User login
According to the researchers, such an algorithm is needed due to the increased demand for fertility treatments, as well as the high day-to-day variability in lab workload.
According to the study investigators, predicting retrieval dates in advance for ongoing cycles is of major importance for both patients and clinicians.
“The population requiring fertility treatments, including genetic testing and fertility preservation, has massively increased, and this causes many more cycles and a high day-to-day variability in IVF activity, especially in the lab workload,” said Rohi Hourvitz, MBA, from FertilAI, an Israeli health care company focused on developing technologies that improve fertility treatments.
“We also need to accommodate and reschedule for non-working days, which causes a big issue with managing the workload in many clinics around the world,” added Mr. Hourvitz, who presented the research highlighting AI’s growing role in reproductive medicine.
In addition, AI has recently emerged as an effective tool for assisting in clinical decision-making in assisted reproductive technology, prompting further research in this space, he said.
The new study used a dataset of 9,550 predictable antagonist cycles (defined as having all necessary data) gathered from one lab with over 50 physicians between August 2018 and October 2022. The data were split into two subsets: one for training the AI model and the other for prospective testing.
To train and test the AI model, data from nearly 6,000 predictable antagonist cycles were used. Key factors used for each cycle included estrogen levels, mean follicle size, primary follicle size, and various patient demographics. Other features were considered, but Mr. Hourvitz noted that primary follicle size influenced the algorithm most, “because that is what most of us use when we want to trigger.”
Mr. Hourvitz explained that these patient data were run through an algorithm that produced a graph predicting the most probable date for a cycle retrieval.
“We could accurately predict when those ‘peak days’ were going to be happening in the clinic, and we could also give a pretty good estimate on how many cycles you’re going to have every day,” Mr. Hourvitz said, explaining that this information could help clinics more efficiently allocate resources and manage patients.
According to Mr. Hourvitz, the predictions derived from this study could improve various aspects of fertility treatments and related procedures, including better staff planning and caseload management in IVF labs, as well as higher-quality eggs at retrieval. Patients would have a clearer timeline for their treatment cycles.
Nikica Zaninovic, PhD, MS, director of the embryology lab at Weill Cornell Medical College, New York City, cautioned that the new findings are not yet ready for clinical application but emphasized the importance of more AI research focusing on the quality of oocytes, not only embryos.
“We’re so focused on the end of the process: the embryo,” Dr. Zaninovic, who was not involved in the research, said in an interview. “I think the focus should be on the beginning – the quality of eggs and sperm, not just the quantity – because that’s what the embryos will depend on.”
He noted the increasing numbers of young women in the United States undergoing egg freezing.
“Cornell is the largest academic IVF center in the United States; 20%-30% of all of the patients that we treat are actually freezing their eggs,” he said. “It’s a huge population.”
“When they come to us, they ask how many eggs they’ll need to guarantee one or two children in the future,” Dr. Zaninovic continued. “We don’t have that answer, so we always tell them [we’ll retrieve] as many as we can. That’s not the answer; we need to be more precise. We’re still lacking these tools, and I think that’s where the research will go.”
The study was funded by FertilAI. Mr. Hourvitz is a shareholder and CEO of FertilAI. Dr. Zaninovic is president of the AI Fertility Society.
A version of this article appeared on Medscape.com.
According to the researchers, such an algorithm is needed due to the increased demand for fertility treatments, as well as the high day-to-day variability in lab workload.
According to the study investigators, predicting retrieval dates in advance for ongoing cycles is of major importance for both patients and clinicians.
“The population requiring fertility treatments, including genetic testing and fertility preservation, has massively increased, and this causes many more cycles and a high day-to-day variability in IVF activity, especially in the lab workload,” said Rohi Hourvitz, MBA, from FertilAI, an Israeli health care company focused on developing technologies that improve fertility treatments.
“We also need to accommodate and reschedule for non-working days, which causes a big issue with managing the workload in many clinics around the world,” added Mr. Hourvitz, who presented the research highlighting AI’s growing role in reproductive medicine.
In addition, AI has recently emerged as an effective tool for assisting in clinical decision-making in assisted reproductive technology, prompting further research in this space, he said.
The new study used a dataset of 9,550 predictable antagonist cycles (defined as having all necessary data) gathered from one lab with over 50 physicians between August 2018 and October 2022. The data were split into two subsets: one for training the AI model and the other for prospective testing.
To train and test the AI model, data from nearly 6,000 predictable antagonist cycles were used. Key factors used for each cycle included estrogen levels, mean follicle size, primary follicle size, and various patient demographics. Other features were considered, but Mr. Hourvitz noted that primary follicle size influenced the algorithm most, “because that is what most of us use when we want to trigger.”
Mr. Hourvitz explained that these patient data were run through an algorithm that produced a graph predicting the most probable date for a cycle retrieval.
