View in EDS

Artificial intelligence and computational methods for modelling and forecasting influenza and influenza-like illness: a scoping review

Saved in:
Bibliographic Details
Title: Artificial intelligence and computational methods for modelling and forecasting influenza and influenza-like illness: a scoping review
Authors: Adekunle Adeoye, Isreal Ayobami Onifade, Michael Bayode, Idowu Michael Ariyibi, Benjamin Akangbe, Oluwabunmi Akomolafe, Tesleem Ajisafe, Delower Hossain, Oluwatope Faith Owoeye
Source: Beni-Suef University Journal of Basic and Applied Sciences, Vol 14, Iss 1, Pp 1-20 (2025)
Publisher Information: SpringerOpen, 2025.
Publication Year: 2025
Collection: LCC:Medicine (General)
LCC:Science
Subject Terms: Artificial Intelligence, Deep Learning, Forecasting Models, Influenza, Predictive Accuracy, Medicine (General), R5-920, Science
Description: Abstract Background The persistent resurgence of influenza and influenza-like illness despite concerted vaccination interventions is a global health burden, thus necessitating accurate tools for early intervention and preparedness. This scoping review aims to map the currently available literature on artificial intelligence (AI)-based forecasting models for seasonal influenza and to identify trends in those published models, approaches, and research gaps. Methods A detailed search was conducted in PubMed, Scopus, and IEEE Xplore to find relevant studies published between 2014 and 2025. The AI techniques (such as machine learning and deep learning) applied in predicting seasonal influenza activity are considered eligible studies. Model types, data inputs, performance metrics, and validation approaches were summarized on data that were extracted and charted. Results Nine studies met the inclusion criteria and were included. Owing to their effectiveness in solving temporal sequence models, many deep learning models have been applied, including the long short-term memory (LSTM) model and the CNN LSTM hybrid model. The data sources are epidemiological records, meteorological variables and social media signals. Most of the models achieved excellent predictive accuracy, but shortcomings in model interpretability, external validation or consistency across performance reporting became issues. Conclusions Although AI-based models show promising capabilities for predicting influenza, there are still issues related to standardization and deployment in the real world. Future work should focus on real-time data integration, external validation and interpretable transferable models appropriate for a wide variety of health settings. Graphical Abstract This graphical abstract encapsulates AI-based forecasting models for seasonal influenza, depicted as a navigational chart through the research terrain. A central magnifying glass over a globe anchors the global health challenge, guiding the viewer through a flowchart-like journey. A funnel filters literature from PubMed, Scopus, and IEEE Xplore (2014–2025), yielding 9 pivotal studies. Layered icons delineate machine learning and deep learning models, with LSTM and CNN-LSTM hybrids highlighted. Interconnected circles symbolize diverse data inputs—epidemiological, meteorological, and social media—converging into a data integration hub. The bar chart connotes high predictive accuracy, tempered by a warning sign flagging interpretability, validation, and reporting challenges. A roadmap at the journey’s end points to future horizons: real-time data integration, external validation, and interpretable models, charting the course for advancing global influenza preparedness.
Document Type: article
File Description: electronic resource
Language: English
ISSN: 2314-8543
Relation: https://doaj.org/toc/2314-8543
DOI: 10.1186/s43088-025-00682-2
Access URL: https://doaj.org/article/bca7942b8be641fc93b152f7e8a88383
Accession Number: edsdoj.bca7942b8be641fc93b152f7e8a88383
Database: Directory of Open Access Journals
Description
Abstract:Abstract Background The persistent resurgence of influenza and influenza-like illness despite concerted vaccination interventions is a global health burden, thus necessitating accurate tools for early intervention and preparedness. This scoping review aims to map the currently available literature on artificial intelligence (AI)-based forecasting models for seasonal influenza and to identify trends in those published models, approaches, and research gaps. Methods A detailed search was conducted in PubMed, Scopus, and IEEE Xplore to find relevant studies published between 2014 and 2025. The AI techniques (such as machine learning and deep learning) applied in predicting seasonal influenza activity are considered eligible studies. Model types, data inputs, performance metrics, and validation approaches were summarized on data that were extracted and charted. Results Nine studies met the inclusion criteria and were included. Owing to their effectiveness in solving temporal sequence models, many deep learning models have been applied, including the long short-term memory (LSTM) model and the CNN LSTM hybrid model. The data sources are epidemiological records, meteorological variables and social media signals. Most of the models achieved excellent predictive accuracy, but shortcomings in model interpretability, external validation or consistency across performance reporting became issues. Conclusions Although AI-based models show promising capabilities for predicting influenza, there are still issues related to standardization and deployment in the real world. Future work should focus on real-time data integration, external validation and interpretable transferable models appropriate for a wide variety of health settings. Graphical Abstract This graphical abstract encapsulates AI-based forecasting models for seasonal influenza, depicted as a navigational chart through the research terrain. A central magnifying glass over a globe anchors the global health challenge, guiding the viewer through a flowchart-like journey. A funnel filters literature from PubMed, Scopus, and IEEE Xplore (2014–2025), yielding 9 pivotal studies. Layered icons delineate machine learning and deep learning models, with LSTM and CNN-LSTM hybrids highlighted. Interconnected circles symbolize diverse data inputs—epidemiological, meteorological, and social media—converging into a data integration hub. The bar chart connotes high predictive accuracy, tempered by a warning sign flagging interpretability, validation, and reporting challenges. A roadmap at the journey’s end points to future horizons: real-time data integration, external validation, and interpretable models, charting the course for advancing global influenza preparedness.
ISSN:23148543
DOI:10.1186/s43088-025-00682-2