Hybrid deep learning model predicts monthly hepatitis B incidence in China

Researchers have developed a hybrid deep learning model that combines long short-term memory networks with convolutional neural networks to forecast the monthly incidence rate of hepatitis B in China, reporting that the combined approach outperformed traditional time series models and individual machine learning methods.

The study, published in BMC Public Health, analyzed monthly hepatitis B incidence data from China covering January 2001 to February 2026. The authors set out to build an optimal prediction model for hepatitis B incidence in China and to provide accurate predictions that could support early monitoring and early warning of infectious diseases.

According to the results, the average monthly incidence rate of hepatitis B in China was approximately 0.7 per 10,000 over the study period. The researchers also report that the prevalence and incidence rate of hepatitis B in the world and in China are decreasing year by year, and that China’s incidence is higher than the global average. The mortality rate among people aged 55 and above in China was reported to be lower than the global average.

To model the data, the team established a series of time series and machine learning models. The comparison set included traditional approaches such as exponential smoothing, ARIMA and support vector regression, alongside individual machine learning models including convolutional neural networks, gated recurrent units and long short-term memory networks.

The hybrid LSTM + CNN model achieved the strongest performance, with a coefficient of determination of 0.603. The authors state that this prediction performance was superior to the traditional time series models and the individual machine learning models tested. They also note that all of the time series models were able to capture the trend of the hepatitis B incidence rate.

In their conclusions, the researchers write that the LSTM + CNN model demonstrates competitive performance in predicting the monthly average incidence rate of hepatitis B in China. They add that the hybrid deep learning model has shown great potential in time series prediction, although they emphasize that further validation is needed.

The authors describe the work as a methodological exploration. They state that the study explores effective methods for predicting infectious disease time series and provides a new perspective for infectious disease prediction studies. The reported performance figure reflects the model’s fit within this study’s data and should be interpreted with the authors’ own caveat about the need for further validation.

Hepatitis B remains an important public health issue facing China, according to the study’s background, and is described as one of the most severe infectious diseases in the world. The authors frame reliable incidence forecasting as a tool for early monitoring and early warning of infectious diseases, which is the stated motivation for comparing multiple modeling approaches in a single systematic analysis.

The article was published as open access research on 08 October 2026, with Minquan Guan, Zhouying Hu, Yichen Lin, Xuehui Meng and Chengcheng Li among the listed authors. The study is presented as a methodological contribution to infectious disease modeling rather than a clinical intervention, and its findings concern statistical prediction of reported incidence rates rather than treatment or prevention outcomes.

Subject of Research: Time series and machine learning prediction of monthly hepatitis B incidence rates in China

Article Title: A methodological exploration of monthly incidence rate of hepatitis B in China: a time series and machine learning analysis

Article References: Guan, M., Hu, Z., Lin, Y., Meng, X., & Li, C. (2026). A methodological exploration of monthly incidence rate of hepatitis B in China: a time series and machine learning analysis. BMC Public Health. https://doi.org/10.1186/s12889-026-29683-1

Image Credits: AI Generated

DOI: 10.1186/s12889-026-29683-1

Keywords: hepatitis B, China, incidence rate, time series analysis, machine learning, LSTM, CNN, deep learning, infectious disease forecasting, public health, ARIMA, epidemiology