The Role of Artificial Intelligence in Automated Embryonic Assessment, Aneuploidy Detection, and Live-Birth Prediction
Artificial intelligence (AI) has increasingly become a cornerstone in various fields, including medicine. One of the most promising applications of AI is in the realm of assisted reproductive technology (ART), specifically in vitro fertilization (IVF). The development and validation of an AI model for evaluating human embryos created through IVF mark a significant advancement in this field. This model leverages images taken at specific time points after insemination, such as day 1 and day 3, and incorporates clinical data like maternal age and BMI to make predictions. This multi-faceted approach aims to enhance the accuracy and reliability of embryo assessment, ultimately improving the success rates of IVF treatments.
The AI model in question is comprised of four distinct modules: grading embryo morphology, predicting blastocyst formation, determining embryo ploidy, and predicting the chances of live birth. Each module serves a specific purpose, collectively contributing to a comprehensive evaluation of the embryo’s potential. Ling Sun from Guangzhou Medical University in China is credited as the creator of this innovative model. The model has been designed for use by IVF clinics and can potentially revolutionize the way embryos are evaluated and selected for transfer. By reducing subjectivity and variability in embryo evaluation, this AI model aims to increase the overall success rates of IVF treatments.
The technology behind this AI model is known as image-based learning. It has been rigorously tested and validated through a series of experiments, demonstrating high accuracy in predicting embryo quality and the likelihood of a successful live birth. This validation process is crucial, as it ensures that the model can be reliably used in clinical settings. The potential impact of this AI model extends beyond just improving success rates; it could also reduce costs for patients and minimize the need for multiple embryo transfers, thereby making IVF treatments more accessible and less stressful for couples.
One of the standout features of this AI model is its ability to detect aneuploidy, a condition where embryos have an abnormal number of chromosomes. Traditionally, detecting aneuploidy involves expensive preimplantation genetic testing (PGT), which can be invasive and stressful for patients. However, the AI model eliminates the need for these invasive biopsies by using time-lapse videos to predict aneuploidy and live-birth outcomes. This non-invasive approach not only reduces costs but also enhances the overall patient experience, making the IVF process less daunting.
The AI model was developed in collaboration with several medical institutions and has been published in a prominent medical journal. This collaborative effort underscores the importance of multi-disciplinary approaches in advancing medical technologies. The model’s accuracy and reliability are the responsibility of the contributing institutions, ensuring that it meets high standards of clinical utility. Ethical considerations have also been taken into account, ensuring that the model is used solely for non-commercial purposes and cannot be modified or adapted without permission.
Another study explored the effectiveness of combining time-lapse culture with AI scoring to improve ongoing pregnancy rates in fresh transfer cycles of single cleavage-stage embryos. This study involved 105 fresh embryo transfer cycles at the Center for Reproductive Medicine. All embryos were cultured using time-lapse technology and scored using an automated AI model known as idascore v2.0. The embryos were then divided into three groups based on their idascore, and the outcomes of clinical treatments, embryonic development, and pregnancy were compared among these groups.
The study found that there were no significant differences in baseline characteristics among the three groups. However, the idascores were significantly higher in group C compared to group B, and higher in group B compared to group A. The average number of high-quality embryos was highest in group C, followed by group B and then group A. Although there was no major difference in ongoing pregnancy rates for single cleavage-stage transfers between group B and group A, there was a slight trend for group B to have higher rates. These findings suggest that combining time-lapse culture with AI scoring may indeed improve ongoing pregnancy rates in single cleavage-stage fresh transfer cycles.
Infertility affects one in six couples globally, significantly impacting their quality of life. Specialized assisted reproductive technologies like IVF are in high demand, but they come with their own set of challenges. IVF is an expensive procedure that relies heavily on the accurate evaluation and selection of the best embryo for implantation. Traditionally, this selection process has been subjective, relying on visual inspection and the physician’s experience. Dr. Kang Zhang and Dr. Ling Sun, along with their collaborators in China, have developed an AI-based system to automate embryo selection, thereby eliminating subjectivity and enhancing accuracy.
The AI system developed by Dr. Zhang and Dr. Sun was published in the Chinese Medical Journal. This system uses multitask learning for embryo morphology assessment at different stages, trained on 19,201 embryo photographs. It includes four modules: embryo grading, blastocyst formation assessment, aneuploidy detection, and live-birth prediction. This comprehensive approach allows for the selection of embryos based on subtle visual features that may be missed by clinicians. The researchers hypothesize that genome aneuploidy can be detected by an AI algorithm due to its influence on embryonic development, further enhancing the model’s utility.
The clinical utility of this AI platform was validated in a prospective cohort, showing superior outcomes compared to traditional PGT-assisted methods. The AI approach was found to be more accurate in predicting embryo aneuploidy than experienced embryologists. Dr. Zhang’s primary aim was to maximize IVF success rates while minimizing the risk of multiple pregnancies. However, the study acknowledges certain limitations, such as the AI model being confined to the Chinese population. Despite these limitations, the use of AI in reproductive medicine shows great promise for improving outcomes and success rates in IVF, offering hope to couples struggling with infertility.
Virtus Fertility Centre Singapore has also embraced the use of AI as a standard of care in their clinic. This initiative aims to improve success rates and shorten the time to a successful pregnancy for patients. The AI technology used at Virtus was developed in collaboration with Harrison.AI and is based on data from over 10,000 embryos cultivated in time-lapse incubators. This data was collected from eight IVF clinics across four countries between 2014 and 2019. The findings were first presented at the American Society of Reproductive Medicine Annual Conference in 2019, highlighting the technology’s potential to revolutionize IVF care.
Virtus Health and Harrison.AI partnered with Vitrolife, a Swedish manufacturer of time-lapse incubation systems, to further develop this technology. Virtus Health has been using AI in IVF care since 2020, making them the first IVF provider in Australia to incorporate this technology into clinical practice. At Virtus Fertility Centre Singapore, embryos are grown in an incubator equipped with time-lapse cameras to monitor their development. Traditionally, embryologists use a standard grading system to assess the embryos’ appearance under a microscope. However, the AI system now analyzes time-lapse imaging data every 10 minutes, assigning each embryo a score to identify those with the highest potential for developing a fetal heart.
The integration of AI into the standard of care at Virtus Fertility Centre Singapore is crucial for enhancing clinical outcomes and achieving successful pregnancies earlier. This can also have a positive impact on the emotional well-being of patients undergoing the IVF process. Established in 2014, VFCS provides personalized fertility care and has access to proven reproductive science techniques and treatment programs through its parent company, Virtus Health. As Virtus Health’s largest self-contained fertility laboratory in the region, VFCS offers comprehensive services and top-tier care for couples dealing with fertility issues. The use of AI in IVF treatments at VFCS exemplifies the potential for AI to improve medical treatments and outcomes, offering hope and support to couples on their journey to parenthood.