Epidemics最新文献_第3页

Enhancing Influenza-Like Illness forecasting: An ensemble approach combining mathematical and deep learning models amidst the COVID-19 pandemic 加强流感样疾病预测：2019冠状病毒病大流行期间结合数学和深度学习模型的集成方法

IF 2.4 3区医学 Q2 INFECTIOUS DISEASES

Epidemics

Pub Date : 2026-03-01 Epub Date: 2026-02-27 DOI: 10.1016/j.epidem.2026.100901

Ganghyun Yoon , Amanda Bleichrodt , Gerardo Chowell , Sunmi Lee

Background:

Timely and accurate short-term forecasting of Influenza-Like Illness (ILI) is crucial for guiding outbreak response, optimizing healthcare resource allocation, and informing public health interventions. The COVID-19 pandemic, which disrupted seasonal ILI dynamics due to widespread nonpharmaceutical interventions (NPI), underscored the urgent need for adaptive and reliable forecasting frameworks.

Method:

In this study, we present a novel ensemble modeling approach that combines a mechanistic n-subepidemic model with a Monte Carlo Dropout Long Short-Term Memory (LSTM) neural network to improve age-specific ILI forecasting performance in South Korea. By capturing both the structured dynamics of disease spread and nonlinear temporal dependencies, our ensemble method adapts to pandemic-altered transmission patterns while offering robust uncertainty quantification. Age-stratified forecasting allows the framework to capture heterogeneity in vulnerability and transmission across demographic groups, providing more targeted insights for policy and planning.

Result:

We evaluated forecasting performance across four epidemic waves using the Weighted Interval Score (WIS), Mean Absolute Error (MAE), consistently finding that the ensemble models outperformed individual approaches.

Conclusion:

These findings underscore the power of hybrid forecasting approaches to improve epidemic preparedness and response, providing a flexible data-driven framework that can evolve with changing transmission dynamics and extend to other emerging infectious threats.

背景：及时准确的流感样疾病（ILI）短期预测对于指导疫情应对、优化卫生资源配置和告知公共卫生干预措施至关重要。COVID-19大流行由于广泛的非药物干预措施（NPI）而扰乱了季节性流感动态，这凸显了迫切需要适应性和可靠的预测框架。方法：在本研究中，我们提出了一种新的集成建模方法，该方法将机制n-亚流行病模型与蒙特卡罗辍学长短期记忆（LSTM）神经网络相结合，以提高韩国特定年龄的ILI预测性能。通过捕获疾病传播的结构化动态和非线性时间依赖性，我们的集成方法适应大流行改变的传播模式，同时提供稳健的不确定性量化。年龄分层预测使该框架能够捕捉脆弱性和跨人口群体传播的异质性，为政策和规划提供更有针对性的见解。结果：我们使用加权区间分数（WIS）和平均绝对误差（MAE）评估了四种流行病波的预测性能，一致发现集成模型优于单个方法。结论：这些发现强调了混合预测方法在改善流行病防范和应对方面的力量，提供了一个灵活的数据驱动框架，可以随着传播动态的变化而发展，并扩展到其他新出现的传染性威胁。

{"title":"Enhancing Influenza-Like Illness forecasting: An ensemble approach combining mathematical and deep learning models amidst the COVID-19 pandemic","authors":"Ganghyun Yoon , Amanda Bleichrodt , Gerardo Chowell , Sunmi Lee","doi":"10.1016/j.epidem.2026.100901","DOIUrl":"10.1016/j.epidem.2026.100901","url":null,"abstract":"<div><h3>Background:</h3><div>Timely and accurate short-term forecasting of Influenza-Like Illness (ILI) is crucial for guiding outbreak response, optimizing healthcare resource allocation, and informing public health interventions. The COVID-19 pandemic, which disrupted seasonal ILI dynamics due to widespread nonpharmaceutical interventions (NPI), underscored the urgent need for adaptive and reliable forecasting frameworks.</div></div><div><h3>Method:</h3><div>In this study, we present a novel ensemble modeling approach that combines a mechanistic n-subepidemic model with a Monte Carlo Dropout Long Short-Term Memory (LSTM) neural network to improve age-specific ILI forecasting performance in South Korea. By capturing both the structured dynamics of disease spread and nonlinear temporal dependencies, our ensemble method adapts to pandemic-altered transmission patterns while offering robust uncertainty quantification. Age-stratified forecasting allows the framework to capture heterogeneity in vulnerability and transmission across demographic groups, providing more targeted insights for policy and planning.</div></div><div><h3>Result:</h3><div>We evaluated forecasting performance across four epidemic waves using the Weighted Interval Score (WIS), Mean Absolute Error (MAE), consistently finding that the ensemble models outperformed individual approaches.</div></div><div><h3>Conclusion:</h3><div>These findings underscore the power of hybrid forecasting approaches to improve epidemic preparedness and response, providing a flexible data-driven framework that can evolve with changing transmission dynamics and extend to other emerging infectious threats.</div></div>","PeriodicalId":49206,"journal":{"name":"Epidemics","volume":"54 ","pages":"Article 100901"},"PeriodicalIF":2.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147348617","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Maternal immunity drives age-related patterns of RSV disease 母亲免疫驱动年龄相关的RSV疾病模式。

