EUVIMEDEUVIMED entdecken
Open Access

Influenza-Like Illness Forecasting Using Multisource Data: Comparative Deep Learning Study

Caixia Dang iD, Yeqing Tong iD, Yanquan Mo iD, Ziqian Zhao iD, Shanhui Li iD, Feng Liu iD, Huiqun Jia iD, Jingya Zhao iD, Yuanyong Xu iD, Hui Chen iD

JMIR Medical Informatics · 11.08.2026 · 14 · e91844-e91844

DOI: 10.2196/91844 ↗

Kurz gefasst

Fachartikel in der Zeitschrift JMIR Medical Informatics, online erschienen am 11.08.2026. Verfasst von Caixia Dang und 9 weiteren Autor:innen. Beteiligte Einrichtung: Chinese PLA Center for Disease Control and Prevention, South Gate, Yard 20, Dongda Street, Fengtai South Road, Fengtai District, Beijing, 100071, Chi…. Hauptthemen laut MEDLINE: Deep Learning, Influenza, Human. Die Arbeit nennt 42 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Accurate forecasting of influenza-like illness (ILI) is crucial for public health. Integrating novel digital data streams (eg, internet searches and human mobility) with traditional surveillance can improve accuracy, but optimal modeling frameworks are underexplored.

Objective

This study aimed to develop and compare multisource data-driven models for forecasting ILI incidence trends.

Methods

Weekly ILI incidence data and multisource variables for Hubei Province from 2020 to 2023 were collected. Multisource predictors included mean temperature, relative humidity, air quality index, a synthesized Baidu Search Index for influenza-related queries, the Baidu Migration Scale Index for population mobility, and the Oxford Stringency Index (SI) for nonpharmaceutical interventions. Predictive models evaluated were Seasonal Autoregressive Integrated Moving Average (SARIMA), long short-term memory (LSTM), a hybrid convolutional neural network–long short-term memory (CNN-LSTM), Transformer, and random forest. Models were constructed using an 85:15 training-test split and optimized via grid search with 5-fold cross-validation. Performance was assessed using mean absolute error, root-mean-squared error (RMSE), mean absolute percentage error, and coefficient of determination ( R ²).

Results

All models incorporating multisource data substantially outperformed univariate time-series benchmarks. Among univariate models, CNN-LSTM achieved superior performance ( R ²=0.7623) over SARIMA ( R ²=−0.2645). In multisource configurations, the LSTM model demonstrated the strongest predictive capability and feature integration, attaining an optimal R ² of 0.8350 when combining environmental, mobility (Baidu Migration Scale), and internet search (synthetic Baidu index [SBI]) data. The inclusion of the Oxford SI consistently reduced prediction error across all models, with mean absolute percentage error decreasing by up to 84.87% in the LSTM model. The Baidu Search Index emerged as the most influential single external predictor, notably enhancing model fit. In contrast, model performance varied with feature composition: the Transformer model excelled with full feature sets, while CNN-LSTM performed best primarily with SBI integration. Over a 26-week prospective projection, the optimal LSTM model forecasted a “rapid decline–gradual decline–stabilization” trend, indicating a return to baseline ILI activity, and demonstrated robust external validation performance ( R 2 =0.7707).

Conclusions

Deep learning models, particularly LSTM, can effectively leverage heterogeneous digital data to improve ILI forecasting. Internet search behavior and policy stringency are critical external predictors. The proposed model demonstrated good external validation performance, and the findings may be applicable to other temperate regions in central China exhibiting similar epidemic characteristics.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningInfluenza, HumanForecastingHumansChinaLong Short Term MemoryPredictive Learning ModelsPrediction AlgorithmsIncidence

Stichwörter der Autor:innen

Influenza-Like Illness forecastingLong Short-Term Memorydeep learninginfluenza-like illnessmultisource data

Autor:innen und Institutionen

  • Chinese PLA Center for Disease Control and Prevention, South Gate, Yard 20, Dongda Street, Fengtai South Road, Fengtai District, Beijing, 100071, China, 86 18149338893.
  • School of Public Health, China Medical University, Shenyang, Liaoning, China.
  • Hubei Center for Disease Control and Prevention, Wuhan, China.

ORCID-Profile: Caixia Dang, Yeqing Tong, Yanquan Mo, Ziqian Zhao, Shanhui Li, Feng Liu, Huiqun Jia, Jingya Zhao, Yuanyong Xu, Hui Chen

Zitieren

APA

Dang, C., Tong, Y., Mo, Y., Zhao, Z., Li, S., Liu, F., Jia, H., Zhao, J., Xu, Y., & Chen, H. (2026). Influenza-Like Illness Forecasting Using Multisource Data: Comparative Deep Learning Study. JMIR Medical Informatics, 14, e91844-e91844. https://doi.org/10.2196/91844

Vancouver

Dang C, Tong Y, Mo Y, Zhao Z, Li S, Liu F, et al. Influenza-Like Illness Forecasting Using Multisource Data: Comparative Deep Learning Study. JMIR Medical Informatics. 2026;14:e91844-e91844. doi:10.2196/91844

BibTeX
@article{Dang2026,
  author = {Dang, Caixia and Tong, Yeqing and Mo, Yanquan and Zhao, Ziqian and Li, Shanhui and Liu, Feng and Jia, Huiqun and Zhao, Jingya and Xu, Yuanyong and Chen, Hui},
  title = {Influenza-Like Illness Forecasting Using Multisource Data: Comparative Deep Learning Study},
  journal = {JMIR Medical Informatics},
  year = {2026},
  volume = {14},
  pages = {e91844--e91844},
  doi = {10.2196/91844}
}
Kostenlos starten →