Assisted fertilization

First baby thanks to artificial intelligence and robots: a breakthrough in male infertility

The technique, developed at Columbia University, detects rare spermatozoa in men with azoospermia, opening up new perspectives in reproductive medicine

by Francesca Cerati

3 November 2025

3' min read

Translated by AI

Versione italiana

3' min read

Translated by AI

Versione italiana

An unprecedented result comes from the United States: a couple who had been trying in vain to have a child for 19 years managed to achieve pregnancy thanks to an artificial intelligence (Ai) and robotics system capable of identifying and recovering extremely rare spermatozoa.

The extraordinary feat is described in the prestigious journal The Lancet and represents the world's first pregnancy achieved through an Ai-guided IVF method.

The method, called Star (an acronym for Sperm tracking and recovery), was developed by a research team from Columbia University Fertility Center in New York, directed by Zev Williams and coordinated by Hemant Suryawanshi. The goal: to find a non-invasive solution to the most severe cases of male infertility, particularly those due to azoospermia, a condition in which the seminal fluid contains very few or no spermatozoa.

"Our team brings together experts in advanced imaging, microfluidics and reproductive endocrinology to address every step necessary to find and isolate viable spermatozoa," explains Suryawanshi, first author of the study.

A sea of cells, but no sperm

For many men with severe infertility, the sperm sample may appear normal, but under the microscope it reveals only a 'sea of cellular debris', with no visible spermatozoa. Until now, solutions were limited to invasive surgical procedures to extract spermatozoa directly from the testicles, with often disappointing outcomes and risks of complications.

The Star system radically changes its approach. Using a combination of ultra-high resolution imaging technology and deep learning algorithms, it can scan millions of images of a sperm sample within minutes, automatically recognising the presence of live sperm.

Once identified, a microchip with channels thinner than a hair isolates the useful part of the sample, and a precision robot extracts the selected sperm in milliseconds. This can then be used to fertilise an oocyte or stored at low temperatures for future use.

The clinical case: 19 years of waiting

The technique was tested on a couple who had been looking for a child for almost 20 years. The woman, 37, had gone through 11 cycles of ovarian stimulation and several attempts at in vitro fertilisation without success. The partner, 39, suffered from non-obstructive azoospermia, with hypotrophic testicles but normal hormone levels.

During the experiment, the Star system analysed 2.5 million images in about two hours, identifying seven spermatozoa, two of which were mobile. These two were injected into as many mature oocytes, generating two embryos. Both were transferred into the woman's uterus and, thirteen days later, the pregnancy test was positive. An ultrasound scan at eight weeks confirmed the pregnancy, with a growing foetus and regular heartbeat.

'It only takes one healthy spermatozoon to create an embryo,' Williams points out. 'This result, although based on a single case, shows the feasibility and potential of Star to give a biological chance to couples who until now had none.

How Star Technology Works

The Star system integrates three key elements: a high-speed imaging system (up to 1.1 million images per hour); a customised microfluidic chip that allows the isolation of tiny areas of the sample; and a deep learning-based artificial intelligence model trained to recognise live spermatozoa in real time even in an apparently cell-free sample.

The model uses the 'Yolo' (You Only Look Once) architecture, capable of detecting moving objects in a single pass, with 90% accuracy. The entire process takes place in a closed and sterile system, minimising the risk of contamination.

A step forward in reproductive medicine

Male infertility accounts for about 40% of infertility cases in couples, and up to 15% of infertile men have azoospermia or extreme cases of cryptozoospermia. For many couples, the diagnosis means facing years of invasive interventions and emotional stress, often with the only alternative of donor sperm or adoption.

The Star platform offers a completely new perspective: an automated, non-invasive and potentially more effective system that can restore hope to those who thought they had lost it. Larger clinical studies are currently underway to test the reliability and reproducibility of the method. If the results are confirmed, it will open a new chapter in the fight against male infertility.

'This is just the beginning,' Williams concludes. 'The integration of artificial intelligence, robotics and human biology could change reproductive medicine forever.