“We could accurately predict when those ‘peak days’ were going to be happening in the clinic, and we could also give a pretty good estimate on how many cycles you’re going to have every day,” Mr. Hourvitz said, explaining that this information could help clinics more efficiently allocate resources and manage patients.
According to Mr. Hourvitz, the predictions derived from this study could improve various aspects of fertility treatments and related procedures, including better staff planning and caseload management in IVF labs, as well as higher-quality eggs at retrieval. Patients would have a clearer timeline for their treatment cycles.
Nikica Zaninovic, PhD, MS, director of the embryology lab at Weill Cornell Medical College, New York City, cautioned that the new findings are not yet ready for clinical application but emphasized the importance of more AI research focusing on the quality of oocytes, not only embryos.
“We’re so focused on the end of the process: the embryo,” Dr. Zaninovic, who was not involved in the research, said in an interview. “I think the focus should be on the beginning – the quality of eggs and sperm, not just the quantity – because that’s what the embryos will depend on.”
He noted the increasing numbers of young women in the United States undergoing egg freezing.
“Cornell is the largest academic IVF center in the United States; 20%-30% of all of the patients that we treat are actually freezing their eggs,” he said. “It’s a huge population.”
“When they come to us, they ask how many eggs they’ll need to guarantee one or two children in the future,” Dr. Zaninovic continued. “We don’t have that answer, so we always tell them [we’ll retrieve] as many as we can. That’s not the answer; we need to be more precise. We’re still lacking these tools, and I think that’s where the research will go.”
The study was funded by FertilAI. Mr. Hourvitz is a shareholder and CEO of FertilAI. Dr. Zaninovic is president of the AI Fertility Society.
A version of this article appeared on Medscape.com.
According to the researchers, such an algorithm is needed due to the increased demand for fertility treatments, as well as the high day-to-day variability in lab workload.
According to the study investigators, predicting retrieval dates in advance for ongoing cycles is of major importance for both patients and clinicians.
“The population requiring fertility treatments, including genetic testing and fertility preservation, has massively increased, and this causes many more cycles and a high day-to-day variability in IVF activity, especially in the lab workload,” said Rohi Hourvitz, MBA, from FertilAI, an Israeli health care company focused on developing technologies that improve fertility treatments.
“We also need to accommodate and reschedule for non-working days, which causes a big issue with managing the workload in many clinics around the world,” added Mr. Hourvitz, who presented the research highlighting AI’s growing role in reproductive medicine.
In addition, AI has recently emerged as an effective tool for assisting in clinical decision-making in assisted reproductive technology, prompting further research in this space, he said.
The new study used a dataset of 9,550 predictable antagonist cycles (defined as having all necessary data) gathered from one lab with over 50 physicians between August 2018 and October 2022. The data were split into two subsets: one for training the AI model and the other for prospective testing.
To train and test the AI model, data from nearly 6,000 predictable antagonist cycles were used. Key factors used for each cycle included estrogen levels, mean follicle size, primary follicle size, and various patient demographics. Other features were considered, but Mr. Hourvitz noted that primary follicle size influenced the algorithm most, “because that is what most of us use when we want to trigger.”
Mr. Hourvitz explained that these patient data were run through an algorithm that produced a graph predicting the most probable date for a cycle retrieval.
“We could accurately predict when those ‘peak days’ were going to be happening in the clinic, and we could also give a pretty good estimate on how many cycles you’re going to have every day,” Mr. Hourvitz said, explaining that this information could help clinics more efficiently allocate resources and manage patients.
According to Mr. Hourvitz, the predictions derived from this study could improve various aspects of fertility treatments and related procedures, including better staff planning and caseload management in IVF labs, as well as higher-quality eggs at retrieval. Patients would have a clearer timeline for their treatment cycles.
Nikica Zaninovic, PhD, MS, director of the embryology lab at Weill Cornell Medical College, New York City, cautioned that the new findings are not yet ready for clinical application but emphasized the importance of more AI research focusing on the quality of oocytes, not only embryos.
“We’re so focused on the end of the process: the embryo,” Dr. Zaninovic, who was not involved in the research, said in an interview. “I think the focus should be on the beginning – the quality of eggs and sperm, not just the quantity – because that’s what the embryos will depend on.”
He noted the increasing numbers of young women in the United States undergoing egg freezing.
“Cornell is the largest academic IVF center in the United States; 20%-30% of all of the patients that we treat are actually freezing their eggs,” he said. “It’s a huge population.”
“When they come to us, they ask how many eggs they’ll need to guarantee one or two children in the future,” Dr. Zaninovic continued. “We don’t have that answer, so we always tell them [we’ll retrieve] as many as we can. That’s not the answer; we need to be more precise. We’re still lacking these tools, and I think that’s where the research will go.”
The study was funded by FertilAI. Mr. Hourvitz is a shareholder and CEO of FertilAI. Dr. Zaninovic is president of the AI Fertility Society.
A version of this article appeared on Medscape.com.
FROM ASRM 2023