Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-0674
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-08-07T06:25:52.515744Z
status: published
lens: trace
sector: health
headline: Machine learning-enabled prediction of ART pregnancy outcomes: a systematic review and meta-analysis
dek: Objective To systematically evaluate the diagnostic accuracy and methodological quality of machine learning (ML) prediction models for pregnancy outcomes after assisted reproductive technology (ART). Methods PubMed, Embase, the Cochrane Library, IEEE Xplore, MEDLINE, ClinicalTrials.gov, CNKI, Wanfang, and VIP were searched from inception to July 2026. Eligible studies developed or validated ML models to predict clinical pregnancy or live birth after ART. For studies reporting complete 2 × 2 contingency data, poo…
gain_title: Machine learning models achieved moderate diagnostic accuracy for predicting clinical pregnancy or live birth after assisted reproductive technology, with pooled sensitivity 0.737 and specificity 0.789.
problem_title: The evidence base for ML prediction of ART outcomes is limited by substantial heterogeneity and frequent high or unclear risk of bias, requiring prospective multi-center external validation before clinical use.
trace_subject: machine learning prediction of pregnancy outcomes after assisted reproductive technology
gain_reading: Machine learning models achieved moderate diagnostic accuracy for predicting clinical pregnancy or live birth after assisted reproductive technology, with pooled sensitivity 0.737 and specificity 0.789.
gain_evidence: ML models show moderate diagnostic accuracy for predicting ART pregnancy outcomes | The pooled sensitivity was 0.737 (95% CI, 0.662-0.799) and the pooled specificity was 0.789 (95% CI, 0.709-0.851)
problem_reading: The evidence base for ML prediction of ART outcomes is limited by substantial heterogeneity and frequent high or unclear risk of bias, requiring prospective multi-center external validation before clinical use.
problem_evidence: evidence base is limited by substantial heterogeneity and frequent high or unclear risk of bias | Overall risk of bias was low in 1 study (5.0%), high in 8 studies (40.0%), and unclear in 11 studies (55.0%) | prioritize prospective multi-center external validation before clinical implementation
quick_read: A systematic review and diagnostic meta-analysis of 20 studies, 14 in quantitative synthesis, evaluated machine learning models to predict clinical pregnancy or live birth after assisted reproductive technology. As of the August 2026 publication, pooled sensitivity was 0.737 and specificity 0.789 with a DOR of 10.49 and acceptable discrimination on SROC, but heterogeneity was very high.

The findings matter because ART counseling could benefit from accurate outcome prediction, yet only 5% of studies had low risk of bias per PROBAST while 40% were high and 55% unclear. The authors state that adherence to TRIPOD+AI, reporting of calibration and clinical utility, and prospective multi-center external validation are needed before clinical implementation, leaving real-world benefit unproven.
limitation: Evidence base limited by substantial heterogeneity (I2 97.7% and 99.0%) and frequent high or unclear risk of bias, with subgroup findings driven by a single prospective study.
tag: Automated dual reading
key_points: Systematic review included 20 studies, 14 contributed to diagnostic meta-analysis of ML models predicting clinical pregnancy or live birth after ART. | Pooled diagnostic performance: sensitivity 0.737, specificity 0.789, DOR 10.49 with SROC indicating acceptable discrimination. | Risk of bias per PROBAST: low in 1 study (5.0%), high in 8 studies (40.0%), unclear in 11 studies (55.0%) with I2 97.7% and 99.0%. | Authors conclude prospective multi-center external validation and TRIPOD + AI reporting needed before clinical implementation.
rundown: Search covered PubMed, Embase, Cochrane Library, IEEE Xplore, MEDLINE, ClinicalTrials.gov, CNKI, Wanfang, and VIP from inception to July 2026 for ML models predicting clinical pregnancy or live birth after ART. Twenty studies were systematically reviewed and 14 with complete 2x2 data were meta-analyzed using random-effects diagnostic methods.

Pooled results showed sensitivity 0.737 and specificity 0.789 with DOR 10.49. Exploratory subgroup analyses found comparable performance for clinical pregnancy versus live birth and no robust difference by algorithm type, center type, or validation status; a significant difference by study design was driven by a single prospective study.
sources:
- peer_reviewed | Journal of Assisted Reproduction and Genetics | https://doi.org/10.1007/s10815-026-03988-x | 2026-08-06
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

b9126d239081962f23332e7304b224b04c36cd869de90c66c300576a4c99ed72

previous

0000000000000000000000000000000000000000000000000000000000000000