Liu Feng, Dang Caixia, Lv Hengliang, Zhao Ziqian, Zhu Sijin, Wang Yang, Song Hongbin, Xu Yuanyong, Chen Hui
Chinese People's Liberation Army Center for Disease Control and Prevention, Beijing, China.
Department of Epidemiology, School of Public Health, China Medical University, Shenyang, China.
BMC Infect Dis. 2025 Aug 14;25(1):1020. doi: 10.1186/s12879-025-11483-4.
To evaluate the application value of the ARIMA-GM(1,1) combined model in predicting the resistance rates of key drug-resistant bacteria in China, providing a scientific basis for optimizing antimicrobial management strategies.
Based on data from the China Antibacterial Resistance Surveillance Network from 2014 to 2023, we selected six types of key drug-resistant bacteria, including methicillin-resistant Staphylococcus aureus (MRSA) and cefotaxime/ceftriaxone-resistant Klebsiella pneumoniae (CTX/CRO-R-KP), to construct the ARIMA-GM(1,1) combined model. The model performance was evaluated using five metrics: Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE) and R, and the resistance rate trends for 2024-2028 were predicted.
The model exhibited strong predictive performance, with MSE, RMSE, MAE, and MAPE below 10 for all six strains, and R values exceeding 0.8 for five strains. Projected resistance rates for 2024 are as follows: MRSA at 27.46% (95% CI: 26.90%-28.02%), CTX/CRO-R-KP at 26.47% (25.81%-27.12%), CRKP at 9.87% (8.99%-10.74%), and CRPA at 15.17% (14.10%-16.24%). By 2028, these rates are expected to decrease to 24.05% (22.81%-25.29%), 22.30% (20.82%-23.77%), 7.58% (6.53%-8.63%), and 11.01% (8.61%-13.40%) (all P < 0.05). For 2024, resistance rates for CREC and CRAB are projected to be 1.63% (1.58%-1.68%) and 53.33% (51.37%-55.28%), respectively, with continued decline expected by 2028.
The ARIMA-GM(1,1) model has been statistically validated in predicting the resistance rates of MRSA, CTX/CRO-R-KP, CRKP, and CRPA, indicating a significant downward trend driven by the National Action Plan for Curbing Bacterial Drug Resistance (2016-2020). While effective in capturing temporal dynamics, future research should integrate antibiotic usage data and other influencing factors for more targeted interventions.
评估自回归积分滑动平均模型与灰色预测模型（1,1）（ARIMA-GM(1,1)）组合模型在中国主要耐药菌耐药率预测中的应用价值，为优化抗菌管理策略提供科学依据。
基于2014年至2023年中国抗菌药物耐药监测网的数据，选取六种主要耐药菌，包括耐甲氧西林金黄色葡萄球菌（MRSA）和对头孢噻肟/头孢曲松耐药的肺炎克雷伯菌（CTX/CRO-R-KP），构建ARIMA-GM(1,1)组合模型。使用均方误差（MSE）、均方根误差（RMSE）、平均绝对误差（MAE）、平均绝对百分比误差（MAPE）和R五个指标评估模型性能，并预测2024年至2028年的耐药率趋势。
该模型表现出较强的预测性能，六种菌株的MSE、RMSE、MAE和MAPE均低于10，五种菌株的R值超过0.8。2024年预测耐药率如下：MRSA为27.46%（95%置信区间：26.90%-28.02%），CTX/CRO-R-KP为26.47%（25.81%-27.12%），耐碳青霉烯类肺炎克雷伯菌（CRKP）为9.87%（8.99%-10.74%），耐碳青霉烯类鲍曼不动杆菌（CRPA）为15.17%（14.10%-16.24%）。到2028年，这些耐药率预计将降至24.05%（22.81%-25.29%）、22.30%（20.82%-23.77%）、7.58%（6.53%-8.63%）和11.01%（8.61%-13.40%）（所有P<0.05）。2024年，耐碳青霉烯类大肠埃希菌（CREC）和耐碳青霉烯类鲍曼不动杆菌（CRAB）的耐药率预计分别为1.63%（1.58%-1.68%）和&53.33%（51.37%-55.28%），预计到2028年将持续下降。
ARIMA-GM(1,1)模型在预测MRSA、CTX/CRO-R-KP、CRKP和CRPA的耐药率方面经过了统计学验证，表明在《遏制细菌耐药国家行动计划（2016-2020年）》推动下呈现出显著下降趋势。虽然该模型在捕捉时间动态方面有效，但未来研究应整合抗生素使用数据和其他影响因素，以进行更有针对性的干预。
