The Pinpoint Test

Problem

The NHS investigates around three million people for cancer every year but only 6% of them actually have it. The result is a system stretched to breaking point: unnecessary referrals, anxious patients enduring invasive procedures they don’t need, and clinicians with no reliable way to separate high-risk cases from low-risk ones at the point of triage.

Urgent cancer referrals are rising 10% year-on-year. Diagnostic capacity is not keeping pace. Patients are seen in first-come, first-served order rather than by clinical urgency, creating a paradox in which the people who need to be seen fastest are not always the people being seen first.

In gynaecology, the pressure is particularly acute. Around 90,000 women are referred annually with post-menopausal bleeding, yet only around 10% will have cancer. Most go on to undergo uncomfortable trans-vaginal procedures which are painful, invasive, and in the majority of cases, ultimately unnecessary. There is currently no simple, affordable way to rule out low-risk patients in primary care before they enter the hospital pathway.

The NHS needs a smarter front door to cancer diagnostics: one that prioritises by biological urgency, protects patients from unnecessary harm, and makes better use of scarce capacity.

Solution

The Pinpoint Test is an AI-powered blood test that gives clinicians a single, reliable cancer risk score at the point of triage generated from a standard blood draw already taken in local pathology labs.

Using machine learning, the platform analyses 30 routinely collected blood biomarkers alongside the patient’s age and sex, aggregating complex signal patterns invisible to the human eye into one clear risk indicator: the actual probability that a symptomatic patient has cancer. The result integrates directly into local LIM systems and requires no new equipment, no capital expenditure, and no change to existing clinical workflows.

The test performs across a broad range of cancer types including head and neck, gastrointestinal, lung, and gynaecological, and is pan-cancer sensitive, meaning it can detect cancers presenting on the wrong pathway.

Impact

• Evaluated across 16,481 urgent suspected cancer referrals in one of the largest real-world assessments of AI-enabled cancer triage in the NHS.
• Correctly identified 99.1% of gynaecological cancers, with a negative predictive value of 99.8% for the lowest-risk group outperforming standard trans-vaginal ultrasound on diagnostic accuracy (AUC 0.832 vs 0.792).
• Could allow one in five symptomatic women to be safely ruled out in primary care, reducing pressure on hospital services.
• In a sub-analysis of 578 women, 116 could have avoided transvaginal ultrasound and hysteroscopy entirely.
• Modelled cost savings of £9.6m per year across 6 pathways in West Yorkshire; ranging from £2m (ICB with fewest USC referrals) to £12m (ICB with most).
• 4 of 9 pathways completed as part of a service evaluation in the West Yorkshire Association of Acute Trusts; Mid Yorks Teaching NHS Trust is progressing a wider pilot.
• Directly aligned to the National Cancer Plan’s call for AI-driven triage tools and the renewed Women’s Health Strategy (April 2026) commitment to offering women more choice ahead of hysteroscopy.