Maintenance Notice

Due to necessary scheduled maintenance, the JMIR Publications website will be unavailable from Wednesday, July 01, 2020 at 8:00 PM to 10:00 PM EST. We apologize in advance for any inconvenience this may cause you.

Who will be affected?

Accepted for/Published in: Journal of Medical Internet Research

Date Submitted: Nov 11, 2022
Date Accepted: Jan 10, 2023

The final, peer-reviewed published version of this preprint can be found here:

Deep-Learning Model for Influenza Prediction From Multisource Heterogeneous Data in a Megacity: Model Development and Evaluation

Yang* L, Li* G, Yang* J, Zhang T, Du J, Liu T, Zhang X, Han X, Li W, Ma# L, Feng# L, Yang# W

Deep-Learning Model for Influenza Prediction From Multisource Heterogeneous Data in a Megacity: Model Development and Evaluation

J Med Internet Res 2023;25:e44238

DOI: 10.2196/44238

PMID: 36780207

PMCID: 9972203

Establishing a new deep learning model for influenza prediction from multi-source heterogeneous data in a megacity

  • Liuyang Yang*; 
  • Gang Li*; 
  • Jin Yang*; 
  • Ting Zhang; 
  • Jing Du; 
  • Tian Liu; 
  • Xingxing Zhang; 
  • Xuan Han; 
  • Wei Li; 
  • Libing Ma#; 
  • Luzhao Feng#; 
  • Weizhong Yang#

ABSTRACT

Background:

In megacities, it is urgent to establish more sensitive forecasting and early warning methods for acute respiratory infectious diseases. The prediction and early warning model for influenza and other acute respiratory infectious diseases still needs to be improved.

Objective:

In this study, we integrated heterogeneous data from different sources in a megacity through improved deep learning models to explore more sensitive and accurate influenza prediction models, such as influenza-like illness case data from hospitals, and meteorological, search engine, and social economy data.

Methods:

We collected multisource and multidimensional data from the 26th week of 2012 to the 25th week of 2019, including influenza-like illness cases, data of virological surveillance, data of climate and demography, and data of search engines. To avoid collinearity, we had the best predictor according to the weight and correlation of each factor. We had established a new Multi-Attention-LSTM (MAL) deep learning model and predicted ILI% and the product of ILI% and influenza positive rate respectively. Finally, we compared models from various sources of data and different prediction methods.

Results:

The highest correlation coefficients were the Baidu query data for ILI% and air quality for ILI positive rate. In addition to these variables, temperature, sunlight, air pressure, humidity and wind speed were high correlation coefficients for ILI%, whereas temperature, Baidu query data, air pressure, wind speed, and humidity for positive rate. Whether it is ILI% or influenza activity, it is best to put data from various sources into the model, and R2 can reach more than 0.7. The comparison between different models showed that the new Multi-Attention-LSTM deep learning model we have established had the best prediction effect.

Conclusions:

The comparison between different models showed that the new MLA deep learning model we have established had the best prediction effect. Natural factors and search engine query data are more helpful in forecasting ILI patterns in megacities. Since a more timely and effective prediction of influenza and other respiratory infectious disease epidemic intensity, early and better preparedness can be made to reduce the health damage to the population


 Citation

Please cite as:

Yang* L, Li* G, Yang* J, Zhang T, Du J, Liu T, Zhang X, Han X, Li W, Ma# L, Feng# L, Yang# W

Deep-Learning Model for Influenza Prediction From Multisource Heterogeneous Data in a Megacity: Model Development and Evaluation

J Med Internet Res 2023;25:e44238

DOI: 10.2196/44238

PMID: 36780207

PMCID: 9972203

Download PDF


Request queued. Please wait while the file is being generated. It may take some time.

© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.