12Mo Eval S1: Q1a.ii, Q2c. (August 2024 scenario)

[djinnome]

Evaluation Scenario 1: Forecasting With NPIs

Background: In April 2020, COVID-19 has spread from small clusters of imported cases to more widespread community transmission. Preventive measures are limited to non-pharmaceutical interventions such as social distancing, isolation of infected individuals, and other forms of source control such as masking. We are interested in determining how masking efficacy and compliance can help reduce cases, hospitalizations, and deaths in New York.
Timepoint (Mapping to Forecasting Challenge Timepoint 1): April 3rd, 2020
Location: New York State
Model: Begin with the following SEIRHD model structure and set of differential equations (a version of this may already exist in the workbench; if not, create it). The general form/structure of the model is below. Also see accompanying code.
Note: p(I → R) in the diagram on the left is pIàR in the equations on the right, and represents the probability of moving between the indicated states. r(state1 → state2) represent rates for how long processes take (i.e. 1/average time to move between states, e.g. 1/ incubation period, etc.). For parameter values for the base model (prior to modification tasks in the scenario question), use the following values: β= 0.4 new infections per infected person/"day"; r(I → R)=0.07/day; r(I → H)=0.1/day; r(E → I)=0.2/day; r(H → R)=0.1/day; r(H → D)= 0.1/day; p(I → R)= 0.8; p(I → H)= 0.2; p(H → R)= 0.88 ;p(H → D)= 0.12. Use N = 19.34 million (approximate population size for NY state in 2020). For initial conditions I(0) and D(0), please pull values from the gold standard cases and deaths data from the Covid-19 ForecastHub. For H(0) use HHS hospitalization data from https://healthdata.gov/Hospital/COVID-19-Reported-Patient-Impact-and-Hospital-Capa/g62h-syeh. Let R(0) = cumulative infections – cumulative deaths, as of April 3rd, 2020. Let E(0) = I(0)/4. Let S(0) = N – E(0) - I(0)-R(0)-H(0)-D(0).

Data: For comparison against historical data or initializing simulations, use gold standard data from the Covid-19 ForecastHub

Questions: For the questions below, you will be asked to implement various policy interventions on April 15th, 2020. Forecasts should begin on April 3rd, 2020.

1. Masking Forecasts: For this question, assume that a masking policy will go into place on April 15th, 2020
   a. Starting from April 3rd, 2020, forecast the next four weeks of the pandemic (for cases, hospitalizations, and deaths) assuming the following constant levels of masking compliance: 40%, 60%, and 80%. Assume that any person who complies with the masking policy is wearing a surgical mask. How does compliance affect forecasted cases, hospitalizations, and deaths?
   - i. (TA1 Search and Discovery Workflow, 1 Hr. Time Limit) Find estimates on the efficacy of surgical masks in preventing onward transmission of SARS-CoV-2 (preferred) or comparable viral respiratory pathogens (e.g., MERS-CoV, SARS), including any information about uncertainty in these estimates. The term surgical mask here refers to the commonly available, disposable procedure mask, not an N95-type respirator. Find 3 credible documents that provide estimates and use your judgment to determine what value (in the deterministic case) or distribution (in the probabilistic case) to use in your forecasts in 1.a.iii.
   - ii. (TA2 Model Modification Workflow) Begin with an SEIRHD model with parameter settings as described in the scenario description. Modify the model to include the ability to implement a masking policy intervention with different compliance levels. Implement this in the following three ways:
   - (1). Introduce a modification term to β as described in Srivastav et. al. (DOI: 10.3934/mbe.2021010): (1-ϵ_m c_m ), where c_m is the fraction of the population that wear face masks correctly and consistently (i.e. comply with masking policies), and ϵ_m is the efficacy of the masks themselves.
   - (2) Adjust the transmission rate following an intervention period and create a time-varying β function, as shown in https://doi.org/10.3390/ijerph18179027
   - (3) Add compartments and transitions to represent mask wearing/non-mask wearing populations. See examples for several similar options of representing this addition; you may implement the update in a way that represents one or a combination of these approaches, or has an otherwise analogous effect: https://doi.org/10.1098/rspa.2020.0376, https://doi.org/10.1016/j.idm.2020.04.001, https://doi.org/10.1016/j.chaos.2020.110599, https://doi.org/10.1177/0272989X211019029