Resource-constrained rolling-horizon scheduling optimization for male fetal NIPT testing timing: a memetic algorithm-based study

TOP · Springer Nature · 2026-11-11 · DOI: 10.1007/s10729-026-09800-7

Аннотация

In male fetal non-invasive prenatal testing (NIPT), the success of result interpretation depends strongly on the gestational age at blood draw, which makes “when to test” both a clinical and an appointment-management issue. We propose a rolling-horizon scheduling framework that integrates patient-specific success-probability curves with capacity, no-show, retest, and execution dynamics. Using a public cohort of 5,543 singleton male fetal samples, a polynomial-feature logistic regression was fitted to 3,880 records and evaluated in 1,663 held-out patients. The held-out area under the receiver operating characteristic curve was 0.799 (95% confidence interval, 0.736–0.856), with calibration intercept $$-0.020$$ and slope 1.018. The operational layer was evaluated as a prospective operational-feasibility simulation on a constructed 105-day timeline. Across ten common random seeds, the Memetic Algorithm (MA) achieved eligible-patient completion of $$99.09\% \pm 0.23$$ percentage points, mean success probability among completed tests of $$98.44\% \pm 0.03$$ percentage points, and eligible-patient risk coverage of $$98.86\% \pm 0.78$$ percentage points. Under prespecified weights, its composite fitness and custom load-distribution score with a front-loading penalty exceeded those of six executable rules, which retained advantages in waiting time, completion, mean success probability, or risk coverage. On four reduced instances, the mean gap from exact mixed-integer linear programming was $$3.37\% \pm 1.94$$ percentage points under the common linearized surrogate objective. The framework offers an integrated and interpretable approach to coordinating individualized testing timing with dynamic appointment capacity in NIPT services.

Читать у издателя →

Похожие статьи