect outcomes.
• * Showcased the potential of machine learning classifiers in clinical decision support.

Conclusions:

• * The proposed method using EHR and LOD can classify RA patient outcomes effectively.
• * This approach offers potential for improving clinical decision-making in RA management.
• * Successful application could reduce the physical and financial burden of RA on patients and healthcare systems.