Can machine learning identify a euploid miscarriage without PGT-A?

A 2026 study used a machine learning model to predict whether a miscarriage was euploid or aneuploid using routine medical data, without PGT-A, and showed moderate ability to distinguish between the two while identifying several associated factors.

Most early pregnancy losses are caused by chromosomal abnormalities (aneuploidy), which is mainly linked to increasing age. Euploid losses, on the other hand, may point to underlying maternal, paternal, immunologic, or metabolic factors that could potentially be addressed.

In practice, it’s hard to know which category a loss falls into. As a result, doctors usually rely on the number of previous losses as a way to estimate risk, which could delay treatment if there’s an underlying issue.

A new study by Banasik et al. (2026) examined whether basic clinical information could predict whether a miscarriage was euploid or aneuploid, without using PGT-A.

The study included over 1,000 women with confirmed pregnancy loss from the Copenhagen Pregnancy Loss cohort. These were mostly unassisted pregnancies from the general population and none did PGT-A. After the miscarriage, researchers used a blood test similar to NIPT to analyze fetal DNA in the mother’s blood and classify each loss as euploid or aneuploid.

They collected 45 common clinical details, such as maternal age, BMI, gestational age, blood pressure, pregnancy history, smoking, and vitamin use. A machine learning model was then trained to classify each loss using only this information.

---

What the model showed

They found that the model was moderately able to distinguish between euploid and aneuploid losses. Its overall predictive accuracy (AUC) was 0.69. An AUC of 1.0 represents perfect prediction, while 0.5 represents chance. An AUC of 0.69 indicates moderate ability to distinguish between the two.

The most important predictors for a euploid loss were ranked:

1. Maternal age. Younger age was associated with a higher chance of euploid loss. This was the strongest predictor.
2. Gestational age calculated from last menstrual period (LMP). Higher gestational age by LMP was associated with euploid loss.
3. Gestational age estimated from ultrasound crown–rump length (CRL). Higher ultrasound-based gestational age was associated with euploid loss.
4. Paternal age. Younger paternal age was associated with euploid loss. Note that this is likely influenced by maternal age, since partners are often close in age.
5. Vitamin D supplementation during pregnancy. Vitamin D use was associated with a lower chance of euploid loss.
6. Type of miscarriage diagnosis. Spontaneous miscarriage and blighted ovum were more likely to be euploid than missed miscarriage.
7. Paternal BMI.
8. Vitamin E supplementation during pregnancy. Vitamin E use was associated with a lower likelihood of euploid loss.
9. Maternal systolic blood pressure.
10. Paternal diastolic blood pressure.
11. Maternal pulse (heart rate).
12. Maternal BMI.
13. Maternal number of sexual partners.
14. Paternal pulse (heart rate).
15. Paternal smoking.

Maternal age was no surprise here: as age increases, losses were more likely to be aneuploid, so younger patients were at a higher risk of a euploid loss than an aneuploid loss.

Gestational age patterns were also important: losses that occurred later in pregnancy were more likely to be euploid, while earlier losses were more often aneuploid.

Vitamin D and vitamin E supplementation were associated with a lower probability of euploid loss in this study. This doesn’t necessarily mean the vitamins prevented miscarriage, but could signal underlying health behaviors.

Conclusion

This model was built using women from the general population who had already experienced a miscarriage. It showed a moderate ability to distinguish between euploid and aneuploid losses.

There were some interesting predictors, but most of the strongest signals, like maternal age and gestational age at the time of loss, aren’t factors that patients can change to potentially lower their risk of a euploid miscarriage. For other factors, like vitamin use, BMI and heart rate measurements, it’s not clear if these directly cause miscarriage or instead are related to general health patterns linked to euploid loss.

At this stage, the model shows that medical record data contains some factors that are associated with euploid miscarriage, but it doesn’t really change how that risk can be prevented. Future studies that look more closely at IVF patients or those with recurrent pregnancy loss might lead to more actionable factors for patients.

Want to read more about euploid losses?

Reference

Banasik K, Tummoszeit I, Bliddal S, Vexø LE, Madsen EP, Hartwig TS, Werge L, Johnsen MG, Gruhn JR, Chan AC, Løkkegaard EC, Ambye L, Jørgensen FS; COPL Consortium; Hoffmann ER, Westergaard D, Nielsen HS. Developing and Validating a Machine Learning Approach for Prediction of Euploid Pregnancy Loss in the Copenhagen Pregnancy Loss Study. Fertil Steril. 2026 Jan 31:S0015-0282(26)00055-5. doi: 10.1016/j.fertnstert.2026.01.026. Epub ahead of print. PMID: 41628849.

If you liked this post and want to support what I do, please consider a paid subscription, Patreon or donate through PayPal!

 

---

About Embryoman

Embryoman (Sean Lauber) is a former embryologist and the founder of Remembryo, an IVF research and fertility education website. After working in an IVF lab in the US, he returned to Canada and now focuses on making fertility research more accessible. He holds a Master’s in Immunology and launched Remembryo in 2018 to help patients and professionals make sense of IVF research. Sean shares weekly study updates on Facebook, Instagram, and Reddit regularly. He also answers questions on Reddit or in his private Facebook group.

---