Investigating Fibromyalgia Detection via SMOTEENN-fused XGBoosted Stacked Learner

Abstract

The Fibromyalgia syndrome is a chronic pain disorder that affects between 2-4%
of global population with diagnostic challenge that relies on subjective symptom
assessment, and absence of specific biomarkers. This leads to delayed diagnosis
and suboptimal result for patients. However, learning schemes have been explored
to aid quick detection of fibromyalgia with limitations including small data set,
single-model dependence, and mishandle of complex clinical data relations. These
have necessitated the deployment of a more robust analytical frameworks, which
this study seeks to advance that will address the gaps. We propose an ensemble
that fuses three (3) base learners with XGBoost. With dataset imbalance resolved
via SMOTEENN, results show superior performance with our ensemble achieving
1.000 for Accuracy, F1 and Precision with 0.999 for Recall and 18secs runtime
efficiency. Report affirms our ensemble's enhanced generalization, and validates
its fusion with proper feature selection and data balancing can substantially
improve fibromyalgia detection to provide clinicians with a robust tool for early
diagnosis to facilitate timely intervention strategies and improved patient care
outcomes in clinical settings.