Figure S1 from Characterization of Non-Monotonic Relationships between Tumor Mutational Burden and Clinical Outcomes

Simulated step data and examples with a polynomial transformation. 15 simulated survival datasets were generated for a step relationship with TMB (A). B shows the hazard ratios and associated log-likelihood ratio tests and associated cutoffs of searching for an optimal cutoff, while C shows the fits of a Cox model, FCN neural network, and a neural network comprised of a single neuron with sigmoid activation. D shows an example of a quadratic relationship with TMB (E). Generating a simulated dataset from the risk relationship in D we explored fitting a Cox model with a two degree polynomial in addition to a neural net. In F we show the fit of a two degree polynomial with non-monotonic data.