Forecasting antimicrobial resistance in China using a hybrid ARIMA-GM(1,1) model

Area of Science:

• Epidemiology
• Infectious Diseases
• Mathematical Modeling

Background:

• Antimicrobial resistance is a growing global health threat.
• Accurate prediction of resistance rates is crucial for effective antimicrobial stewardship.

Purpose of the Study:

• To evaluate the ARIMA-GM(1,1) combined model for predicting resistance rates of key drug-resistant bacteria in China.
• To provide a scientific basis for optimizing antimicrobial management strategies.

Main Methods:

• Utilized data from the China Antibacterial Resistance Surveillance Network (2014-2023).
• Constructed the ARIMA-GM(1,1) model using six key drug-resistant bacteria, including MRSA and CTX/CRO-R-KP.
• Evaluated model performance using MSE, RMSE, MAE, MAPE, and R²; predicted trends for 2024-2028.

Main Results:

• The ARIMA-GM(1,1) model demonstrated strong predictive performance (R² > 0.8 for five strains).
• Projected resistance rates for MRSA and CTX/CRO-R-KP in 2024 are 27.46% and 26.47%, respectively.
• Expected significant decreases in resistance rates by 2028 for all studied bacteria.

Conclusions:

• The ARIMA-GM(1,1) model is validated for predicting resistance rates of major drug-resistant bacteria.
• A significant downward trend in resistance rates is observed, linked to national action plans.
• Future research should incorporate antibiotic usage data for enhanced intervention strategies.