The dimensions of healthcare has widened with the adoption of modern thinking and the acceptance of medical innovation. Over the past few years, there has been an increasing utility and a growth of AI and digital technologies in reproductive medicine for new standardization, automation, and precision.
With automatic annotation of embryo development, embryo grading, and embryo selection for implantation is evolving as one of the best uses of Artificial intelligence applications. The selection process of the best embryo from the larger cohort of the fertilized egg is majorly by the embryologist by grading the embryos.
“The embryo selection process is on the morphology and the photography of that particular embryo. Now, these grading because of the manual intervention will change based on embryologist to the embryologist, lab to lab and, city to city. Thus, the standardizing and automizing the process using AI becomes very, very important,” said, Nitiz Murdia, the co-founder and Director of Marketing & Embryology at Indira IVF Group while, explaining the grading of the embryos in acquiring the best suitable matches.
Multiple organizations are opting towards this development of the Artificial intelligence process to objectivize and standardize these embryos grading to predict the higher implantation potential embryo compared to any X-Y-Z or the other embryo. According to Murdia, the analysis of the data sets plays a vital role in selecting the embryo with higher efficiency.
“If I am an embryologist, I can only do a particular number of cases, but the AI prediction has at least one lakh cases in his database. It can perform a better prediction compared to me, and it has more experience compared to me. I might have experienced it for ten years but, the AI will have a combined experience of 150 years. So that is the biggest advantage I think with this kind of system that exists,” Murdia said.
Overcoming the Issues with AI
As AI increases the evaluation and selection for the embryos, the major problem lies in the grading performed by the individual embryologist. Each system delivers the different results of the same embryo image feed into the system by the multiple embryologists. Thus, the focus has shifted to machine learning which automatically device this grading system with data, not from a particular clinic or a country, but data from all the labs.
“People have developed the system and are still using the system. If you talk about the accuracy part of it, it can predict up to 70% accuracy. So, there is still a lot of room left for improvement of this particular technology and, I think there are a lot of variables which still exist, which people need to solve,” said Murdia.
AI for Ovary and Uterus Evaluation: Other than embryo evaluation and selection, AI is helping to predict the correct gonadotropin levels in the ovary by using ultrasound to feed that image into an AI system. It predicts the volume of the ovary to suggest a particular dose of gonadotropin injection, just like cancer therapy. The process enables stimulating the ovaries to produce eggs as per the patient.
With not discounting on the uterus, AI identifies the patient’s uterine conditions for implantation by fitting the ultrasound image into the AI system. It predicts the endometrium state for implantation and suggests a suitable surgery if required.
Miscarriages and Pre-Genetic Testing: As the miscarriage’s cases are on the rise, AI models are used to locate the embryo with higher chances of miscarriages by reversing the data feed before the implantation. “The process contains 77% accuracy of avoiding the embryos as it is done on a small set,” added Murdia. He pointed out that PGT or Pre-Genetic testing is the way-beyond technology that screens chromosomes and eliminates abnormal embryos. The technology helps the patients to conceive a healthy child despite experiencing a failed implantation circumstances in the past.
AI in embryologist efficiency
Indira IVF Group is working on the model to study and compare the AI and embryologist predictions for long-term purposes. The data management has helped the organization to strengthen the system and train the embryologist.
“We have installed all the electronic witness systems across all our labs. We have tried to capture that data on the cloud and made dashboards. The electronic system, all the process is the SOP and, the process for each level is finalized with steps. It is giving us good data in terms of every single embryologist’s performance in every lab. It helps to evaluate the duration taken by the embryologist for a particular step. It is helping us to train the embryologist who is taking more time to do a particular process. So, this has helped us in that particular way,” said Murdia.
