Machine Learning-Based Prediction of Seasonal Influenza Trends in Saudi Arabia - A Tool for Regional Public Health Planning

Aug 6, 2026

Upload by Maeve

Seasonal influenza remains a major public health challenge, especially in densely populated settings where timely early-warning mechanisms are limited. This study assesses the forecasting performance of Support Vector Regression (SVR) and Random Forest (RF) for weekly influenza case counts in Saudi Arabia across 2017–2022, using 313 weekly records from the WHO Global Influenza Surveillance and Response System (GISRS). Model quality is evaluated using R², MAE, MSE, and RMSE. SVR yields higher training accuracy (R²=0.96), while RF provides stronger generalization (R²=0.818) and more stable peak predictions, supporting real-time surveillance and data-driven preparedness.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages9

Reading Time23 min

AccessFree to download

Introduction

Influenza burden and transmission risks in Saudi Arabia

Limits of classical time-series models

Motivation for machine learning approaches

Study aim and model comparison

Machine Learning-Based Prediction of Seasonal Influenza Trends in Saudi Arabia: A Tool for Regional Public Health Planning Alshaikh A. Shokeralla 1,* , Fathelrhman El Guma 1 , Ali H. Abdalla 1 , Amal E.Y. Hagsddig2 , Rahma AbuBakr Musa3,4 , M.A. M. Eltaweel 1,5 , Abdelaziz H. Elawad 1 and Ibrahim Elshamy6,7 1Department of Mathematics, Faculty of Science, Al-Baha University, Al-Aqiq 65931, Saudi Arabia 2Department of Statistics, Faculty of Engineering, Mashreq University, Khartoum, Sudan 3Biology Department, Faculty of Science, Al-Baha University, Saudi Arabia 4Biology Department, Faculty of Applied and Industrial Science, Bahri University, Sudan 5Faculty of Science Ain Shams University Cairo Egypt 6Higher Institute of Engineering & Technology, Almanzala, Egypt 7Mathematics Department, Faculty of Science, Al Baha University, KSA Abstract: Influenza has continued to be a worldwide social health problem, especially in high-density population areas with minimal early-warning mechanisms. The present study evaluates the predictive ability of two machine learning models Support Vector Regression (SVR) and Random Forest (RF) to predict weekly influenza cases in Saudi Arabia, spanning from 2017 to 2022, on 313 weekly influenza records in the WHO Global Influenza Surveillance and Response System (GISRS) . The performance of these models was measured with R² , MAE, MSE, and RMSE. Although SVR hada better training accuracy (R² = 0.96), RF had a better generalization (R² = 0.818) and more consistent predictions at the peaks of the seasons. These observations show that RF is appropriate to real-time influenza surveillance and can provide a replicable and versatile framework to assist data-driven epidemic preparedness in Saudi Arabia and other similar contexts throughout MENA and Asia-Pacific. Keywords: Forecasting, Influenza, Machine learning, Random Forest, Saudi Arabia, Support Vector Regression. 1. INTRODUCTION Seasonal influenza in the world is one of the leading causes of morbidity and mortality with an estimated 290,000 to 650,000 respiratory deaths every year [1] . It is transmitted by airborne droplets and is particularly common in highly populated zones, as well as among the vulnerable populations, such as children, older people and those with chronic conditions . Influenza surveillance is a social health concern in the Middle East where the variability of climatic conditions, crossborder travel, and urban congestion are making it challenging to contain the disease . In the Kingdom of Saudi Arabia (KSA), the frequent large-scale crowds like Hajj and Umrah increase the transmission risks, and well-developed forecasting tools are essential in early detection and interventions [2,3] . Early detection of influenza pandemics supports immunization measures, distribution of resources and preparedness to epidemics. Even though classical time-series forecasting models, including SARIMA, ARIMA, and Holt-Winters exponential smoothing, have *Address correspondence to this author at the Department of Mathematics, Faculty of Science, Al-Baha University, Al-Aqiq 65931, Saudi Arabia; E-mail: [sshokeralla@gmail.com](sshokeralla@gmail.com) been used to predict influenza, they generally assume stationarity and linearity. These assumptions make them less effective in the representation of the multivariate, nonlinear and complex nature of the transmission of infectious diseases [4,6] . Machine learning (ML) methods have become more versatile and data-driven alternatives to traditional public health datasets as they are growing more multidimensional and dynamic. Certain ML models, including Support Vector Regression (SVR) and Random Forest (RF), are capable of processing nonlinear patterns of data on a large scale and a multiplicity of predictors with increased robustness than classical models [7,9] . Recent reports have shown that ML can be used to predict influenza-like illness (ILI) cases in the countries of China and South Ko

FAQs

Which machine learning models were evaluated for forecasting weekly influenza cases in Saudi Arabia?

Support Vector Regression (SVR) and Random Forest (RF) were compared for weekly influenza case prediction.

What data source and time span were used in the study?

The study used 313 weekly records from the WHO Global Influenza Surveillance and Response System (GISRS) covering 2017 to 2022.

How did SVR and RF perform in terms of predictive accuracy and generalization?

SVR achieved better training accuracy (R²=0.96), while RF showed improved generalization (R²=0.818) and more consistent peak predictions.

Aug 6, 2026

Upload by Maeve