Every year, about 90,000 women in England are sent to hospital after their GP notices signs that could point to womb cancer. The most common trigger is heavy bleeding after menopause. That referral starts a process that usually includes an internal scan, and sometimes a biopsy or a direct examination of the womb. It is uncomfortable. It takes time. And in roughly nine out of ten cases, the result is that there is no cancer.
That nine-out-of-ten figure is not a system failure. It is how cancer screening is supposed to work: cast a wide net, miss as little as possible. But it creates a serious practical problem. The NHS is running well behind on cancer waiting time targets. As of December 2025, fewer than three-quarters of urgent cancer referrals were resolved within the target of 28 days, against a goal of 75%. The 62-day target from referral to first treatment has not been consistently met since 2013. A system already short on capacity is doing a very large volume of tests on people who do not have cancer.
PinPoint Data Science, a company based in Leeds, built its blood test to address exactly this. The test analyses around 30 standard blood markers using a machine-learning system trained on data from hundreds of thousands of patients. It produces a single risk score: low, elevated, or high. The idea is not to replace the hospital investigation. It is to identify, before anyone goes to hospital, which women are at low enough risk that the investigation can be safely skipped.
The trial that led to this deployment involved over 16,000 patients referred through urgent cancer pathways across Yorkshire. The test correctly placed 99.1% of actual cancer cases in the elevated or high-risk group. For women it classified as lowest risk, the chance that cancer was missed was 0.2%. Those are strong numbers by clinical standards.
PinPoint says the test could spare about one in five referred women from needing an internal scan. That works out to roughly 18,000 women a year across England. For the NHS, that is 18,000 fewer appointments that need a specialist, a probe, and a follow-up pathway. For the women involved, it is a faster, less distressing answer.
Two NHS trusts are now preparing to use it: Mid Yorkshire NHS Teaching Trust for six types of cancer including gynaecological and upper digestive cancers, and Leeds Teaching Hospitals NHS Trust for gynaecological cancers. The test costs around £30, which is far below the cost of the specialist scans it would replace for the women cleared at the blood-test stage.
The broader NHS AI rollout in diagnostics is moving steadily. The government has committed £20 million to deploy AI-assisted chest X-ray tools to all NHS trusts in England by 2029; those tools have already been used in over 4 million patient assessments for lung cancer. A separate AI tool being added to the NHS App is expected to reach over 200,000 patients within 12 months. The PinPoint test fits into a clear direction: use AI early in the diagnostic process to sort patients by risk, so that specialist resources go to people who actually need them.
What is still missing is longer-term outcome data. Cancer Research UK called the test promising but noted that more research is needed to understand its real-world effect on patients and NHS capacity. That caution is fair. The trial results are strong, but a trial is a controlled environment. The question is whether the accuracy holds up across a wider and more varied patient population, and whether GPs and hospitals can integrate the score into their existing decision-making without adding new friction.
For anyone working in healthcare administration, insurance, or employee benefits, the pattern worth watching is this: AI is being used not to make a diagnosis, but to reduce the number of people who need to go through the full diagnostic process. That is a meaningful shift in how health systems can manage demand, and the PinPoint test is a concrete, affordable example of it working in practice.