Ask Onix
Updated 1 May 2026 to correct an institutional affiliation.
AI system detects elusive sperm, enabling pregnancy after years of infertility
A New Jersey couple, Penelope and Samuel, celebrated their first pregnancy in November 2025 after a pioneering artificial-intelligence platform identified viable sperm in Samuel's tissue-something doctors had previously ruled out. Samuel, diagnosed with Klinefelter syndrome, was told he had only a 20 % chance of fathering a biological child.
"His face was just a wave of emotion. He cried... to finally get to that point, because it took so much effort, time and research," Penelope recalled.
The Star system: how it works
Developed at Columbia University, the Sperm Track and Recovery (Star) system employs machine learning to scan microscopic images at 300 frames per second. It distinguishes rare sperm cells from cellular debris in real time, achieving 100 % sensitivity-meaning it can detect a single sperm if one exists.
A robotic microfluidic chip then isolates the sperm within milliseconds, separating it from the surrounding fluid. In controlled tests, Star located 40 times more sperm than trained human technicians could find manually.
From diagnosis to delivery
Samuel's Klinefelter syndrome, a genetic condition marked by an extra X chromosome, typically results in azoospermia-an absence of sperm in ejaculate. After nine months of hormone therapy, he underwent testicular extraction surgery. The extracted tissue was rushed to Columbia's andrology laboratory, where Star identified eight viable sperm cells.
Penelope's eggs were retrieved the same day. One of the eight sperm successfully fertilised an egg, forming a blastocyst that led to the couple's ongoing pregnancy. Their son is due at the end of July 2026.
Expanding access and ethical questions
Since the first Star-assisted birth in late 2025, Columbia's fertility centre has treated 175 patients, finding sperm in nearly 30 % of cases where conventional methods had failed. The waiting list now spans hundreds of hopeful couples worldwide.
Yet experts caution that larger clinical trials are needed to confirm long-term safety and efficacy. Siobhan Quenby, professor of obstetrics at the University of Warwick, emphasised that while the technology is "very exciting," its value must be validated across broader patient populations.
Data privacy, accountability, and the risk of overpromising outcomes also remain key concerns. "Couples on long fertility journeys can be vulnerable to expensive treatments of unproven value," Quenby noted.
Beyond sperm detection: AI's growing role in fertility
Star is not the only AI application transforming reproductive medicine. Machine learning now personalises hormone dosages in ovarian stimulation, while deep learning aids in selecting viable embryos. However, experts stress the need for rigorous oversight as these tools evolve.
A future with possibilities
For Samuel and Penelope, the technology has already reshaped their lives. "We want another child," Samuel said, "and now we have hope where there was none before."
"It's just finding something where we couldn't see it before," said Zev Williams, director of Columbia University Fertility Center.
