Related Experiment Videos
Post-pandemic forecasting of pediatric acute respiratory infections with deep learning: a multi-pathogen,
Anna Cheng1, Leijun Meng2, Jing Wang1
1Department of Emergency, Shanghai Children's Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China.
Insights
Deep learning models significantly outperform traditional methods for forecasting pediatric acute respiratory infections (ARIs). Tailoring model selection by pathogen and timeframe is key for effective public health preparedness in the post-COVID-19 era.
Area of Science:
- Epidemiology
- Infectious Diseases
- Data Science
Background:
- Acute respiratory infections (ARIs) are a major global cause of childhood illness and hospitalization.
- The post-COVID-19 era presents new challenges for forecasting ARIs due to shifting pathogen patterns.
- Accurate forecasting is crucial for effective clinical and public health preparedness.
Purpose of the Study:
- To compare the predictive performance of traditional statistical and deep learning models for pediatric ARIs.
- To enhance the accuracy of ARI forecasting for improved public health and clinical decision-making.
- To evaluate forecasting models across various pathogens, time horizons, and error metrics.
Main Methods:
- Utilized 29,260 pediatric hospitalization records (2021-2024) with pathogen screening data.
- Excluded Chlamydia pneumoniae due to low case numbers, analyzing 10 respiratory pathogens.
- Evaluated 13 time-series models (including SARIMA and 11 deep learning models) for short-, medium-, and long-term forecasting.
Main Results:
- Deep learning models showed lower error rates than SARIMA in 72.2% of comparisons.
- DLinear excelled in short-term forecasting (e.g., human bocavirus, human coronavirus); TSMixer led in long-term forecasting.
- No single model consistently outperformed others for medium-term forecasting; performance varied by pathogen and horizon.
Conclusions:
- Deep learning models offer substantial improvements for pediatric ARI pathogen surveillance in the post-pandemic era.
- Tailored model selection based on pathogen and forecasting horizon is essential for timely interventions and resource allocation.
- Enhanced forecasting supports clinical decisions, public health strategies, and vaccination planning.
Background:
Acute respiratory infections (ARIs) remain a major cause of morbidity and hospitalization in children worldwide. In the post-COVID-19 era, the epidemiological patterns of respiratory pathogens have changed considerably, posing new challenges to conventional forecasting approaches. This study aimed to compare the performance of traditional statistical and deep learning models in forecasting pediatric ARIs and to improve prediction accuracy for clinical and public health preparedness.
Methods:
A total of 29,260 pediatric hospitalization records with multiplex polymerase chain reaction screening results for 11 respiratory pathogens were collected from a large children's hospital in China between January 2021 and December 2024. Because Chlamydia pneumoniae (C. pneumoniae) had a small number of positive cases (n = 70), it was excluded before model fitting; therefore, forecasting analyses were conducted for 10 pathogens. Thirteen time-series forecasting models were evaluated, including a baseline model (NaiveSeasonal), SARIMA, and 11 deep learning models (DeepAR, BlockRNN, N-BEATS, NHiTS, TCN, Transformer, TFT, DLinear, NLinear, TiDE, and TSMixer). Forecasting performance was assessed over short-term (4-week), medium-term (12-week), and long-term (26-week) horizons using mean absolute error (MAE), root mean square error (RMSE), and symmetric mean absolute percentage error (sMAPE).
Results:
Across 990 deep-learning-versus-SARIMA comparisons (11 deep learning models x 10 pathogens x 3 forecasting horizons x 3 error metrics), deep learning models had lower error values than SARIMA in 715 comparisons (72.2%). DLinear was most frequently selected as the lowest-error model in short-term forecasting, particularly for human bocavirus and human coronavirus, whereas TSMixer was most frequently selected in long-term forecasting. For medium-term forecasting, no single model demonstrated consistent superiority. Predictive performance varied by pathogen type, forecasting horizon, and error metric.
Conclusions:
Deep learning-based forecasting models may substantially enhance surveillance of the 10 pediatric ARI pathogens included in the forecasting analyses in the post-pandemic era and support clinical and public health decision-making. Model selection should be tailored to the specific pathogen and forecasting horizon to facilitate timely intervention, optimize healthcare resource allocation, and improve vaccination planning.
Related Concept Videos
01:29Respiratory Syncytial Virus Disease
01:18Steps in Outbreak Investigation
01:21Acute Respiratory Failure-I
Definition: It is defined by specific criteria based on blood gas measurements. Hypoxemia happens when the partial pressure of oxygen (PaO2) falls below 60 mmHg. At the same time,...
01:21Acute Respiratory Failure-II
The underlying physiological abnormalities that contribute to hypoxemic respiratory failure include:
01:30Acute Respiratory Failure-III
01:29Acute Respiratory Failure-V
Ensure that patients are monitored continuously for their response to therapy, including changes in...