DARPA-ASKEM / program-milestones

Repository for materials related to program milestone hackathon and evaluation events
6 stars 10 forks source link

September problem 3 part 1: 12Mo Hackathon S2 Q1a-c (Model extraction and simulation for SV2AIR3 basic version); #79

Open djinnome opened 2 months ago

djinnome commented 2 months ago

Hackathon Scenario 2: Multiple Vaccines and Variants

Models from early in the Covid-19 pandemic were relatively simple, as we had no vaccines available, only NPIs to mitigate the spread of the disease, and only one strain (the original wild type) to worry about. By mid-2021, the situation was much more complex, as we had access to multiple vaccines, each with multiple doses, and the emergence of multiple variants. You are interested in modeling this situation with SV2(AIR)3, one of the first models to represent multiple Covid-19 variants, and the competition between them.

Question 1: Extract the simplest version of the model with 1 vaccine and 1 variant (shown in Fig. 1A)

a. (TA1 Model Extraction Workflow) Extract the model from Matlab code ‘svair_simple.m’ and ‘get_beta.m’ and with the publication equations and Fig. 1A (with the exception of one small difference between code and figure). Note that in this code, the units of all state variables are in terms of fraction of the total population. b. (TA3 Simulation Workflow) Simulate the first 14 months of the pandemic beginning on January 1st, 2020. Consider only the Wild Type variant, and no vaccination during this period. Use parameter values from the supplementary material, for the wild type. Use the population of Ontario in 2019: 14.57 million people, and let the natural death rate μ= 2.05e-5. For initial conditions, let I(0) = 1e-6,S(0) = 1-sum(I(0)+A(0)), and all other values be 0. When do I(t) and A(t) peak, and what is the value of these variables at their peaks? How do the I(t) and A(t) profiles compare with Fig. 3d,f (which considered 3 variants and 2 types of vaccines)? Given that this question assumes only the presence of the Wild Type variant, do the results seem reasonable? c. (TA2 Domain Knowledge Grounding; ASKEM Workbench Only) Demonstrate that the domain knowledge groundings accurately reflect new concepts with the extensions such as “variants”.

Input
  • Publication with equations, and simplest SVAIR version of the model described in Fig. 1A
  • Accompanying code
Tasks
  • Model extraction
  • Set parameters according to 1b and supplementary materials
  • Simulate model
Outputs
  • Extracted model grounded to the domain knowledge graph, and simulation outcomes
  • (S, V, A, I, R)