AI and Liquid Biopsy for Cancer

A liquid biopsy uses blood-based material to look for signals associated with disease. In multi-cancer early detection research, cell-free DNA fragments are sequenced and machine-learning classifiers can evaluate methylation patterns that differ across tissues. A test may report whether a cancer signal was detected and predict a likely signal origin. This is a screening result, not a tissue diagnosis. A positive result needs a diagnostic workup chosen by qualified clinicians; a negative result cannot rule out cancer or replace established screening for breast, cervical, colorectal or other diseases. Clinical utility is a separate question from detecting a molecular signal. The randomized NHS-Galleri trial enrolled more than 142,000 adults and evaluated a blood test added to usual care. In May 2026, the trial team reported that its primary goal—a reduction in combined stage III and IV cancer diagnoses—was not met, although analyses found fewer stage IV diagnoses in later screening rounds. A separate performance paper published in September reported secondary test-performance endpoints across three annual rounds; these descriptive analyses do not establish that the test reduces cancer deaths. It also reported sensitivity for all cancers below 40% in the evaluated rounds, so many cancers were not detected by the test. The U.S. FDA reviewed a premarket application for Galleri at a September 2026 advisory meeting. An advisory committee vote is non-binding and is not the same as FDA authorization; the FDA presentation stated that no devices were then authorized for multi-cancer early-detection screening. Patients and clinicians should weigh false reassurance, false positives, follow-up procedures and uncertain mortality benefit. The technology is promising research, but a blood signal should not replace an exam, symptom evaluation or guideline-recommended screening.