Algorithmic Time Rationing: The New Inverse Care Law

The patient with five conditions, low health literacy, and limited German waits longer, gets less time, and leaves with more unfinished business than the wealthy patient with one complaint. Soon, an algorithm will codify this. AI-driven scheduling systems increasingly predict appointment duration from historical data, and when optimised for throughput they risk allocating consultation time on the basis of what is typical rather than what is needed. We introduce algorithmic time rationing: the systematic under-allocation of consultation time to patients whose clinical needs exceed the statistical average predicted by scheduling algorithms. We propose T(p) = f(C(p), R(p), A(p)) as a descriptive function decomposing time allocation into clinical complexity, reimbursement profile, and AI prediction. Three scheduling scenarios are analysed: uniform-slot, efficiency-first, and complexity-aware. When AI scheduling is trained on historical durations in which billing categories already constrained consultation length, the system inherits and perpetuates these constraints, establishing a new channel through which administrative norms reshape clinical reality. Five patient groups at highest risk of systematic under-allocation are identified: multimorbid patients, those with low health literacy or mental health comorbidity, socioeconomically disadvantaged populations, and non-native speakers. This paper develops a theoretical framework, not an empirical study; T(p) is a conceptual decomposition and the three scenarios are illustrative, not estimated from real scheduling data. The central argument is that AI-driven appointment scheduling, optimised for throughput, threatens to re-implement Tudor Hart's inverse care law as code.