IF 2.4 3区医学 Q2 INFECTIOUS DISEASES

Epidemics

Pub Date : 2026-03-01 Epub Date: 2026-02-28 DOI: 10.1016/j.epidem.2026.100903

Clara Brigitta , David Hodgson , Graham F. Medley , Rosalind M. Eggo

Respiratory Syncytial Virus (RSV) is a leading cause of respiratory illness in young children. While maternal antibodies offer temporary protection in early infancy, their interaction with age-dependent disease risk remains poorly quantified. The COVID-19 pandemic, which disrupted RSV transmission, provides a unique opportunity to explore these dynamics and the potential impact of maternal vaccination. We developed a compartmental model of childhood RSV disease incorporating maternal infection history, maternally-derived immunity, waning protection in infants, and age-dependent disease risk. Calibrated to Scottish surveillance data (2016–2024), the model estimated non-linear functions for maternal immunity and RSV risk by age, and projected burden from 2024 to 2028 under vaccination and no-vaccination scenarios. Following pandemic-related disruption, RSV burden shifted to older children due to delayed primary exposure. Infants who missed their typical first RSV season in 2020 experienced higher disease rates at later ages, in subsequent seasons. Maternal immunity conferred protection only when infection occurred in late pregnancy, with infant protection waning to negligible levels by six months of age. Maternal vaccination at current coverage rates was projected to reduce RSV disease cases in infants ≤ 6 months by 15.3% (95% CI: 11.9–18.5%) in 2024–25 and 18.4% (95% CI: 12.8–23.6%) in 2025–26. Our findings highlight the role of the timing and immunological mechanisms of maternally-derived immunity in shaping RSV dynamics in young children and demonstrate how disruptions—whether through pandemic-related changes or maternal vaccination—can alter age-specific disease patterns.

呼吸道合胞病毒（RSV）是幼儿呼吸道疾病的主要原因。虽然母体抗体在婴儿早期提供暂时的保护，但它们与年龄依赖性疾病风险的相互作用仍然缺乏量化。COVID-19大流行阻断了RSV传播，为探索这些动态和孕产妇接种疫苗的潜在影响提供了独特的机会。我们建立了一个儿童呼吸道合胞病毒疾病的区室模型，包括母亲感染史、母亲来源的免疫力、婴儿的保护减弱和年龄相关的疾病风险。根据苏格兰监测数据（2016-2024年）校准，该模型估计了按年龄划分的孕产妇免疫和RSV风险的非线性函数，并在接种疫苗和不接种疫苗的情况下预测了2024年至2028年的负担。在与大流行相关的破坏之后，由于初次接触延迟，呼吸道合流病毒负担转移到年龄较大的儿童身上。在2020年错过典型的第一个RSV季节的婴儿在随后的季节中年龄较大，发病率较高。只有当感染发生在妊娠后期时，母体免疫才能提供保护，婴儿的保护在6个月大时减弱到可以忽略不计的水平。按照目前的覆盖率，预计2024-25年≤ 6个月婴儿的RSV疾病病例减少15.3% (95% CI: 11.9-18.5%)， 2025-26年减少18.4% （95% CI: 12.8-23.6%）。我们的研究结果强调了母源免疫的时间和免疫机制在塑造幼儿RSV动力学中的作用，并证明了无论是通过大流行相关的变化还是通过母亲接种疫苗，破坏是如何改变年龄特异性疾病模式的。

{"title":"Maternal immunity drives age-related patterns of RSV disease","authors":"Clara Brigitta , David Hodgson , Graham F. Medley , Rosalind M. Eggo","doi":"10.1016/j.epidem.2026.100903","DOIUrl":"10.1016/j.epidem.2026.100903","url":null,"abstract":"<div><div>Respiratory Syncytial Virus (RSV) is a leading cause of respiratory illness in young children. While maternal antibodies offer temporary protection in early infancy, their interaction with age-dependent disease risk remains poorly quantified. The COVID-19 pandemic, which disrupted RSV transmission, provides a unique opportunity to explore these dynamics and the potential impact of maternal vaccination. We developed a compartmental model of childhood RSV disease incorporating maternal infection history, maternally-derived immunity, waning protection in infants, and age-dependent disease risk. Calibrated to Scottish surveillance data (2016–2024), the model estimated non-linear functions for maternal immunity and RSV risk by age, and projected burden from 2024 to 2028 under vaccination and no-vaccination scenarios. Following pandemic-related disruption, RSV burden shifted to older children due to delayed primary exposure. Infants who missed their typical first RSV season in 2020 experienced higher disease rates at later ages, in subsequent seasons. Maternal immunity conferred protection only when infection occurred in late pregnancy, with infant protection waning to negligible levels by six months of age. Maternal vaccination at current coverage rates was projected to reduce RSV disease cases in infants ≤ 6 months by 15.3% (95% CI: 11.9–18.5%) in 2024–25 and 18.4% (95% CI: 12.8–23.6%) in 2025–26. Our findings highlight the role of the timing and immunological mechanisms of maternally-derived immunity in shaping RSV dynamics in young children and demonstrate how disruptions—whether through pandemic-related changes or maternal vaccination—can alter age-specific disease patterns.</div></div>","PeriodicalId":49206,"journal":{"name":"Epidemics","volume":"54 ","pages":"Article 100903"},"PeriodicalIF":2.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147373234","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Spatiotemporal transmission of influenza in the US during the 2022/23 season 美国2022/23流感季节的时空传播

IF 2.4 3区医学 Q2 INFECTIOUS DISEASES

Epidemics

Pub Date : 2026-03-01 Epub Date: 2026-02-14 DOI: 10.1016/j.epidem.2026.100896

Alexia Couture , Matthew Biggerstaff , Michael Sheppard , Alicia Budd , Aaron Kite-Powell , Sinead E. Morris

Understanding the spatiotemporal dynamics of seasonal influenza spread across the United States (US) is crucial for informed public health planning. We explored patterns of influenza transmission during the 2022/23 season in the US and used a mathematical model to infer potential drivers and underlying mechanisms. Leveraging emergency department visit data, we first estimated the timing of influenza onset for the 2022/23 season at the Health Service Area (HSA) level. We then combined the estimated onset times in a gravity-based mechanistic model with covariates that could be associated with influenza spread, including demographics, climate, mobility, and school opening information. We compared multiple models to find the best fit to the onset times, infer factors driving transmission, and identify potential geographic hubs that were most influential in generating early chains of transmission. From the estimated onset times, we found a spatiotemporal pattern that was characterized by early transmission in southern and southeastern HSAs, followed by localized spread to other regions. The best-fit model included absolute humidity and local transmission modulated by school opening times, with four out of five potential hubs located in the southern US (two in Georgia and one each in North Carolina and Texas) and one in the Northwest (Washington). In conclusion, we found a regional pattern for the spatiotemporal spread of 2022/23 seasonal influenza and identified potential key drivers of this pattern. These findings, and similar studies from other influenza seasons, may improve our understanding of the spatiotemporal spread of influenza in the US.

