Startup hopes to use its IVF-predictive software against coronavirus
search

Startup hopes to use its IVF-predictive software against coronavirus

Israel’s Embryonics uses AI to help fertility specialists choose best embryos for womb implantation and develops personalized hormonal treatments for IVF patients

An illustrative image of an embryo (Zffoto; iStock by Getty Images)
An illustrative image of an embryo (Zffoto; iStock by Getty Images)

An early-stage Israeli startup that uses artificial intelligence and machine learning to help fertility specialists select the best embryos for implantation into the womb, and develops personalized hormonal treatments for in vitro fertilization (IVF) patients, is now hoping that a similar technology can be used to help diagnose and treat COVID-19 patients.

“We want to see if we can use our technology to help with the treatment, the diagnosis and progress of COVID-19 patients and other diseases,” said Yael Zamir, the co-founder and CEO of Embryonics, a startup she set up with partners in 2018. COVID-19 is the illness caused by the coronavirus, which has killed over 400,000 people worldwide.

The startup has developed two algorithms. One is a commercially available patented AI-technology that enables trawling through healthcare databases, studying tens of thousands of embryos and their implant-success rate, and then predicting which embryos will be the most likely to succeed.

The other analyzes clinical data using newly invented geometric deep learning technology to personalize hormonal treatments for IVF patients.

Dr Yael Zamir, the co-founder and CEO of Embryonics (Courtesy)

Geometric deep learning is a new field of machine learning that can gain insights from complex data, like graphs and multi-dimensional points, and, according to the company, “has shown big promise in other areas as it outperforms the classical widely used AI algorithms.”

Embryonics now wants to use geometric deep learning to predict which COVID-19 patients will need to be put on ventilators, which needs to be kept in the hospital, and which can be treated at home.

“We believe that our geometrical methods that have shown big promise in IVF could be applicable here,” said Zamir.

The startup has started working with Shaare Zedek Medical Center in Jerusalem — the hospital that has treated the second-largest number of COVID-19 patients in Israel — to get access to its database of virus-stricken patients. The idea is to build new predictive software based on the same geometric deep-learning tools it uses for personalizing hormonal treatments for IVF patients.

Working with the hospital will allow the startup to adapt its embryo-selecting algorithm so it will be able to examine hospital data, looking at patients’ characteristics, from whether they are smokers to where they live, their age, diseases and other clinical parameters to make predictions about their outcomes: at what day will their illness peak? Will their illness be easier than for others?

“If patients are similar, we want to team them up and compare data points,” Zamir said. When a new patient arrives, she explained, the system should be able to look for what happened to patients that had similar traits and make forecasts and treatment recommendations.

Shaare Zedek is the first hospital to use the platform as a pilot for COVID-19 patients, she said.

Medical personnel in protective gear bring a patient suspected of having the coronavirus to Shaare Zedek Medical Center in Jerusalem on April 30, 2020. (Nati Shohat/Flash90)

Renana Ofan, the director of the Shaare Zedek Innovation Center, said the hospital has granted Embryonics access to patients’ data under strict privacy regulations, and is working closely with the firm to make sure the data is clear, reliable and relevant to their needs.

“We must have collaborations with tech firms” to make much-needed breakthroughs, she said. “We bring our clinical knowledge, and they bring the technology. We are happy to join forces with anyone who wants to, and have very many of these kinds of collaborations.”

Bringing transparency to IVF

Zamir set up Embryonics with her co-founders to bring transparency to the field of in vitro fertilization, in which eggs are fertilized with sperm in a lab.

The human fertility field “has many gray areas,” said Zamir. “Traditional fertility treatments don’t have good answers and this causes a lot of suffering for couples.”

Innovative technologies can be used to answer complicated questions, she said, like how to choose the best embryos to implant into the womb after the IVF process, a process that “has not changed in years,” said Zamir.

Illustrative. In vitro fertilization (IVF) of an egg cell. (iStock by Getty Images/ man_at_mouse)

Doctors rely on observation through a microscope, looking for the rate of cell division, cell symmetry, and certain tables of rules regarding the morphology of the embryo, she said, noting that the IVF success rate is only around 30 percent.

What the startup has done is bring objectivity to the process, Zamir explained, with an algorithm that is able to look at tens of thousands of embryos and their implant-success rate, and then predict which embryos are the most likely to succeed.

The firm has been testing its software platform at clinics in Israel and Europe, including Kaplan medical center of Rehovot, Carmel hospital of Haifa, Nadiya international center of Kyiv, Ukraine, and other clinics in France, Spain, China, Malaysia and the United States, Zamir said.

Zamir said the results show that the algorithm can “outperform embryologists.”

The firm is also expanding the scope of the algorithm to help set out the best protocol to improve the process of hormonal treatments by personalizing recommendations to suit candidates in order to minimize side effects and shorten time to pregnancy.

Embryonics, which has raised $4 million in seed money from a US-based VC and in grants from the Israel Innovation Authority, is in the process of getting US Food and Drug Administration approval for its IVF software, Zamir said, and is starting to sign its first contracts with clients.

read more:
comments