The AI Bela system, used in assessing IVF (in vitro fertilization) embryos, applies artificial intelligence to improve the accuracy and efficiency of embryo evaluation, a critical step in IVF treatment. The goal is to enhance the selection of embryos with the highest potential for successful implantation, pregnancy, and live birth. BELA does not need to consider embryologists' subjective assessments of embryos. It offers an objective, generalizable measure. It was developed by Weill Cornell Medicine Institute
The way it functions is as follows:
1. Embryo Imaging and Time-Lapse Monitoring
• AI Bela integrates with time-lapse imaging systems that continuously capture images of embryos as they develop in the laboratory.
• These images provide data on the embryos’ growth patterns, morphology (structure), and development stages, which are crucial indicators of viability.
• The system may use time-lapse incubators that provide a stable environment while documenting key stages like fertilization, cell division, and blastocyst formation.
2. Data Analysis Using AI Algorithms
• The AI system analyzes thousands of images and datasets from the embryos’ development, detecting patterns invisible to the human eye.
• It uses deep learning algorithms trained on vast datasets of embryo images, coupled with clinical outcomes (e.g., successful pregnancies), to make predictions.
• The system evaluates key features, such as:
• Morpho kinetics: The timing and sequence of cell divisions.
• Morphology: The appearance and structure of the embryo (e.g., symmetry, fragmentation, blastocyst quality).
• Cell Count: Accuracy of cell division stages and the number of cells in key stages.
3. Scoring and Ranking Embryos
• AI Bela assigns each embryo a viability score based on its analysis. This score reflects the embryo’s potential for implantation, pregnancy success, and live birth.
• The AI also compares embryos in a cohort, helping embryologists rank them based on their likelihood of success.
• Some AI systems might also detect anomalies that indicate potential genetic issues or abnormalities.
4. Reducing Subjectivity in Embryo Selection
• Traditionally, embryologists have relied on manual observation and experience to assess embryos, which can introduce subjectivity.
• AI systems like Bela aim to standardize this process by providing objective, data-driven assessments, improving consistency in decision-making.
• AI can also reduce the risk of human error and fatigue from the demanding task of embryo observation.
5. Integration with Genetic Testing
• AI can complement or integrate with preimplantation genetic testing (PGT), where embryos are screened for genetic abnormalities. AI Bela may help prioritize which embryos undergo testing by predicting viability beforehand.
6. Outcome Prediction
• The AI system uses historical data (e.g., previous pregnancies and clinical outcomes) to improve the prediction of success rates for each individual embryo.
• Machine learning algorithms continuously learn and improve from new data, refining the model over time.
By combining these advanced imaging and data analytics techniques, AI systems like Bela enhance embryo selection, aiming to increase IVF success rates while potentially reducing costs by optimizing embryo transfers.