了解季节性流感在美国传播的时空动态对于知情的公共卫生规划至关重要。我们探索了美国2022/23年流感季节的传播模式，并使用数学模型来推断潜在的驱动因素和潜在机制。利用急诊就诊数据，我们首先在卫生服务区（HSA）层面估计了2022/23年流感发病时间。然后，我们将基于重力的机制模型中估计的发病时间与可能与流感传播相关的协变量（包括人口统计、气候、流动性和学校开放信息）结合起来。我们比较了多个模型，以找到最适合的发病时间，推断驱动传播的因素，并确定对产生早期传播链最具影响力的潜在地理中心。从估计的发病时间来看，我们发现了南部和东南部HSAs早期传播的时空格局，随后向其他地区局部传播。最合适的模型包括绝对湿度和根据学校开学时间调制的本地传播，五个潜在中心中有四个位于美国南部（佐治亚州两个，北卡罗来纳州和德克萨斯州各一个），一个位于西北部（华盛顿州）。综上所述，我们发现了2022/23年季节性流感时空传播的区域格局，并确定了这种格局的潜在关键驱动因素。这些发现，以及其他流感季节的类似研究，可能会提高我们对美国流感时空传播的理解。

{"title":"Spatiotemporal transmission of influenza in the US during the 2022/23 season","authors":"Alexia Couture , Matthew Biggerstaff , Michael Sheppard , Alicia Budd , Aaron Kite-Powell , Sinead E. Morris","doi":"10.1016/j.epidem.2026.100896","DOIUrl":"10.1016/j.epidem.2026.100896","url":null,"abstract":"<div><div>Understanding the spatiotemporal dynamics of seasonal influenza spread across the United States (US) is crucial for informed public health planning. We explored patterns of influenza transmission during the 2022/23 season in the US and used a mathematical model to infer potential drivers and underlying mechanisms. Leveraging emergency department visit data, we first estimated the timing of influenza onset for the 2022/23 season at the Health Service Area (HSA) level. We then combined the estimated onset times in a gravity-based mechanistic model with covariates that could be associated with influenza spread, including demographics, climate, mobility, and school opening information. We compared multiple models to find the best fit to the onset times, infer factors driving transmission, and identify potential geographic hubs that were most influential in generating early chains of transmission. From the estimated onset times, we found a spatiotemporal pattern that was characterized by early transmission in southern and southeastern HSAs, followed by localized spread to other regions. The best-fit model included absolute humidity and local transmission modulated by school opening times, with four out of five potential hubs located in the southern US (two in Georgia and one each in North Carolina and Texas) and one in the Northwest (Washington). In conclusion, we found a regional pattern for the spatiotemporal spread of 2022/23 seasonal influenza and identified potential key drivers of this pattern. These findings, and similar studies from other influenza seasons, may improve our understanding of the spatiotemporal spread of influenza in the US.</div></div>","PeriodicalId":49206,"journal":{"name":"Epidemics","volume":"54 ","pages":"Article 100896"},"PeriodicalIF":2.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146259889","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Combining an agent-based model with Gaussian process emulation to model the emergence of the SARS-CoV-2 Omicron variant in Norway 结合基于主体的模型和高斯过程仿真来模拟挪威SARS-CoV-2 Omicron变体的出现。

IF 2.4 3区医学 Q2 INFECTIOUS DISEASES

Epidemics

Pub Date : 2026-03-01 Epub Date: 2026-02-17 DOI: 10.1016/j.epidem.2026.100895

Alfonso Diz-Lois Palomares , Birgitte Freiesleben de Blasio , Louis Yat Hin Chan , Francesco Di Ruscio , Jørgen E. Midtbø

Agent-based models (ABMs) are powerful for simulating detailed infectious disease dynamics, but they are computationally intensive, limiting parameter inference. We present a Bayesian history matching approach with Gaussian process emulation to efficiently calibrate an ABM of SARS-CoV-2 transmission in Norway, enabling inference of numerous parameters and reducing computational costs.

We apply this technique to model the emergence and subsequent winter wave of the Omicron BA.1/BA.2 variants in Norway from September 2021 to April 2022. By fitting age-specific hospitalisation data, we infer 15 key parameters, including age-specific susceptibility, variant transmissibility, and the infection-hospitalisation ratio. The calibrated model accurately reproduces epidemic curves and provides insights into Omicron’s spread.

We estimate that around 60% of the Norwegian population was infected with Omicron BA.1/BA.2 during the 2021/2022 winter wave. The model suggests the variant circulated in Norway before its detection in South Africa in late November 2021. Our findings indicate that Omicron’s transmission advantage over Delta was primarily due to immune evasion rather than a substantial increase in intrinsic transmissibility. Counterfactual scenarios reveal the critical role of the November 2021 booster vaccine rollout in mitigating the ensuing winter wave, indicating near-optimal timing for balancing immunisation and waning. Our estimates also suggest that the December 2021 interventions had limited impact beyond altering general mobility.

This study demonstrates the power of combining ABMs with advanced calibration to derive detailed epidemiological insights from limited data. Efficiently fitting parameters enables comprehensive epidemic modelling. Emulator-based calibration greatly enhances ABMs’ utility for retrospective analysis and real-time decision support during outbreaks, providing a valuable tool for public health planning and response.

