Predicting Childhood Diabetes from Microarray Data: A Cuckoo Search Algorithm - Support Vector Machine Approach

Diabetes is a medical condition characterized by elevated blood sugar levels, which the body struggles to regulate effectively. It manifests in three primary forms: type 1 diabetes, type 2 diabetes, and gestational diabetes. Type 1 diabetes, specifically, arises from the autoimmune destruction of insulin-producing cells, resulting in insufficient insulin production. Remarkably, diabetes ranks as the third leading cause of death in Indonesia, contributing to 6.7% of overall mortality. Timely detection of diabetes plays a pivotal role in mitigating its considerable mortality rate. In light of advancing technology, the integration of machine learning techniques with microarray data has emerged as a promising avenue for early diabetes detection. This study employs the Cuckoo Search (CS) algorithm in conjunction with the Support Vector Machine (SVM) to predict diabetes based on microarray data. Hyperparameter tuning is applied to three different kernels: RBF, Poly, and Linear, with the aim of enhancing model performance. Our findings reveal that SVM with the RBF kernel yields the most promising outcomes, achieving accuracy and F1-score values of 0.809 and 0.878, respectively, on the test dataset. These results signify a significant step forward in the realm of diabetes prediction, holding potential for early intervention and improved patient outcomes.