Predicting Infectious Disease Using Deep Learning and Big Data.

Area of Science:

• Public Health
• Epidemiology
• Data Science

Background:

• Infectious diseases pose significant individual and societal risks.
• Current surveillance systems face challenges with delayed and missing reports, hindering timely interventions.
• Predicting infectious disease trends is difficult with existing methods.

Purpose of the Study:

• To develop and evaluate deep learning models for predicting infectious diseases.
• To compare the performance of deep neural network (DNN) and long-short term memory (LSTM) models against ARIMA.
• To utilize big data, including social media, for improved infectious disease forecasting.

Main Methods:

• Implemented and optimized deep learning algorithms: Deep Neural Network (DNN) and Long-Short Term Memory (LSTM).
• Compared DNN and LSTM model performance with the Autoregressive Integrated Moving Average (ARIMA) model.
• Utilized big data sources, incorporating social media data, for predictive modeling.

Main Results:

• DNN and LSTM models demonstrated superior performance compared to the ARIMA model in predicting infectious diseases.
• Top-performing DNN and LSTM models showed significant performance improvements: 24% and 19% for chickenpox prediction, respectively.
• DNN models exhibited stable performance, while LSTM models were more accurate during infectious disease outbreaks.

Conclusions:

• The proposed DNN and LSTM models offer a promising approach to infectious disease prediction.
• These advanced models can mitigate reporting delays inherent in traditional surveillance systems.
• Successful implementation can lead to reduced societal costs associated with infectious disease outbreaks.