基于智能体的模型（ABMs）在模拟详细的传染病动力学方面功能强大，但它们计算量大，限制了参数推理。我们提出了一种基于高斯过程仿真的贝叶斯历史匹配方法，以有效校准挪威SARS-CoV-2传播的ABM，实现了众多参数的推断并降低了计算成本。我们应用该技术模拟了Omicron BA.1/BA的出现和随后的冬季波。从2021年9月到2022年4月在挪威有2个改型。通过拟合年龄特异性住院数据，我们推断出15个关键参数，包括年龄特异性易感性、变异传播率和感染住院率。经过校准的模型准确地再现了流行病曲线，并为欧米克隆的传播提供了见解。我们估计约60%的挪威人感染了欧米克隆BA.1/BA。2在2021/2022年冬季浪潮。该模型表明，在2021年11月下旬在南非被发现之前，该变体在挪威传播。我们的研究结果表明，Omicron相对于Delta的传播优势主要是由于免疫规避，而不是内在传播率的大幅增加。反事实情景揭示了2021年11月推出的加强疫苗在缓解随后的冬季疫情方面的关键作用，表明了平衡免疫接种和减弱的接近最佳时机。我们的估计还表明，2021年12月的干预措施除了改变一般流动性外，影响有限。这项研究表明，将ABMs与先进的校准相结合，可以从有限的数据中获得详细的流行病学见解。有效地拟合参数使全面的流行病建模成为可能。基于模拟器的校准极大地增强了ABMs在疫情爆发期间的回顾性分析和实时决策支持的效用，为公共卫生规划和应对提供了有价值的工具。

{"title":"Combining an agent-based model with Gaussian process emulation to model the emergence of the SARS-CoV-2 Omicron variant in Norway","authors":"Alfonso Diz-Lois Palomares , Birgitte Freiesleben de Blasio , Louis Yat Hin Chan , Francesco Di Ruscio , Jørgen E. Midtbø","doi":"10.1016/j.epidem.2026.100895","DOIUrl":"10.1016/j.epidem.2026.100895","url":null,"abstract":"<div><div>Agent-based models (ABMs) are powerful for simulating detailed infectious disease dynamics, but they are computationally intensive, limiting parameter inference. We present a Bayesian history matching approach with Gaussian process emulation to efficiently calibrate an ABM of SARS-CoV-2 transmission in Norway, enabling inference of numerous parameters and reducing computational costs.</div><div>We apply this technique to model the emergence and subsequent winter wave of the Omicron BA.1/BA.2 variants in Norway from September 2021 to April 2022. By fitting age-specific hospitalisation data, we infer 15 key parameters, including age-specific susceptibility, variant transmissibility, and the infection-hospitalisation ratio. The calibrated model accurately reproduces epidemic curves and provides insights into Omicron’s spread.</div><div>We estimate that around 60% of the Norwegian population was infected with Omicron BA.1/BA.2 during the 2021/2022 winter wave. The model suggests the variant circulated in Norway before its detection in South Africa in late November 2021. Our findings indicate that Omicron’s transmission advantage over Delta was primarily due to immune evasion rather than a substantial increase in intrinsic transmissibility. Counterfactual scenarios reveal the critical role of the November 2021 booster vaccine rollout in mitigating the ensuing winter wave, indicating near-optimal timing for balancing immunisation and waning. Our estimates also suggest that the December 2021 interventions had limited impact beyond altering general mobility.</div><div>This study demonstrates the power of combining ABMs with advanced calibration to derive detailed epidemiological insights from limited data. Efficiently fitting parameters enables comprehensive epidemic modelling. Emulator-based calibration greatly enhances ABMs’ utility for retrospective analysis and real-time decision support during outbreaks, providing a valuable tool for public health planning and response.</div></div>","PeriodicalId":49206,"journal":{"name":"Epidemics","volume":"54 ","pages":"Article 100895"},"PeriodicalIF":2.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147372681","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A robust compartmental modeling framework for infectious disease monitoring and analysis via fractional differential equations 通过分数阶微分方程进行传染病监测和分析的鲁棒分区建模框架。

IF 2.4 3区医学 Q2 INFECTIOUS DISEASES

Epidemics

Pub Date : 2026-03-01 Epub Date: 2026-01-13 DOI: 10.1016/j.epidem.2026.100887

Farrukh A. Chishtie , John Drozd , X. Li , A. Benterki , Sree R. Valluri

This study presents a comprehensive framework for infectious disease monitoring using fractional differential equations, specifically developing the SEIQRDP (Susceptible, Exposed, Infected, Quarantined, Recovered, Deceased, Protected) model. Traditional compartmental models are extended by incorporating fractional calculus, with orders

α \in (0, 2]

, which provides enhanced flexibility in capturing memory effects and non-local behaviors inherent in disease transmission dynamics. The framework demonstrates improved accuracy when fitted to Canadian COVID-19 data compared to classical integer-order models, with Wave 1 achieving 22.1% improvement (95% CI: 17.4–26.8%) and Wave 2 achieving 6.2% improvement (95% CI: 3.1–9.3%) in predictive accuracy (average

\sim

14%). Fractional orders both below and above unity yield superior fits to empirical data depending on epidemic phase, successfully capturing multi-wave dynamics across different pandemic phases. The model incorporates time-dependent parameters to account for varying intervention strategies. Rigorous mathematical analysis including existence, uniqueness, and stability of solutions is provided alongside comprehensive sensitivity analysis. Out-of-sample validation using rolling-origin cross-validation demonstrates robust forecasting performance across 7-, 14-, and 21-day horizons. This research provides public health authorities with an evidence-based tool for epidemic modeling, with proposed extensions for AI-enhanced surveillance, interoperability standards, and Long COVID monitoring discussed as future research directions.

本研究提出了一个使用分数阶微分方程进行传染病监测的综合框架，特别是开发了SEIQRDP（易感、暴露、感染、隔离、恢复、死亡、保护）模型。传统的区室模型通过加入分数阶微积分得到扩展，阶数α∈(0,2)，从而增强了捕捉记忆效应和疾病传播动力学中固有的非局部行为的灵活性。与经典整阶模型相比，该框架在拟合加拿大COVID-19数据时显示出更高的准确性，其中Wave 1在预测准确性方面提高了22.1% (95% CI: 17.4-26.8%)， Wave 2在预测准确性方面提高了6.2% (95% CI: 3.1-9.3%)（平均~ 14%）。单位以下和单位以上的分数阶对经验数据的拟合都优于单位，这取决于流行阶段，成功地捕获了不同流行阶段的多波动力学。该模型纳入了与时间相关的参数，以考虑不同的干预策略。对解的存在性、唯一性和稳定性进行了严格的数学分析，并进行了全面的灵敏度分析。使用滚动原点交叉验证的样本外验证在7天、14天和21天的范围内展示了稳健的预测性能。这项研究为公共卫生当局提供了一种基于证据的流行病建模工具，并提出了人工智能增强监测、互操作性标准和长期COVID监测的扩展建议，作为未来的研究方向。

