Toward a Dynamic Ovarian Recruitment Phenotype in Assisted Reproduction

Anti-Müllerian hormone (AMH) and antral follicle count (AFC) predict ovarian response, while serial ultrasound documents the evolving antral cohort during controlled stimulation. Yet most visit-level summaries do not preserve which follicle is which across scans. We propose a dynamic ovarian recruitment phenotype, defined as a treatment-conditioned description of identity-linked antral follicle trajectories rather than a new measure of primordial reserve. Our primary, falsifiable hypothesis is that individual follicle identity contributes reproducible information about mature-oocyte yield beyond baseline predictors and the complete routine monitoring history available at the same early prediction landmark. A stringent test compares three nested models: baseline characteristics; baseline plus contemporaneous visit-level measurements, hormones and treatment exposure; and the same information augmented by identity-linked growth, persistence and synchrony, with a rank-matched pseudo-identity model as an intermediate comparator. Scans, prediction timing and modelling flexibility must be matched, with identity permutation as a negative control. A separate hypothesis asks whether any trajectory feature retains a patient-level component across comparable stimulation cycles. Spatial features are exploratory and are not evidence of primordial-follicle activation. Support requires reliable tracking, external validation, clinically relevant incremental prediction and, ultimately, evidence of benefit for a specific clinical decision. Equal performance of identity-free monitoring models, sensitivity to matching error or absent cross-cycle reproducibility would substantially weaken the proposed phenotype. This essay offers an operational definition and a validation pathway, not new clinical data or treatment recommendations.