Seismic predictions in the mediterranean: machine learning insights and a meta-analysis review of recent studies

¹ Laboratory of Informatics, Robotics and Complex Systems (RISC), National Engineering School of Tunis (ENIT), University of Tunis El Manar (UTM), B.P N°37, Le Belvedere, 1002, Tunis, Tunisia
² CN&S Research Laboratory, Higher School of Communications of Tunis (Sup’Com), University of Carthage, ElGhazala Technopark, Route de Raoued Km3.5, 2083 Ariana, Tunisia
³ Higher Institute of Information and Communication Technologies (ISTIC), University of Carthage, B.P N°123, Hammam Chatt 1164, Tunisia

^* Corresponding author: imen.ziadi@istic.ucar.tn

Received: 25 September 2024
Accepted: 28 May 2025

Abstract

Earthquake prediction is a critical aspect of seismology, especially for regions prone to seismic activity. This systematic review, conducted following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, evaluates the efficacy of machine learning techniques in earthquake prediction within Mediterranean contexts. Through a comprehensive search across academic databases and seismic archives, this review examines the application of various machine learning algorithms for earthquake prediction. The findings highlight the potential of machine learning models, trained on historical seismic data, to predict earthquake occurrences in Mediterranean regions. The methods used include linear regression, time series analysis and the Informer model. However, the review also underscores challenges and limitations, including the need for high-quality and diverse datasets, as well as robust validation methods. By adhering to PRISMA standards, this review provides a comprehensive analysis of the current state of earthquake prediction using machine learning techniques in Mediterranean regions, offering insights for future research and methodological advancements.