DOI: 10.12688/f1000research.189842.1 ISSN: 2046-1402

From Separation to Synergy: A Systematic Literature Review on Machine Learning–IoT Integration for Disaster Risk Management

Nur Wachid Hidayatulloh, Harliyan Tri Mardian, Musdalifah Musdalifah, Zulfah Binti Toyibah, Indrianti Susanto, Andrian Kaspari, Ervananda Putri Juliana, Corry Norbertha Atwandan Komeep, Alfin Baso, Abillio Attalarik Syech, Muh. Bayu Syafei, Elsi Zeheskiel Balla
Show PDF Cite
Abstract Natural disasters such as earthquakes, floods, landslides, and wildfires continue to increase in frequency and impact globally, necessitating risk management systems that are faster, more accurate, and based on real-time data. Machine Learning (ML) and the Internet of Things (IoT) are widely used to support disaster prediction, early detection, and mitigation; however, most research still treats these two technologies separately, so patterns of ML–IoT collaboration have not yet been systematically studied. This study employs a Systematic Literature Review (SLR) approach following the PRISMA 2020 guidelines to map trends, approaches, and challenges in the integration of ML and IoT for disaster risk management. The search was conducted in the ScienceDirect, Scopus, IEEE Xplore, and Springer databases, yielding 533 initial records that were filtered down to 402 relevant articles; ultimately, 91 articles met all inclusion criteria for in-depth analysis. The study’s results indicate that deep learning is the most widely used approach (36.3%), with floods (36.3%) and landslides (23.1%) being the most frequently studied types of disasters. IoT sensors were identified as the dominant data source (42.9%), while early warning systems were the most commonly implemented ML–IoT integration architecture (34.1%). Nevertheless, the integration of ML and IoT within a single, cohesive framework remains relatively limited, and the main challenges identified include model accuracy and generalization, limitations in data quality, and the lack of maturity in operational aspects such as scalability, interoperability, and IoT device energy consumption. These findings underscore the need for further research focused on a more holistic ML–IoT integration that is ready for operational implementation in disaster risk management.

More from our Archive