Epi-DNNs: Epidemiological priors informed deep neural networks for modeling COVID-19 dynamics

Ning, Xiao; Jia, Linlin; Wei, Yongyue; Li, Xian; Chen, Feng

doi:10.1016/j.compbiomed.2023.106693

Cited by 10 publications

(7 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A long tail of the epidemic in Shanghai was observed and the 16 districts were divided into two groups according to the baseline estimation and prediction results. Our fitting results on the data can be mutually verified with two studies [ 40 , 49 ], including the simulation results based on deep neural networks. Then to avoid a big resurgence, we considered a step-by-step relaxation strategy with two patterns: the city-based and the regional linked (district-based) relaxation strategies after 28 April.…”

Section: Discussionsupporting

confidence: 74%

Lessons drawn from Shanghai for controlling highly transmissible SARS-CoV-2 variants: insights from a modelling study

Wang

Gao

et al. 2023

BMC Infect Dis

View full text Add to dashboard Cite

Background The continuous emergence of novel SARS-CoV-2 variants with markedly increased transmissibility presents major challenges to the zero-COVID policy in China. It is critical to adjust aspects of the policy about non-pharmaceutical interventions (NPIs) by searching for and implementing more effective ways. We use a mathematical model to mimic the epidemic pattern of the Omicron variant in Shanghai to quantitatively show the control challenges and investigate the feasibility of different control patterns in avoiding other epidemic waves. Methods We initially construct a dynamic model with a core step-by-step release strategy to reveal its role in controlling the spread of COVID-19, including the city-based pattern and the district-based pattern. We used the least squares method and real reported case data to fit the model for Shanghai and its 16 districts, respectively. Optimal control theory was utilized to explore the quantitative and optimal solutions of the time-varying control strength (i.e., contact rate) to suppress the highly transmissible SARS-CoV-2 variants. Results The necessary period for reaching the zero-COVID goal can be nearly 4 months, and the final epidemic size was 629,625 (95%CI: [608,049, 651,201]). By adopting the city-based pattern, 7 out of 16 strategies released the NPIs more or earlier than the baseline and ensured a zero-resurgence risk at the average cost of 10 to 129 more cases in June. By adopting the district-based pattern, a regional linked release can allow resumption of social activity to ~ 100% in the boundary-region group about 14 days earlier and allow people to flow between different districts without causing infection resurgence. Optimal solutions of the contact rate were obtained with various testing intensities, and higher diagnosis rate correlated with higher optimal contact rate while the number of daily reported cases remained almost unchanged. Conclusions Shanghai could have been bolder and more flexible in unleashing social activity than they did. The boundary-region group should be relaxed earlier and more attention should be paid to the centre-region group. With a more intensive testing strategy, people could return to normal life as much as possible but still ensure the epidemic was maintained at a relatively low level.

show abstract

Section: Discussionsupporting

confidence: 74%

Lessons drawn from Shanghai for controlling highly transmissible SARS-CoV-2 variants: insights from a modelling study

Wang

Gao

et al. 2023

BMC Infect Dis

View full text Add to dashboard Cite

show abstract

“…A number of studies based on compartmental models have been explored by the researchers to model the spread trend of the COVID-19 [24], [30], [32]. The Susceptible-Infected-Recovered (SIR) [16] model has been extended by a number of researchers by introducing a few new compartments related to the COVID-19 spread.…”

Section: Related Workmentioning

confidence: 99%

Epidemic Modeling using Hybrid of Time-varying SIRD, Particle Swarm Optimization, and Deep Learning

Kumar,

Susan

2023

2023 14th International Conference on Computing Communication and Networking Technologies (ICCCNT)