{"title":"A robust compartmental modeling framework for infectious disease monitoring and analysis via fractional differential equations","authors":"Farrukh A. Chishtie , John Drozd , X. Li , A. Benterki , Sree R. Valluri","doi":"10.1016/j.epidem.2026.100887","DOIUrl":"10.1016/j.epidem.2026.100887","url":null,"abstract":"<div><div>This study presents a comprehensive framework for infectious disease monitoring using fractional differential equations, specifically developing the SEIQRDP (Susceptible, Exposed, Infected, Quarantined, Recovered, Deceased, Protected) model. Traditional compartmental models are extended by incorporating fractional calculus, with orders <span><math><mrow><mi>α</mi><mo>∈</mo><mrow><mo>(</mo><mn>0</mn><mo>,</mo><mn>2</mn><mo>]</mo></mrow></mrow></math></span>, which provides enhanced flexibility in capturing memory effects and non-local behaviors inherent in disease transmission dynamics. The framework demonstrates improved accuracy when fitted to Canadian COVID-19 data compared to classical integer-order models, with Wave 1 achieving 22.1% improvement (95% CI: 17.4–26.8%) and Wave 2 achieving 6.2% improvement (95% CI: 3.1–9.3%) in predictive accuracy (average <span><math><mo>∼</mo></math></span>14%). Fractional orders both below and above unity yield superior fits to empirical data depending on epidemic phase, successfully capturing multi-wave dynamics across different pandemic phases. The model incorporates time-dependent parameters to account for varying intervention strategies. Rigorous mathematical analysis including existence, uniqueness, and stability of solutions is provided alongside comprehensive sensitivity analysis. Out-of-sample validation using rolling-origin cross-validation demonstrates robust forecasting performance across 7-, 14-, and 21-day horizons. This research provides public health authorities with an evidence-based tool for epidemic modeling, with proposed extensions for AI-enhanced surveillance, interoperability standards, and Long COVID monitoring discussed as future research directions.</div></div>","PeriodicalId":49206,"journal":{"name":"Epidemics","volume":"54 ","pages":"Article 100887"},"PeriodicalIF":2.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146020351","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

UnMuted: Defining SARS-CoV-2 lineages according to temporally consistent mutation clusters in wastewater samples 根据废水样本中暂时一致的突变簇来定义SARS-CoV-2谱系

IF 2.4 3区医学 Q2 INFECTIOUS DISEASES

Epidemics

Pub Date : 2026-03-01 Epub Date: 2025-12-16 DOI: 10.1016/j.epidem.2025.100876

Devan Becker

SARS-CoV-2 lineages are defined according to placement in a phylogenetic tree, but approximated by a list of mutations based on sequences collected from clinical sampling. Wastewater lineage abundance is generally found under the assumption that the mutation frequency is approximately equal to the sum of the abundances of the lineages to which it belongs. By leveraging numerous samples collected over time, I am able to estimate the temporal trends of the abundance of lineages as well as the definitions of those lineages. This is accomplished by assuming that collections of mutations that appear together over time can be used to define lineages.

Three main models are considered: One that does not imposes a temporal structure, one that includes an explicit temporal component but allows for missing lineages, and one with an explicit temporal component that attempts to estimate all lineages. It is found that the temporal trend of estimated lineage definitions approximately corresponds to the trend of lineage definitions determined by clinical samples, despite having no information from clinical samples.

SARS-CoV-2谱系是根据系统发育树中的位置来定义的，但通过基于从临床抽样中收集的序列的突变列表来近似定义。废水谱系丰度通常是在假设突变频率近似等于其所属谱系丰度之和的情况下发现的。通过利用随时间收集的大量样本，我能够估计谱系丰度的时间趋势，以及这些谱系的定义。这是通过假设随时间一起出现的突变集合可以用来定义谱系来实现的。考虑了三种主要的模型：一种不强加时间结构，一种包括显式时间成分但允许缺失的血统，一种具有显式时间成分，试图估计所有血统。研究发现，尽管没有来自临床样本的信息，但估计的谱系定义的时间趋势大致对应于临床样本确定的谱系定义的趋势。

引用次数: 0

Insights from a Ventilation-Aware Pandemic and Outbreak Risk model (VAPOR) 来自通风感知的流行病和爆发风险模型的见解（蒸气）

IF 2.4 3区医学 Q2 INFECTIOUS DISEASES

Epidemics

Pub Date : 2026-03-01 Epub Date: 2025-12-09 DOI: 10.1016/j.epidem.2025.100878

Natalie J. Wilson , Callandra Moore , Clara Eunyoung Lee , Ashleigh R. Tuite , David N. Fisman

Transmission of airborne pathogens in indoor spaces is strongly modulated by heterogeneity in ventilation. Understanding the role indoor air plays in pandemic risk is limited in part due to differing modeling approaches used in engineering and epidemiology. Here we present the VAPOR (Ventilation-Aware Pandemic and Outbreak Risk) model, a hybrid transmission framework that integrates Reed-Frost close-contact dynamics with Wells-Riley aerosol-mediated risk. Using a meta-population structure to simulate multi-patch environments (e.g., separate workplaces or schools), we explore how ventilation disparities shape epidemic potential. A fixed minority of individuals are modeled as “aerosolizers,” consistent with overdispersed real-world transmission patterns (e.g., SARS-CoV-2). Simulations reveal that both improving ventilation in high-risk patches and raising baseline ventilation across environments independently reduces risk. Parameter sweeps across air changes per hour (ACH, 2–12) demonstrate non-linear benefits with early saturation. These findings emphasize the need for targeted ventilation strategies and show how small-world effects amplify heterogeneity-driven transmission. VAPOR offers a framework for linking ventilation equity to epidemic control.

