Two medical trusts of the UK's National Health Service (NHS) are preparing to implement an innovative AI blood test called PinPoint, designed to screen women suspected of having uterine cancer. Clinical trial results are impressive: in a group of 3,313 participants, the machine learning model demonstrated 99% accuracy in both detecting and ruling out gynecological cancers.
Each year in England, around 90,000 patients are referred for examination due to heavy postmenopausal bleeding. Approximately 10,000 of them receive a diagnosis of uterine (or endometrial) cancer. The disease kills 2,700 women annually. The standard diagnostic procedure includes transvaginal ultrasound to measure endometrial thickness, and if suspicions persist, biopsy and hysteroscopy. Many patients find these procedures painful and invasive.
Notably, 20% of those referred do not have cancer confirmed. According to developers, the test could rule out the disease at the initial referral stage, sparing around 18,000 women per year from unnecessary invasive procedures.
How the test works
The PinPoint algorithm analyzes 30 biomarkers in a blood sample. Based on their combination, the model determines the risk level: low, elevated, or high. The technology was developed by PinPoint Data Science, a company based in Leeds specializing in statistical analysis of medical data. The test began implementation after a large-scale trial involving 16,481 patients referred by general practitioners from 170 clinics in Yorkshire.
"99% accuracy is outstanding by any clinical standards. But equally important is another factor: the test reliably rules out cancer in women with low risk," said Sean Duffy, Chief Medical Officer of PinPoint Data Science.
Mid Yorkshire NHS Teaching Trust plans to use the test for six types of gynecological cancer and upper gastrointestinal tract cancer. Leeds Teaching Hospitals NHS Trust will implement it for screening gynecological cancers.
Doctors' response
General practitioner Jacinta Walsh from a practice in West Yorkshire noted that the current process of ruling out cancer requires up to six doctor visits. According to her, PinPoint will speed up diagnosis and free up time for other patients. Consultant gynecologist Tracy Jackson from Leeds Trust emphasized that most referred women turn out to be healthy, and the new technology will enable more effective triage.
"Women who actually have cancer will see a doctor faster, receive a diagnosis, and start treatment. That's where our focus should be," she stressed.
Cancer Research UK called the PinPoint test results "promising," but the organization's representative Samantha Harrison pointed out the need for additional research to assess the long-term benefits for patients and the healthcare system.
My analysis: The introduction of AI tests in oncology is not just technological progress but a paradigm shift in early diagnosis. 99% accuracy in ruling out uterine cancer could dramatically reduce the burden on the healthcare system and improve patients' quality of life by sparing them from unnecessary procedures. However, as experts rightly note, larger and longer-term studies will be needed to confirm the effectiveness and safety of PinPoint in routine clinical practice. Nevertheless, this is a striking example of how machine learning is transforming medicine.