Beyond the Average Patient: The Ergodicity Problem in Evidence-Based Medicine and a Proposal for Individual-Dynamic...

Evidence-based medicine (EBM) replaced authority with appraisal and has reduced harm. Its own methodologists have documented a persistent gap between what its methods conclude and what its practice does: average effects that are marginal for most treated patients, guidelines that misfit the individual, widespread difficulty converting relative into absolute effects, and no built-in trigger for stopping a treatment that is not helping.

To identify a common source of these problems in the application of population-level estimates to individuals, and to propose a framework that corrects it without abandoning controlled evidence.

Conceptual and critical analysis drawing on the methodological critique of EBM published by its founders and leading epidemiologists; the theory of ergodicity in dynamical systems and its empirical tests; the reference class problem; and existing clinical existence proofs of individual-level measurement and closed-loop intervention.

The assumption that a population estimate applies to an individual is the ergodic assumption, which holds when a process is homogeneous across persons and stationary over time. Homogeneity fails whenever treatment response differs between people, which is common; where an individual's responses are reproducible, individual-level inference is both necessary and feasible. Large within-person variability weakens every inference and must be measured rather than assumed away. The N-of-1 randomised trial, which estimates an effect in the individual, was published by Guyatt and Sackett in 1986 and ranks at Level 1 of the Oxford 2011 Levels of Evidence, yet a systematic review found 108 published trials in twenty-five years.

I propose Individual-Dynamic Medicine, a biocybernetic framework of six principles, each with an existence proof in mainstream medicine, specifying how population evidence and within-person trials divide the work of an individual decision, and state its assumptions, limitations, and the predictions on which it could fail.