Optimized Extreme Learning Machine with Bacterial Colony Optimization Algorithm for Disease Diagnosis in Clinical Datasets

• Published in: SN computer science Vol. 5; no. 5; p. 584
• Main Authors: Vigneshvaran, P., Kathiravan, A. Vijaya
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 26.05.2024; Springer Nature B.V
• Subjects: Accuracy; Algorithms; Artificial neural networks; Bias; Colonies; Computer Imaging; Computer Science; Computer Systems Organization and Communication Networks; Convergence; Data Structures and Information Theory; Datasets; Diagnostic systems; Disease; Illnesses; Information Systems and Communication Service; Machine learning; Medical diagnosis; Neural networks; Optimization; Original Research; Pattern Recognition and Graphics; Software Engineering/Programming and Operating Systems; Support vector machines; Vision; Convergence rate; Bacterial colony optimization; Artificial neural network; Extreme learning machine; Disease prediction
• ISSN: 2661-8907; 2662-995X; 2661-8907
• DOI: 10.1007/s42979-024-02864-8

Currently, the computational disease diagnostic models are mainly for approximating the non-linear complex sensitive patterns in the medical data. For the most accurate diagnostic, it is to be processed by more sophisticated algorithms. The clinical data is obtained from various clinical tests conducted on both infected and healthy individuals. It is crucial to precisely identify the presence of sickness to provide treatment promptly. Disease detection models must accurately differentiate between true positive and negative outcomes. Manual interpretations may result in either false positives or false negatives. This study aims to create a universal framework for diagnosing diseases on clinical datasets. For this scenario, the Extreme Learning Machine (ELM) is combined with the Bacterial Colony Optimization (BCO) method to boost prediction accuracy and speed up global convergence. For efficient selection of the extreme learning machine model’s optimal weights and biases, it is recommended to use the BCO algorithm's neighborhood-based communication strategy. Eight different UCI medical datasets are used to analyze the performance of the developed optimized ELM method. The experimental results showed that the suggested BCO-ELM outperformed in terms of accuracy, sensitivity, specificity, precision, f-measure, ROC-AUC, convergence speed. It obtained approximately 5 to 10 percent maximal accuracy and f-measure than the existing traditional approaches and obtained an AUC of 0.9 to 1 on all datasets. The experimental results confirmed that the developed approach produced high classification accuracy when compared with other existing approaches.