室内空间中空气传播的病原体受到通风不均一性的强烈调节。对室内空气在大流行风险中所起作用的理解有限，部分原因是工程和流行病学中使用的不同建模方法。在这里，我们提出了蒸汽（通风感知大流行和爆发风险）模型，这是一个混合传播框架，将Reed-Frost密切接触动力学与Wells-Riley气溶胶介导的风险集成在一起。使用元人口结构来模拟多斑块环境（例如，单独的工作场所或学校），我们探讨了通风差异如何影响流行病的可能性。固定的少数个体被建模为“气溶胶”，这与过度分散的现实世界传播模式（例如SARS-CoV-2）一致。模拟结果表明，改善高危区域的通风和提高不同环境的基线通风都能降低风险。参数扫描每小时空气变化（ACH, 2-12）显示出早期饱和的非线性效益。这些发现强调了有针对性的通风策略的必要性，并显示了小世界效应如何放大异质性驱动的传播。VAPOR为将通风公平与流行病控制联系起来提供了一个框架。

{"title":"Insights from a Ventilation-Aware Pandemic and Outbreak Risk model (VAPOR)","authors":"Natalie J. Wilson , Callandra Moore , Clara Eunyoung Lee , Ashleigh R. Tuite , David N. Fisman","doi":"10.1016/j.epidem.2025.100878","DOIUrl":"10.1016/j.epidem.2025.100878","url":null,"abstract":"<div><div>Transmission of airborne pathogens in indoor spaces is strongly modulated by heterogeneity in ventilation. Understanding the role indoor air plays in pandemic risk is limited in part due to differing modeling approaches used in engineering and epidemiology. Here we present the VAPOR (Ventilation-Aware Pandemic and Outbreak Risk) model, a hybrid transmission framework that integrates Reed-Frost close-contact dynamics with Wells-Riley aerosol-mediated risk. Using a meta-population structure to simulate multi-patch environments (e.g., separate workplaces or schools), we explore how ventilation disparities shape epidemic potential. A fixed minority of individuals are modeled as “aerosolizers,” consistent with overdispersed real-world transmission patterns (e.g., SARS-CoV-2). Simulations reveal that both improving ventilation in high-risk patches and raising baseline ventilation across environments independently reduces risk. Parameter sweeps across air changes per hour (ACH, 2–12) demonstrate non-linear benefits with early saturation. These findings emphasize the need for targeted ventilation strategies and show how small-world effects amplify heterogeneity-driven transmission. VAPOR offers a framework for linking ventilation equity to epidemic control.</div></div>","PeriodicalId":49206,"journal":{"name":"Epidemics","volume":"54 ","pages":"Article 100878"},"PeriodicalIF":2.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145791943","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Machine learning-based short-term forecasting of COVID-19 hospital admissions using routine hospital patient data 基于机器学习的基于常规医院患者数据的COVID-19住院率短期预测

IF 2.4 3区医学 Q2 INFECTIOUS DISEASES

Epidemics

Pub Date : 2026-03-01 Epub Date: 2025-12-18 DOI: 10.1016/j.epidem.2025.100877

Martin S. Wohlfender , Judith A. Bouman , Olga Endrich , Alban Ramette , Alexander B. Leichtle , Guido Beldi , Christian L. Althaus , Julien Riou

During the COVID-19 pandemic, the field of infectious disease modeling advanced rapidly, with forecasting tools developed to track trends in transmission dynamics and anticipate potential shortages of critical resources such as hospital capacity. In this study, we compared short-term forecasting approaches for COVID-19 hospital admissions that generate forecasts one to five weeks ahead, using retrospective electronic health records. We extracted different features (e.g., daily emergency department visits) from an individual-level patient dataset covering six hospitals located in the region of Bern, Switzerland, from February 2020 to June 2023. We then applied five methods – last-observation carried forward (baseline), linear regression, XGBoost and two types of neural networks – to time series using a leave-future-out training scheme with multiple cutting points and optimized hyperparameters. Performance was evaluated using the root mean square error between forecasts and observations. Generally, we found that XGBoost outperformed the other methods in predicting future hospital admissions. Our results also show that adding features such as the number of hospital admissions with fever and augmenting hospital data with measurements of viral concentration in wastewater improves forecast accuracy. This study offers a thorough and systematic comparison of methods applicable to routine hospital data for real-time epidemic forecasting. With the increasing availability and volume of electronic health records, improved forecasting methods will contribute to more precise and timely information during epidemic waves of COVID-19 and other respiratory viruses, thereby strengthening evidence-based public health decision-making.

在2019冠状病毒病大流行期间，传染病建模领域发展迅速，开发了预测工具来跟踪传播动态趋势并预测医院能力等关键资源的潜在短缺。在这项研究中，我们比较了COVID-19住院率的短期预测方法，这些方法使用回顾性电子健康记录提前一到五周进行预测。从2020年2月至2023年6月，我们从覆盖瑞士伯尔尼地区六家医院的个人患者数据集中提取了不同的特征（例如，每日急诊室就诊次数）。然后，我们应用了五种方法-最后一次观测（基线），线性回归，XGBoost和两种类型的神经网络-使用具有多个切割点和优化超参数的留未来训练方案来处理时间序列。使用预测和观测之间的均方根误差来评估性能。一般来说，我们发现XGBoost在预测未来住院率方面优于其他方法。我们的研究结果还表明，增加发烧住院人数等特征，以及通过测量废水中的病毒浓度来增加医院数据，可以提高预测的准确性。本研究对适用于医院常规数据的流行病实时预测方法进行了全面、系统的比较。随着电子健康记录的可用性和数量的增加，改进的预测方法将有助于在2019冠状病毒病和其他呼吸道病毒流行期间提供更准确和及时的信息，从而加强基于证据的公共卫生决策。

