For decades, astrophysicists have used increasingly sophisticated ways to search the universe for objects buried in the vastness of space, and now that technology has been adapted to look for something here on Earth, much smaller but a key to life.

The Sperm Tracking and Recovery System, or STAR, is a technology that combines high-powered imaging, AI, robotics and microfluidics to find viable sperm in men with extraordinarily low or undetectable sperm counts.

“We’re using the same technologies that are used to search for life in the universe to help create new life right here on Earth,” says Dr. Zev Williams, director of the Columbia University Fertility Center. “It’s truly awe-inspiring to now be able to help patients build the family they always dreamed of.”

Dr. Williams and Dr. Hemant Suryawanshi, an assistant professor of reproductive sciences at Columbia, have been leading the research team. The center and New York-Presbyterian Hospital operate as a unified academic medical system. Dr. Williams is also chief of the Division of Reproductive Endocrinology and Infertility at New York-Presbyterian.

For men with azoospermia, there is no measurable sperm in their semen. The condition can occur because the body is not producing sperm, because a blockage is preventing sperm from reaching the semen, or because hormonal problems are interfering with sperm production. Azoospermia affects about 1% of all men and about 10% to 15% of infertile men, but a diagnosis does not necessarily mean biological fatherhood is impossible. In some cases, viable sperm are still being produced inside the testicles — just in extremely small numbers.

Finding those rare cells is where the STAR system shines. It was developed over five years, and the first confirmed clinical pregnancy in which it was utilized was announced in 2025. A couple who had spent 19 years trying to conceive was finally able to using sperm recovered by the system. The STAR system scanned 2.5 million images of the man’s samples and located two viable sperm cells.

“With our method, many men who were previously told they have no chance of having a biological child now have that chance,” Dr. Williams said.

Until recently, physicians had limited options. One is surgery to retrieve sperm directly from the testes. The procedure can be painful and unsuccessful and can lead to complications, including vascular problems, inflammation and temporarily reduced testosterone levels.

Another approach involves processing semen through a centrifuge and having specially trained technicians painstakingly search the concentrated sample under a microscope. The technique is time-consuming, expensive and available at only a small number of laboratories. The process also can damage sperm, potentially rendering it unusable.

The STAR team brought together researchers and clinicians with expertise in machine learning, robotics, microfabrication and reproductive medicine.

First, high-powered imaging scans the entire sample, generating more than 8 million images in less than an hour. AI analyzes those images to identify potential sperm cells. But finding the sperm is only half the problem. The cell must then be isolated without damaging it. To accomplish that, the team developed a microfluidic chip containing a network of channels as thin as a human hair. The system separates the portion of the sample containing the rare sperm cells. Once a sperm is identified, a robot can retrieve it within milliseconds. The process eliminates the centrifugation, lasers, dyes and other harsh treatments sometimes used in conventional sperm-recovery techniques.

“By avoiding centrifugation, lasers, dyes or other harsh systems, the STAR system provides a gentler environment that minimizes stress on sperm cells, giving them the best possible chance of being identified and maintaining their viability,” Dr. Williams said.

The recovered sperm can then be injected directly into an egg through in vitro fertilization, or IVF, or frozen for future use.

The promise of AI in fertility medicine extends beyond finding elusive sperm. A 2025 review published in Cureus and available through the National Library of Medicine’s PubMed Central examined the expanding use of AI across assisted reproductive technology. Researchers described applications that include using machine-learning algorithms to help personalize ovarian stimulation, predict the optimal time for triggering ovulation, evaluate sperm morphology and motility, estimate sperm velocity, assess embryos and predict IVF outcomes.

AI systems can analyze microscopic images to identify abnormalities in sperm that may be difficult or time-consuming for embryologists to detect. Other systems analyze sperm movement from microscope video, attempting to distinguish progressive, nonprogressive and immobile sperm and estimate how quickly individual sperm are moving.

In the IVF process, algorithms are being studied to help determine which embryos have the greatest potential for implantation, assess the uterine lining and predict the likelihood of a successful pregnancy based on large amounts of clinical data.

The implications are significant because reproductive medicine is a numbers game.

“The good news is that, with IVF, you only need one healthy sperm to be able to create an embryo,” Dr. Williams said.