Several NHS hospitals are preparing to use a machine-learning blood test to assess women referred with possible womb cancer before invasive investigations. PinPoint Data Science’s test analyses around 30 markers and classifies risk as low, elevated or high.
Its most valuable role may be ruling out very-low-risk patients. Around 90,000 postmenopausal women in England are referred annually after heavy bleeding, while most do not have cancer and may undergo uncomfortable examinations.
A low-cost blood test could improve triage
The PinPoint test costs about £30 and produces a risk score for existing urgent referral pathways. Clinicians can use it when deciding who may be monitored, who needs further assessment and who should be prioritised.
A trial included 16,481 patients across Yorkshire cancer pathways. For women in the lowest-risk group, the reported negative predictive value was 99.8%, while 99.1% of cancers were classified elevated or high risk.
Those metrics support triage but need careful interpretation. Predictive value depends on prevalence and population, so performance may change across regions, ethnic groups, ages and clinical settings.
Mid Yorkshire plans use across several gynaecological and upper-gastrointestinal pathways, while Leeds will apply it in gynaecological referrals. Real-world deployment should collect outcomes beyond the original trial.
The goal is fewer procedures without delayed diagnosis
Current assessment often includes transvaginal ultrasound, followed when needed by biopsy or hysteroscopy. Some women find these procedures painful or distressing.
PinPoint estimates that one in five referrals could avoid ultrasound—around 18,000 women annually. Low-risk patients might be ruled out in primary care, freeing hospital capacity and shortening repeated GP visits.
Safety-netting is essential. A low score should not close the pathway when symptoms persist, worsen or conflict with clinical judgement. Patients need clear instructions about when to return.
Higher-risk results can prioritise specialist capacity, potentially reducing delay for women most likely to have cancer. The test should support, not replace, a clinician’s assessment and informed discussion.
False reassurance and unnecessary alarm must both be monitored. Communication should explain that a risk category is not a diagnosis.
NHS adoption needs independent evidence and equitable access
Cancer Research UK called the test promising while emphasising the need for more research on patient outcomes and NHS capacity. Implementation should compare diagnosis time, missed cancers, procedures avoided and patient experience.
Calibration across populations is crucial. If markers perform differently for a group underrepresented in development data, a system intended to reduce burden may widen diagnostic inequality.
Patients should help shape the pathway. Some may prefer imaging despite a low score for reassurance, while others may value avoiding an invasive procedure. Shared decision-making and clear language about residual risk are part of clinical quality, not optional communication around the algorithm.
Data governance should cover laboratory results, model version and how scores enter records. Clinicians need to know when the model changes and whether thresholds were revalidated.
The NHS is deploying AI in infection prediction, triage and imaging as well, making shared evaluation and monitoring capacity increasingly important.
Operational savings should be reinvested in timely diagnostics rather than treated only as a budget reduction. The purpose is a kinder and faster pathway.
PinPoint’s test offers a compelling use of AI because it addresses a specific decision with measurable consequences. If independent real-world evidence confirms the trial results, thousands of women could avoid an invasive procedure while those at higher risk move faster. The technology earns trust not by sounding intelligent, but by helping clinicians rule out cancer safely and preserving a clear route back when symptoms say otherwise.
The NHS should publish those consequences transparently as the rollout expands beyond the original hospitals.