{"title":"Machine learning-based short-term forecasting of COVID-19 hospital admissions using routine hospital patient data","authors":"Martin S. Wohlfender , Judith A. Bouman , Olga Endrich , Alban Ramette , Alexander B. Leichtle , Guido Beldi , Christian L. Althaus , Julien Riou","doi":"10.1016/j.epidem.2025.100877","DOIUrl":"10.1016/j.epidem.2025.100877","url":null,"abstract":"<div><div>During the COVID-19 pandemic, the field of infectious disease modeling advanced rapidly, with forecasting tools developed to track trends in transmission dynamics and anticipate potential shortages of critical resources such as hospital capacity. In this study, we compared short-term forecasting approaches for COVID-19 hospital admissions that generate forecasts one to five weeks ahead, using retrospective electronic health records. We extracted different features (e.g., daily emergency department visits) from an individual-level patient dataset covering six hospitals located in the region of Bern, Switzerland, from February 2020 to June 2023. We then applied five methods – last-observation carried forward (baseline), linear regression, XGBoost and two types of neural networks – to time series using a leave-future-out training scheme with multiple cutting points and optimized hyperparameters. Performance was evaluated using the root mean square error between forecasts and observations. Generally, we found that XGBoost outperformed the other methods in predicting future hospital admissions. Our results also show that adding features such as the number of hospital admissions with fever and augmenting hospital data with measurements of viral concentration in wastewater improves forecast accuracy. This study offers a thorough and systematic comparison of methods applicable to routine hospital data for real-time epidemic forecasting. With the increasing availability and volume of electronic health records, improved forecasting methods will contribute to more precise and timely information during epidemic waves of COVID-19 and other respiratory viruses, thereby strengthening evidence-based public health decision-making.</div></div>","PeriodicalId":49206,"journal":{"name":"Epidemics","volume":"54 ","pages":"Article 100877"},"PeriodicalIF":2.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927169","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Using wastewater surveillance to improve infectious disease control in correctional facilities and congregate living settings: A modeling perspective 利用废水监测改善惩教设施和聚集生活环境中的传染病控制：建模视角。

IF 2.4 3区医学 Q2 INFECTIOUS DISEASES

Epidemics

Pub Date : 2026-03-01 Epub Date: 2026-02-14 DOI: 10.1016/j.epidem.2026.100898

Daniel de-la-Rosa-Martinez , Kara L. Nelson , Joyce Lee , Rose S. Kantor , Ashley Hazel , Seth Blumberg

Wastewater surveillance is a valuable tool for monitoring infectious disease dynamics. However, its integration into outbreak control strategies in congregate settings requires further exploration. As observed during the SARS-CoV-2 pandemic, these high-risk environments can facilitate large outbreaks, further exacerbated by residents’ heightened vulnerability. Congregate settings exhibit distinct epidemiological dynamics that influence wastewater surveillance. For instance, their semi-closed populations and reduced mobility can lower environmental noise in wastewater signals, but small population sizes also increase stochastic fluctuations, complicating the interpretation of disease trends. In this context, mathematical modeling helps translate wastewater signals into actionable insights for outbreak response. This work synthesizes key benefits and challenges in applying wastewater surveillance in congregate settings and identifies modeling approaches that have potential to improve outbreak detection, enhance monitoring of transmission dynamics, and optimize infection control strategies. This provides a conceptual framework for expanding wastewater surveillance to strengthen infectious disease control in these high-risk populations.

废水监测是监测传染病动态的重要工具。然而，将其集成到聚集环境中的爆发控制策略中需要进一步探索。正如在SARS-CoV-2大流行期间所观察到的那样，这些高风险环境可能助长大规模疫情，居民的脆弱性进一步加剧。聚集环境表现出影响废水监测的不同流行病学动态。例如，它们的半封闭种群和减少的流动性可以降低废水信号中的环境噪声，但较小的种群规模也会增加随机波动，使疾病趋势的解释复杂化。在这种情况下，数学建模有助于将废水信号转化为应对疫情的可行见解。这项工作综合了在聚集环境中应用废水监测的主要好处和挑战，并确定了有可能改善疫情检测、加强传播动力学监测和优化感染控制策略的建模方法。这为扩大废水监测以加强这些高危人群的传染病控制提供了一个概念性框架。

{"title":"Using wastewater surveillance to improve infectious disease control in correctional facilities and congregate living settings: A modeling perspective","authors":"Daniel de-la-Rosa-Martinez , Kara L. Nelson , Joyce Lee , Rose S. Kantor , Ashley Hazel , Seth Blumberg","doi":"10.1016/j.epidem.2026.100898","DOIUrl":"10.1016/j.epidem.2026.100898","url":null,"abstract":"<div><div>Wastewater surveillance is a valuable tool for monitoring infectious disease dynamics. However, its integration into outbreak control strategies in congregate settings requires further exploration. As observed during the SARS-CoV-2 pandemic, these high-risk environments can facilitate large outbreaks, further exacerbated by residents’ heightened vulnerability. Congregate settings exhibit distinct epidemiological dynamics that influence wastewater surveillance. For instance, their semi-closed populations and reduced mobility can lower environmental noise in wastewater signals, but small population sizes also increase stochastic fluctuations, complicating the interpretation of disease trends. In this context, mathematical modeling helps translate wastewater signals into actionable insights for outbreak response. This work synthesizes key benefits and challenges in applying wastewater surveillance in congregate settings and identifies modeling approaches that have potential to improve outbreak detection, enhance monitoring of transmission dynamics, and optimize infection control strategies. This provides a conceptual framework for expanding wastewater surveillance to strengthen infectious disease control in these high-risk populations.</div></div>","PeriodicalId":49206,"journal":{"name":"Epidemics","volume":"54 ","pages":"Article 100898"},"PeriodicalIF":2.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147318762","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

ML-ABC: Machine-learning assisted Approximate Bayesian Computation for efficient calibration of agent-based models for pandemic outbreak analysis ML-ABC：机器学习辅助的近似贝叶斯计算，用于大流行爆发分析中基于主体的模型的有效校准

IF 2.4 3区医学 Q2 INFECTIOUS DISEASES

Epidemics

Pub Date : 2026-03-01 Epub Date: 2026-01-05 DOI: 10.1016/j.epidem.2025.100881

Thomas Bayley , Tony Ward , Fabian Sturman , Akashaditya Das , Luca Imeneo , Cliff Kerr , Christophe Fraser , Simon Maskell , Jasmina Panovska-Griffiths

