Figure 5
From: Algorithmic discovery of dynamic models from infectious disease data
Comparison between rubella incidence data and coefficients of the best SINDy-discovered model using a function library of polynomials up to 2nd order, and showing the model with the lowest AIC scores across the \({S}_{0}-\lambda \) parameter grid. The algorithm was unable to discover a model that exhibited the multi-annual cycle observed in the data, instead returning an annual cycle. Despite this, strong dependence on the mass action incidence term is again present. The sparse regression excluded 14 terms, giving \(r=0.7\). The parameter grids appear in SI Appendix, Fig. 10. The results in the table display the SINDy-discovered coefficients of the corresponding terms in (Eqs. 1–3).
