An Approach to Optimizing Machine Learning Models for the Diagnosis of COVID-19 via Hyperparameter Optimization

Aug 5, 2026

Upload by Maeve

Hyperparameter selection is a critical and challenging step in building effective classification and prediction models, directly affecting generalization and classifier performance. This study evaluates five machine learning methods—SVM, AdaBoost, RandomForest, XGBoost, and DecisionTree—on a COVID-19 dataset collected from Albert Einstein Hospital in São Paulo, Brazil. Two experiments are compared: default parameter settings versus GridSearch-based tuning. Metrics include accuracy, precision, recall, AUC, and F1-score, showing that improved hyperparameters increase recall by 18%.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Introduction

Literature survey

Dataset and machine learning algorithms

Proposed prediction methodology

Experimental results and discussion

Conclusion and future work

African Journal of Management Engineering and Technology Copyright © 2023, SupMoTI Rabat Morocco Afr. J. Manag. Engin. Technol., 2023, Vol. 1, N° 2, 149-160 An Approach to Optimizing Machine Learning Models for the Diagnosis ofCOVID-19 via Hyperparameter Optimization Hamida S. 1, 2*, El-Gannour O. 3, 4, Cherradi B. 3, 5 1 2IACS Laboratory, ENSETof Mohammedia, Hassan II University of Casablanca, Mohammedia, Morocco. 2 GENIUS Laboratory, SupMTIof Rabat Rabat, Morocco. 3 EEIS Laboratory, ENSETof Mohammedia, Hassan II University of Casablanca Mohammedia, Morocco. 4 LIASSE Laboratory, ENSA of Fez, Sidi Mohamed Ben Abdellah University, Fez, Morocco. 5 CRMEF Casablanca-Settat, provincial section of El Jadida, 24000, El Jadida, Morocco. *Corresponding author, Email address: [hamida@enset-media.ac.ma](hamida@enset-media.ac.ma) Received 11 Nov 2023, Revised 21 Dec 2023, Accepted 23 Dec 2023 Citation: Hamida S., ElGannour O., Cherradi B. (2023) An Approach to Optimizing Machine Learning Models for the Diagnosis of COVID-19 via Hyperparameter Optimization, Afr. J. Manag. Engg. Technol., 1(2), 149-160 Abstract: The process of picking appropriate hyper-parameters for classification or prediction algorithms is a tough endeavor in the field of modeling. This selection is essential for the capacity for generalization and the performance of classifiers. Over the course of two tests, this article examines and evaluates the performance of five different Machine Learning (ML) algorithms: Support Vector Machine (SVM), AdaBoost, RandomForest, XGBoost, and DecisionTree. When it comes to training and testing, the first experiment makes use of the default settings, while the second experiment makes use of the GridSearch function to locate the most effective configurations. The tests make use of a dataset that was gathered from Albert Einstein Hospital in Sao Paulo, Brazil, and the dataset contains anonymous individuals that either have or do not have COVID-19. Usage of evaluation metrics includes things like accuracy, precision, recall, area under the curve (AUC), and F1-score. According to the findings, improving hyper-parameters results inan 18% improvement in recall. Keywords: Optimization; Hyper-parameters; Coronavirus; Machine learning; Model 1. Introduction As the coronavirus disease (COVID-19) posed a worldwide risk to lives, businesses, and travel, reports on the virus are in a constant state of flux. On January 30, 2020, and March 11, 2020, respectively, the World Health Organization (WHO) classified COVID-19 as a public health emergency of international concern and a pandemic. Assembling solutions to this health crisis requires immediate action. Hours are required to analyze a nasopharyngeal specimen submitted to a laboratory for virologic or PCR tests that detect the virus (Qiu et al., 2016) . Despite this, many nations continue to lack adequate COVID-19 testing capabilities, which necessitates the development of additional early detection tools to prevent the virus's spread (Laghmati et al., 2020; Moujahid et al., 2020; Terrada, et al., 2020; Touzani et al., 2020; Toubi et al., 2024) . Understanding the virus's transmission and implementing evidence-based measures to limit the pandemic are facilitated by early diagnosis. A WHO report from February 2020 posited that the utilization of big data analysis and Artificial Intelligence (AI) might prove indispensable in the fight against the COVID-19 pandemic (Terrada, et al., 2020) . As a subfield of AI, Machine Learning (ML) is concerned with the design, analysis, and implementation of automated learning methods for machines. Significantly utilized in the medical sphere, ML has demonstrated its worth in predicting positive cases of numerous diseases (Song et al., 2019) . A machine learning model is a mathematical construct whose parameters require data-driven tuning. Users select hyper-parameters, which are not amenable to direct examination, through testing an

FAQs

Which machine learning algorithms are evaluated for COVID-19 diagnosis?

The study evaluates Support Vector Machine (SVM), AdaBoost, RandomForest, XGBoost, and DecisionTree.

How does the paper compare hyperparameter tuning strategies?

It compares training with default settings against training with tuned configurations found using GridSearch.

Which evaluation metrics are used to assess model performance?

Accuracy, precision, recall, AUC (area under the curve), and F1-score are used as evaluation metrics.

Aug 5, 2026

Upload by Maeve