Model Design Clause Samples
Model Design.
1.2.2.1. Supervised learning analysis pipeline Figure F1 illustrates the pipeline adopted for the supervised learning analysis towards the design and development of robust and generalizable predictive models to minimize training errors while considering the bias-variance tradeoff. These steps are described in more detail below. Initially, raw data were rescaled to zero mean and unit variance and ordinal variables were recoded into dummy binary variables. Cases and variables with more than 10% of missingness were excluded from the final dataset. Remaining missing values were replaced by the global median value.
Model Design. We developed a deterministic, age structured, dynamic model of NoV transmission that includes symptomatic and asymptomatic infections. The full model is illustrated by the flow diagrams in Figures 2 and 3, with parameters as defined in Table 4. The full model equations are described below. We assume that newborn infants have no maternal immunity; therefore all births enter into the susceptible class. All individuals in the susceptible class can be infected at rate λ(t) and they progress from the exposed class into the infected symptomatic class at rate μs. Individuals in the infected symptomatic class progress to the infected asymptomatic class at rate μa and move on to the recovered class at rate ρ. Individuals in the recovered class are assumed to have immunity to symptomatic disease, as opposed to asymptomatic infection, and they move from the recovered class back to the susceptible class at rate θ. However, individuals from the recovered class can also cycle back into the infected asymptomatic class at rate λ(t). One scenario required a slight adjustment to the model, and this version is also included below.