Cox Model Go Brrr - A Journey to Performance

Learn how to optimize the Cox model implementation in Julia through a comprehensive performance journey that transforms a naive implementation into one that outperforms historical R/C equivalents. Explore the Cox model as a standard parametric model for censored time-to-event data in survival analysis, starting with an off-the-shelf numerical solver using automatic differentiation before progressing through manual derivative and Hessian implementation, custom solving loops, profiling techniques, and allocation reduction strategies. Discover how Julia's ecosystem tools facilitate type stability and allocation reductions while building fair performance comparisons, and examine the mathematical analysis that led to departing from standard Newton-Raphson minimization for crucial performance improvements. Master the development of a high-performance Cox algorithm implementation destined for SurvivalModels.jl that serves as a pillar of numerical survival analysis across fields including public health and medicine.