View full text Add to dashboard Cite

Epidemiological models are best suitable to model an epidemic if the spread pattern is stationary. To deal with nonstationary patterns and multiple waves of an epidemic, we develop a hybrid model encompassing epidemic modeling, particle swarm optimization, and deep learning. The model mainly caters to three objectives for better prediction: 1. Periodic estimation of the model parameters. 2. Incorporating impact of all the aspects using data fitting and parameter optimization 3. Deep learning based prediction of the model parameters. In our model, we use a system of ordinary differential equations (ODEs) for Susceptible-Infected-Recovered-Dead (SIRD) epidemic modeling, Particle Swarm Optimization (PSO) for model parameter optimization, and stacked-LSTM for forecasting the model parameters. Initial or one time estimation of model parameters is not able to model multiple waves of an epidemic. So, we estimate the model parameters periodically (weekly). We use PSO to identify the optimum values of the model parameters. We next train the stacked-LSTM on the optimized parameters, and perform forecasting of the model parameters for upcoming four weeks. Further, we fed the LSTM forecasted parameters into the SIRD model to forecast the number of COVID-19 cases. We evaluate the model for highly affected three countries namely; the USA, India, and the UK. The proposed hybrid model is able to deal with multiple waves, and has outperformed existing methods on all the three datasets.

show abstract

“…6 Likewise, the capability of epidemic models is constrained to the problem of parameter dimensions and estimation. 2 Currently, discussions persist as to whether knowledgeinformed learning (KIL) is a viable avenue to bridge the gap between ML and dynamic models-to inform decision support systems for ML practitioners, policy developers, and decision makers. KIL, 7 also framed as physics-informed ML (PIML), 6 entails the synthesis of multiple viewpoints, principles, and evidence to provide informative priors for modeling.…”

mentioning

confidence: 99%

“…The determinants of a given infectious disease are complex, and they include but are not limited to 1) environmental factors (i.e., weather and climate) and 2) human behavioral characteristics (i.e., public mobility) and political factors such as government policies and interventions. Previous studies have shown that machine learning (ML) models 1 and epidemic compartmental models 2 formulated using deterministic differential equations are the dominant tools for examining, modeling, and analyzing the transmission of infectious diseases.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Informed Random Forest to Model Associations of Epidemiological Priors, Government Policies, and Public Mobility

Thapelo,

Mpoeleng,

Hillhouse

2023

MDM Policy & Practice

View full text Add to dashboard Cite

Background. Infectious diseases constitute a significant concern worldwide due to their increasing prevalence, associated health risks, and the socioeconomic costs. Machine learning (ML) models and epidemic models formulated using deterministic differential equations are the most dominant tools for analyzing and modeling the transmission of infectious diseases. However, ML models can be inconsistent in extracting the dynamics of a disease in the presence of data drifts. Likewise, the capability of epidemic models is constrained to parameter dimensions and estimation. We aimed at creating a framework of informed ML that integrates a random forest (RF) with an adapted susceptible infectious recovered (SIR) model to account for accuracy and consistency in stochasticity within the dynamics of coronavirus disease 2019 (COVID-19). Methods. An adapted SIR model was used to inform a default RF on predicting new COVID-19 cases (NCCs) at given intervals. We validated the performance of the informed RF (IRF) using real data. We used Botswana’s pharmaceutical interventions (PIs) and non-PIs (NPIs) adopted between February 2020 and August 2022. The discrepancy between predictions and observations is modeled using loss functions, which are minimized, interpreted, and used to assess the IRF. Results. The findings on the real data have revealed the effectiveness of the default RF in modeling and predicting NCCs. The use of the effective reproductive rate to inform the RF yielded an excellent predictive power (84%) compared with 75% by the default RF. Conclusion. This research has potential to inform policy and decision makers in developing systems to evaluate interventions for infectious diseases. Highlights This framework is initiated by incorporating model outputs from an epidemic model to a machine learning model. An informed random forest (RF) is instantiated to model government and public responses to the COVID-19 pandemic. This framework does not require data transformations, and the epidemic model is shown to boost the RF’s performance. This is a baseline knowledge-informed learning framework for assessing public health interventions in Botswana.

show abstract

Epi-DNNs: Epidemiological priors informed deep neural networks for modeling COVID-19 dynamics

Cited by 10 publications

References 40 publications

Lessons drawn from Shanghai for controlling highly transmissible SARS-CoV-2 variants: insights from a modelling study

Lessons drawn from Shanghai for controlling highly transmissible SARS-CoV-2 variants: insights from a modelling study

Epidemic Modeling using Hybrid of Time-varying SIRD, Particle Swarm Optimization, and Deep Learning

Informed Random Forest to Model Associations of Epidemiological Priors, Government Policies, and Public Mobility

Contact Info

Product

Resources

About