Doctors Use AI to Spot 'Hidden' Sperm In Men

In brief

• AI is helping doctors locate sperm in men previously diagnosed with none.
• The AI found sperm in about 30% of tested cases, researchers said.
• Experts say larger clinical trials are still needed.

Artificial intelligence developed at Columbia University is helping doctors find sperm cells in men who were previously told they had none, opening a new path to biological parenthood for some couples, according to a report by the BBC.

The method, called Star, short for Sperm Track and Recovery, developed by the Columbia University Fertility Center, uses artificial intelligence to scan semen or tissue samples to locate extremely rare sperm cells that standard lab methods often miss. The approach could allow some men once considered infertile to use their own sperm in treatment.

Azoospermia, the condition the method targets, means no sperm can be detected in a man’s semen using conventional testing. It affects about 10% of infertile men and roughly 1% of all men overall, according to the BBC.

First announced in 2025, Star combines imaging, AI, and robotics to search for sperm. Samples move through microfluidic chips, small devices etched with channels as thin as a human hair that guide fluid in a controlled way. As the sample flows through, an imaging system captures about 300 images per second. A machine learning algorithm, a type of AI trained to recognize patterns, analyzes those images in real time to identify sperm among debris and other cells.

Researchers said a robot isolates sperm within milliseconds, avoiding centrifugation, a spinning method that can damage fragile cells. Doctors can then use the sperm in in vitro fertilization, or IVF, where an egg is fertilized outside the body.

The news comes as doctors and researchers are expanding the use of AI in medicine. In April, OpenAI said a version of ChatGPT designed for clinicians outperformed human physicians on certain clinical tasks, while Mayo Clinic researchers reported an AI model that can detect pancreatic cancer years earlier than doctors by identifying subtle changes in routine scans.