New AI-Powered Urine Test Detects Prostate Cancer with Higher Precision than PSA

Researchers from Karolinska Institutet developed a non-invasive urine test for early prostate cancer detection, utilizing AI, and pseudotime analysis. Validated in over 2,000 samples, the biomarkers showed an AUC of 0.92, outperforming PSA. This non-invasive method may enable earlier, more precise diagnosis and reduce unnecessary biopsies.

In the relentless pursuit of earlier, more accurate cancer diagnosis, prostate cancer has remained a stubborn frontier. Despite being one of the leading causes of cancer-related death among men worldwide, its detection still leans heavily on prostate-specific antigen (PSA) testing—an imperfect biomarker that has led to countless unnecessary biopsies, overdiagnoses, and patient anxiety.

Now, a new study published in Cancer Research presents compelling evidence that a simple urine test could transform prostate cancer diagnosis. By integrating artificial intelligence, with spatial transcriptomics and pseudotime analysis, researchers identified highly accurate urinary biomarkers that outperform PSA—offering a more precise, non-invasive method to detect prostate cancer, even in its early stages.

From Biopsy Rooms to Bathroom Samples

“There are many advantages to measuring biomarkers in urine,” says Dr. Mikael Benson, senior researcher at Karolinska Institutet and principal investigator of the study. “It’s non-invasive and painless and can potentially be done at home. The sample can then be analysed using routine methods in clinical labs.”

But behind the apparent simplicity of this test lies a deeply sophisticated scientific framework. To build it, the researchers analyzed the activity of every gene in thousands of individual cells from prostate tumor samples. By mapping gene expression to the physical location and grade of cancer in each cell, they constructed dynamic, three-dimensional "digital models" of how cancer evolves in tissue over time.

Using pseudotime analysis, they traced the molecular transitions from healthy to malignant states, identifying genes most tightly linked to disease progression.

From this complex data, machine learning algorithms were then trained to flag which genes could serve as reliable biomarkers. These candidates were subsequently validated in over 2,000 samples of prostate tissue, blood, and urine. The results were striking: the urine-based biomarkers not only indicated the presence of cancer with an AUC (area under the curve) of 0.92—a measure of predictive accuracy—but also correlated with the cancer’s severity.

Why PSA Falls Short—and This Test May Not

PSA screening has long been under scrutiny for its limited specificity. Elevated PSA levels can stem from benign conditions such as prostatitis or benign prostatic hyperplasia, leading to unnecessary follow-ups and biopsies. Moreover, PSA testing often misses aggressive cancers in their earliest stages, precisely when intervention would be most beneficial.

“New, more precise biomarkers than PSA can lead to earlier diagnosis and better prognoses for men with prostate cancer,” says Dr. Benson. “Moreover, it can reduce the number of unnecessary prostate biopsies in healthy men.”

The urine biomarkers uncovered in this study address several of PSA's shortcomings. They offer greater molecular specificity and provide information not just on the presence but also the grade of cancer—an essential factor in treatment decision-making.

The AI Advantage

What sets this research apart isn’t just the identification of biomarkers—it’s how they were found. Machine learning and spatial transcriptomics allowed the researchers to navigate the genomic and spatial complexity of prostate tumors with unprecedented resolution. By recognizing patterns invisible to the human eye, AI models could link biomarker expression directly to disease evolution, regardless of how heterogeneous or multifocal the tumor may be.

From Bench to Bedside

The findings are already being translated into next steps. The team is working with Professor Rakesh Heer at Imperial College London, who leads the UK’s national TRANSFORM study on prostate cancer. Plans are underway to test the biomarkers in prospective clinical trials at scale—a crucial step toward regulatory approval and eventual clinical integration.