Naegele’s rule is not a 280-day count: generative AI conflates two distinct pregnancy-dating algorithms
Journal of Perinatal Medicine · 17.11.2026
DOI: 10.1515/jpm-2026-0477 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Perinatal Medicine, online erschienen am 17.11.2026. Verfasst von Amos Grünebaum und 1 weiteren Autor:innen. Beteiligte Einrichtung: Northwell Health, New Hyde Park, NY, USA. Die Arbeit nennt 8 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Objectives
Naegele’s calendar rule and an exact 280-day count are commonly treated as equivalent methods for estimating the due date (EDD), although their calendar arithmetic differs. We quantified this difference and evaluated whether generative artificial intelligence (AI) systems disclosed and accurately labeled the method used.
Methods
A deterministic script calculated EDDs for every last menstrual period (LMP) date from January 1, 1812, through December 31, 2025, using Naegele’s calendar rule and a fixed 280-day count. Five generative AI systems answered four prespecified LMP prompts, representing agreement and differences of 1, 2, and 3 days. Each prompt was submitted in three separate conversations, yielding 60 responses.
Results
Among 78,163 LMP dates, the methods produced identical EDDs in 34,609 cases (44.3 %) and different EDDs in 43,554 (55.7 %). Naegele’s calendar rule produced dates 1 day later in 27.0 %, 2 days later in 21.0 %, and 3 days later in 7.7 %. The methods never agreed among 14,828 intervals containing February 29. Of 60 AI responses, 57 (95.0 %) included the exact 280-day date and 48 (80.0 %) stated a calculation method. Only 18 of 45 responses to divergent dates distinguished the algorithms. Twelve of 48 method-stating responses contained a method-label mismatch.
Conclusions
Naegele’s calendar rule and fixed 280-day counting are distinct algorithms. Their small numerical differences are unlikely to be clinically important, but they provide a deterministic test of computational transparency. Generative AI systems should disclose the calculation used, distinguish non-equivalent methods, and label them accurately.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Northwell Health, New Hyde Park, NY, USA.
Zitieren
APA
Grünebaum, A., & Chervenak, F. A. (2026). Naegele’s rule is not a 280-day count: generative AI conflates two distinct pregnancy-dating algorithms. Journal of Perinatal Medicine. https://doi.org/10.1515/jpm-2026-0477
Vancouver
Grünebaum A, Chervenak FA. Naegele’s rule is not a 280-day count: generative AI conflates two distinct pregnancy-dating algorithms. Journal of Perinatal Medicine. 2026. doi:10.1515/jpm-2026-0477
BibTeX
@article{Grnebaum2026,
author = {Grünebaum, Amos and Chervenak, Frank A},
title = {Naegele’s rule is not a 280-day count: generative AI conflates two distinct pregnancy-dating algorithms},
journal = {Journal of Perinatal Medicine},
year = {2026},
doi = {10.1515/jpm-2026-0477}
}
EUVIMED