To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.

 
 

Description

Purpose: Influenza virus is responsible for a recurrent, yearly epidemic in most temperate regions of the world. For the 2021-2022 season the CDC reports 5000 deaths and 100,000 hospitalizations, a significant number despite the confounding presence of SARS-CoV-2. The mechanisms behind seasonal variance in flu burden are not well understood. Based on a previously validated model, this study seeks to expand understanding of the impact of variable climate regions on seasonal flu trends. To that end, three climate regions have been selected. Each region represents a different ecological region and provides different weather patterns showing how the climate variables impact flu transmission in different regions.

Methods: An LSTM-Based recurrent neural network was used to predict influenza-like-illness trends for three separate locations: Hawaii, Vermont, and Nevada. Flu data were gathered from the CDC as weekly influenza-like-illness (ILI) percents. Weather data were collected from Visual Crossing and included temperature, UV index, solar radiation, precipitation, and humidity. These weather data sets were chosen based on previous work results and a literature search. Data were prepared and the model trained as described previously.

Results: All three regions showed strong seasonality of flu trends with Hawaii having the largest absolute ILI values. Temperature showed a moderate negative correlation with ILI in all three regions (Vermont = -54, Nevada = -0.56, Hawaii = -0.44). Humidity was moderately correlated in Nevada (0.47) and weakly correlated with ILI in Hawaii (0.22). Vermont ILI did not correlate with humidity. Precipitation and wind speed were weakly correlated in all three regions. Solar radiation and UV index showed moderate correlation in Vermont (-0.33, -0.36) and Nevada (-0.5263, -0.55) however only weak correlation in Hawaii (-0.15, -0.18). When trained on the complete data set model performance at +1 week was comparable to the previously validated model.

Conclusions: Preliminary results indicate that temperature is a moderate predictor of ILI rates. Additionally, humidity, solar radiation, and UV index present promising prediction variables. Initial modeling attempts revealed acceptable performance in all regions. While seasonality appeared similar in each region, differences in correlation with weather variables may reveal variability in the driving forces behind ILI rates.

Disciplines

Epidemiology | Influenza Humans | Longitudinal Data Analysis and Time Series

Download

Share

COinS
 

LSTM-based Recurrent Neural Network Predicts Influenza-like-illness in Variable Climate Zones

Purpose: Influenza virus is responsible for a recurrent, yearly epidemic in most temperate regions of the world. For the 2021-2022 season the CDC reports 5000 deaths and 100,000 hospitalizations, a significant number despite the confounding presence of SARS-CoV-2. The mechanisms behind seasonal variance in flu burden are not well understood. Based on a previously validated model, this study seeks to expand understanding of the impact of variable climate regions on seasonal flu trends. To that end, three climate regions have been selected. Each region represents a different ecological region and provides different weather patterns showing how the climate variables impact flu transmission in different regions.

Methods: An LSTM-Based recurrent neural network was used to predict influenza-like-illness trends for three separate locations: Hawaii, Vermont, and Nevada. Flu data were gathered from the CDC as weekly influenza-like-illness (ILI) percents. Weather data were collected from Visual Crossing and included temperature, UV index, solar radiation, precipitation, and humidity. These weather data sets were chosen based on previous work results and a literature search. Data were prepared and the model trained as described previously.

Results: All three regions showed strong seasonality of flu trends with Hawaii having the largest absolute ILI values. Temperature showed a moderate negative correlation with ILI in all three regions (Vermont = -54, Nevada = -0.56, Hawaii = -0.44). Humidity was moderately correlated in Nevada (0.47) and weakly correlated with ILI in Hawaii (0.22). Vermont ILI did not correlate with humidity. Precipitation and wind speed were weakly correlated in all three regions. Solar radiation and UV index showed moderate correlation in Vermont (-0.33, -0.36) and Nevada (-0.5263, -0.55) however only weak correlation in Hawaii (-0.15, -0.18). When trained on the complete data set model performance at +1 week was comparable to the previously validated model.

Conclusions: Preliminary results indicate that temperature is a moderate predictor of ILI rates. Additionally, humidity, solar radiation, and UV index present promising prediction variables. Initial modeling attempts revealed acceptable performance in all regions. While seasonality appeared similar in each region, differences in correlation with weather variables may reveal variability in the driving forces behind ILI rates.