Presented by AWS Machine Learning


As it matures, machine learning has been applied to more and bigger challenges. One of the latest is women’s health tech, a space where research has traditionally lagged, and that few companies have addressed, until now. In 2015, women’s health tech startups raised only $82 million in funding from investment firms. Since then, that number has risen to $1.1 billion.

AI health care company Presagen is one of the companies stepping up in this essential health space, with scalable machine learning that can be used by clinics and patients anywhere in the world.

Their first product is Life Whisperer, a cloud-based web application that uses machine learning built on AWS to offer better outcomes for patients undergoing the fertility treatment IVF, or in vitro fertilization.

“The average age of pregnancy is increasing right at the point when fertility is naturally decreasing,” says Michelle Perugini, co-founder and CEO of Presagen. “Our vision is to improve the success rates for IVF and allow greater access at lower cost for patients, so that they can achieve that goal of having a family sooner.”

More than 8 million babies have been conceived via IVF since the technology was first introduced in 1978. But IVF is not a precise science. For a patient over 35 years old, the chances of delivering a single full-term baby of normal birth weight with IVF are less than 20% per cycle.

That’s where Life Whisperer comes in. With computer vision technology that helps physicians identify the most viable embyos for implantation, the company’s product can reduce the time to pregnancy for patients on average by 15%.

The IVF process, and the Life Whisperer difference

The IVF process has always been complex. One cycle of treatment takes several months and can be extremely stressful for the patient, from the medications they must take at the outset to the extraction of their eggs. Following fertilization in the lab, the embryologist selects the most viable embryos to implant in the patient’s uterus.

This is where Life Whisperer’s machine learning technology comes into play. The success of an IVF cycle relies on selecting the most viable embryos. Unfortunately, the science has been imprecise, meaning the doctor’s choice is subjective, says Perugini. Currently the selection relies on the embryologist conducting a visual assessment of embryos using a microscope. However, there are only a few macro features the clinician can use to grade those embryos.

With machine learning, Life Whisperer helps embryologists choose the most viable embryo or embryos the first time — significantly increasing the chance of a successful IVF cycle.

The Life Whisperer system is relatively simple on the outside, using standard imaging equipment – a powerful camera attached to a microscope. The embryologist sends captured images of the candidate’s embryos to the cloud-based Life Whisperer application, where they are assessed by computer vision algorithms. Trained on thousands of historical IVF cases, the algorithm identifies aspects of the embryo that are most critical in determining viability, but are invisible to the human eye.

In real time, within 10 to 15 seconds, the platform returns a report which helps the doctor select the most viable embryos for implantation. The stronger the score, the more likely the chance of pregnancy for an embryo. The report also gives patients a window into the assessment process being used to guide their treatment.

With their machine learning algorithms, Presagen has improved accuracy in choosing an embryo by 25 to 30% over the current standard of care.

How Life Whisperer works

To build the Life Whisperer machine learning algorithm, Presagen used Pytorch on AWS for deep learning and Amazon EC2 and AWS Fargate for AI training and inference. The algorithm analyzed thousands of embryo images taken from actual IVF cycles, both successful and unsuccessful. The information was extracted from Presagen’s globally linked clinical data platform in the AWS cloud, which securely stores data from clinics around the world.

The goal: to learn how to evaluate both individual features of the embryo that signify its health, as well as how to assess the embryo as a whole. With deep learning and medical data shared from around the world, the custom algorithm continues to learn and refine itself over time. Critical to a global undertaking, the AWS cloud gave Presagen the opportunity to use this global medical data without crossing any patient data privacy laws in the country of origin, says Don Perugini, co-founder of Presagen.

With a distributed cloud solution, the AWS cloud offers a secure, reliable, and accessible platform with the massive amounts of computing power it takes to deliver the system in practice. A distributed cloud system keeps data secure in its origin point while federated learning, the machine learning technique that trains an algorithm across multiple decentralized servers, addresses the critical data privacy and security issues.

In the end however, Michelle Perugini says that AI is not all science — it also requires art.

“The utility of tools to build AI products is only as good as your ability to target and understand the context in which you’re applying the AI – and how to best do that is an art form,” she explains. “It requires multidisciplinary experience and insight from different types of people in different disciplines to create an effective AI.”

The future of IVF and women’s health data

The WHO has called infertility a global health crisis, affecting over 10% of people trying to conceive, and the rates haven’t decreased in the last 20 years.

But the field of reproductive medicine and endocrinology is rapidly growing, and the addition of machine learning to the mix is enabling new innovation. Life Whisperer’s success rates are encouraging, and the company is set to expand into India, the U.K., and Europe.

“Babies conceived with the help of Life Whisperer are being born now,” says Perungini. “Many of those patients went through repeated failed treatments before Life Whisperer, and would not necessarily have had the opportunity to have a child without this type of technology.”

Dig deeper: See more ways machine learning is being used to tackle today’s biggest social, humanitarian, and environmental challenges.


Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. Content produced by our editorial team is never influenced by advertisers or sponsors in any way. For more information, contact sales@venturebeat.com.