EUVIMEDEUVIMED entdecken
Crossref

L26/P-876 Clinical performance of AI-based follicular monitoring in routine IVF care

J. Griffiths, V. Ferguson, P. Atkinson, E. Ebinger

Human Reproduction · 01.07.2026 · 41 (Supplement_1)

DOI: 10.1093/humrep/deag083.1200 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Human Reproduction, online erschienen am 08.07.2026. Verfasst von J. Griffiths und 3 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Study question Can AI-based automated follicular monitoring reduce measurement variability without altering clinical decision-making or laboratory performance in routine IVF practice? Summary answer AI-based follicular monitoring reduced measurement variability, did not change clinical decisions, and maintained laboratory key performance indicators in routine IVF care. What is known already Manual follicular measurement during ovarian stimulation is subject to inter-operator, inter-machine, and inter-site variability, particularly in high-volume fertility networks. Such variability may influence trigger timing, oocyte maturity, and workflow efficiency. Increasing clinical demand further strains sonographer capacity and operational resources. Artificial intelligence-based ultrasound analysis has been proposed to standardise follicular assessment; however, evidence from real-world implementation demonstrating unchanged clinical decision-making and preserved laboratory performance remains limited. Study design, size, duration This clinical validation study assessed the real-world implementation of an AI-based follicular monitoring system in Adora Fertility Clinic chain. Internal validation consisted of a blinded clinical decision review of 100 ultrasound scans and an analysis of per-visit follicle size variability. Laboratory key performance indicators were evaluated over a two-month period following conversion to exclusive AI-based monitoring in selected regional clinics. Participants/materials, setting, methods 100 ovarian ultrasound scans were anonymised and blinded as either AI-based (FOLLISCAN) or conventional sonographer measurements. The scans were independently reviewed by the Medical Director to assess whether clinical decisions would have potentially differed based on the two sets of follicle size data. Follicle sizes from each scan were extracted to evaluate per-visit measurement variability and were visualised. Laboratory key performance indicators from Queensland clinics following exclusive AI-based monitoring were reviewed over a two-month period. Main results and the role of chance Blinded review of 100 ultrasound scans demonstrated that differences between AI-based and conventional follicular measurements did not result in potential changes to clinical decision-making regarding ovarian stimulation management or trigger timing. Per-visit analysis of follicle sizes showed reduced variability in follicle measurements and follicle counts when AI-based monitoring was used, including consistent capture of follicles <10 mm and reduced variation to oocyte pick-up. Following implementation, laboratory key performance indicators remained within established competency thresholds for IVF and ICSI fertilisation, blastocyst development, and ICSI oocyte maturation across centres. In Queensland clinics, ICSI oocyte maturation rate increased from 72.2% pre-implementation to 73.5% post-implementation, while IVF fertilisation rate remained stable (76.9% vs 76.6%). In Western Australia clinics, ICSI oocyte maturation rate increased from 79.2% to 81.2%, and IVF fertilisation rate remained stable (78.6% vs 78.2%). No increase in cycle cancellation rates was observed following implementation. Operationally, AI-based monitoring improved workflow efficiency, reducing ultrasound assessment time by up to 50% compared with conventional scanning and enabling more consistent sonographer scheduling. The combination of stable laboratory outcomes, unchanged clinical decisions, and improved efficiency across centres suggests robust real-world clinical performance. Limitations, reasons for caution The study was retrospective and limited to internal validation within a single clinical network. Clinical pregnancy and live birth outcomes were not assessed. Longer follow-up and external validation are required to confirm broader generalisability. Wider implications of the findings AI-based follicular monitoring may support standardised ovarian stimulation management, reduce operator-dependent variability, and enable scalable IVF workflows without compromising laboratory performance or patient safety. Trial registration number No

Zitieren

APA

Griffiths, J., Ferguson, V., Atkinson, P., & Ebinger, E. (2026). L26/P-876 Clinical performance of AI-based follicular monitoring in routine IVF care. Human Reproduction, 41(Supplement_1). https://doi.org/10.1093/humrep/deag083.1200

Vancouver

Griffiths J, Ferguson V, Atkinson P, Ebinger E. L26/P-876 Clinical performance of AI-based follicular monitoring in routine IVF care. Human Reproduction. 2026;41(Supplement_1). doi:10.1093/humrep/deag083.1200

BibTeX
@article{Griffiths2026,
  author = {Griffiths, J. and Ferguson, V. and Atkinson, P. and Ebinger, E.},
  title = {L26/P-876 Clinical performance of AI-based follicular monitoring in routine IVF care},
  journal = {Human Reproduction},
  year = {2026},
  volume = {41},
  number = {Supplement_1},
  doi = {10.1093/humrep/deag083.1200}
}
Kostenlos starten →