A study published in June 2026 introduced Blasto-Net, an artificial-intelligence model designed to perform several tasks in blastocyst assessment during IVF.

The system combines structure segmentation, morphological grading and implantation prediction. On a public dataset it achieved strong segmentation scores and an implantation-prediction F1 score of 80%.

Explainability is part of the goal

The authors also used visualisations to check whether predictions were based on anatomically coherent regions. The work remains experimental and does not replace embryologist assessment or by itself demonstrate improved live-birth rates.

Source: arXiv, Blasto-Net, June 2026.