Mathematical modelling with agent-based models (ABMs) has gained popularity during the COVID-19 pandemic, but their complexity makes efficient and robust calibration to data challenging, particularly when applying Bayesian methods to quantify parameter uncertainty. We propose a method for calibrating ABMs that combines a Machine-Learning step with Approximate Bayesian Computation (ML-ABC). We showcase ML-ABC application with a proof-of-principle case study, in which we calibrate the Covasim -a stochastic ABM that has been used to model the English COVID-19 epidemic and inform policy at important junctions. Benchmarking against traditional Rejection-ABC (R-ABC), we illustrate the advantage of ML-ABC application in calibrating Covasim to data on hospitalisations and deaths from COVID-19 during the first and the second COVID-19 epidemic waves of 2020 and early 2021. Across scenarios, we demonstrate that using an ML screening step allows us to derive identical posterior distributions of the calibrated Covasim parameters as with the traditional R-ABC method, but faster. Specifically, we derive posterior distributions for input parameters around 52% faster when calibrating to the first epidemic wave and around 33% faster when calibrating parameters for the second epidemic wave, compared to the traditional R-ABC. Policy modelling requires calibration which is both efficient to adapt to fast-changing pandemic environments and robust to ensure confidence in policy decisions. However, existing ABM calibration often relies on myopic non-exhaustive searches in order to remain tractable, resulting in point parameter estimates. In this preliminary study, ML-ABC strictly improves upon existing ABC calibration approaches in all tested scenarios, indicating its potential to make ABC competitive with point-estimate calibration approaches. This novel approach offers a pathway to effectively calibrate ABMs in a way which is both efficient and quantifies parameter uncertainty, crucial for realising the potential of ABMs for timely and responsively modelling during an emerging epidemic.

基于主体的模型（ABMs）的数学建模在COVID-19大流行期间得到了普及，但其复杂性使得对数据的有效和稳健校准具有挑战性，特别是在应用贝叶斯方法量化参数不确定性时。我们提出了一种校准ABMs的方法，该方法将机器学习步骤与近似贝叶斯计算（ML-ABC）相结合。我们通过一个原理验证案例研究展示了ML-ABC的应用，在该案例研究中，我们校准了Covasim（一种随机ABM，已用于模拟英国COVID-19流行病并在重要节点为政策提供信息）。以传统的拒绝abc （R-ABC）为基准，我们说明了ML-ABC应用在将Covasim校准为2020年和2021年初第一次和第二次COVID-19流行期间COVID-19住院和死亡数据方面的优势。在各种情况下，我们证明使用ML筛选步骤可以使我们获得与传统R-ABC方法相同的校准Covasim参数的后验分布，但速度更快。具体而言，与传统的R-ABC相比，我们在校准第一流行波时获得输入参数的后验分布速度约为52%，在校准第二流行波参数时获得输入参数的后验分布速度约为33%。政策建模需要校准，这既能有效地适应快速变化的大流行环境，又能确保对政策决定的信心。然而，现有的ABM校准为了保持可处理性，往往依赖于短视的非穷举搜索，导致点参数估计。在本初步研究中，ML-ABC在所有测试场景中都严格改进了现有的ABC校准方法，这表明它有可能使ABC与点估计校准方法竞争。这种新方法提供了一种有效校准ABMs的途径，这种方法既有效又量化参数不确定性，这对于实现ABMs在新出现的流行病期间及时和响应性建模的潜力至关重要。

{"title":"ML-ABC: Machine-learning assisted Approximate Bayesian Computation for efficient calibration of agent-based models for pandemic outbreak analysis","authors":"Thomas Bayley , Tony Ward , Fabian Sturman , Akashaditya Das , Luca Imeneo , Cliff Kerr , Christophe Fraser , Simon Maskell , Jasmina Panovska-Griffiths","doi":"10.1016/j.epidem.2025.100881","DOIUrl":"10.1016/j.epidem.2025.100881","url":null,"abstract":"<div><div>Mathematical modelling with agent-based models (ABMs) has gained popularity during the COVID-19 pandemic, but their complexity makes efficient and robust calibration to data challenging, particularly when applying Bayesian methods to quantify parameter uncertainty. We propose a method for calibrating ABMs that combines a Machine-Learning step with Approximate Bayesian Computation (ML-ABC). We showcase ML-ABC application with a proof-of-principle case study, in which we calibrate the Covasim -a stochastic ABM that has been used to model the English COVID-19 epidemic and inform policy at important junctions. Benchmarking against traditional Rejection-ABC (R-ABC), we illustrate the advantage of ML-ABC application in calibrating Covasim to data on hospitalisations and deaths from COVID-19 during the first and the second COVID-19 epidemic waves of 2020 and early 2021. Across scenarios, we demonstrate that using an ML screening step allows us to derive identical posterior distributions of the calibrated Covasim parameters as with the traditional R-ABC method, but faster. Specifically, we derive posterior distributions for input parameters around 52% faster when calibrating to the first epidemic wave and around 33% faster when calibrating parameters for the second epidemic wave, compared to the traditional R-ABC. Policy modelling requires calibration which is both efficient to adapt to fast-changing pandemic environments and robust to ensure confidence in policy decisions. However, existing ABM calibration often relies on myopic non-exhaustive searches in order to remain tractable, resulting in point parameter estimates. In this preliminary study, ML-ABC strictly improves upon existing ABC calibration approaches in all tested scenarios, indicating its potential to make ABC competitive with point-estimate calibration approaches. This novel approach offers a pathway to effectively calibrate ABMs in a way which is both efficient and quantifies parameter uncertainty, crucial for realising the potential of ABMs for timely and responsively modelling during an emerging epidemic.</div></div>","PeriodicalId":49206,"journal":{"name":"Epidemics","volume":"54 ","pages":"Article 100881"},"PeriodicalIF":2.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146078037","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0