Researchers at the Johns Hopkins Kimmel Cancer Center published findings on July 31, 2026 showing that an AI-powered liquid biopsy blood test successfully detected liver cancer in patients from Guatemala and Romania, two geographically, genetically and behaviorally distinct populations, according to reporting by Medical Xpress and News-Medical. The study analyzed blood samples from 377 people, a validation step researchers say is critical because a screening test that only works in one population is of limited use for a disease that affects patients under very different risk conditions worldwide.
How the DELFI test works
The test relies on Hopkins’ DELFI platform — DNA Evaluation of Fragments for Early Interception — which uses AI to analyze patterns in how cell-free DNA fragments circulate in the bloodstream. Tumors shed DNA into the blood in fragmentation patterns that differ from those of healthy cells, and DELFI’s AI model has been trained to recognize those genome-wide fragmentation signatures rather than searching for a single specific mutation, an approach the Hopkins team argues generalizes better across different underlying causes of cancer.
Why researchers chose these two specific populations
Participants in Romania largely developed liver disease from viral hepatitis or alcohol use, the classic risk pathway for hepatocellular carcinoma in much of Europe and North America. Participants in Guatemala primarily had metabolic liver disease tied to obesity and diabetes, with many also exposed to aflatoxin, a naturally occurring toxin produced by mold on stored grains and a well-documented liver cancer risk factor in parts of Central America and sub-Saharan Africa. According to the Hopkins team, the AI model detected liver cancer consistently across both groups despite these very different biological pathways to disease, which the researchers say strengthens the case that the underlying fragmentation signal reflects cancer biology broadly rather than one population’s specific risk profile.
Why liver cancer screening has lagged behind other cancers
Hepatocellular carcinoma is typically diagnosed late, in part because current screening relies on ultrasound and a blood protein marker called AFP that misses a substantial share of early tumors, particularly in patients whose liver disease stems from metabolic causes rather than viral hepatitis — a growing share of cases globally as obesity-driven fatty liver disease rises. A blood test that works across both viral and metabolic disease pathways could close a screening gap that has left many at-risk patients, especially in low-resource settings without easy ultrasound access, undiagnosed until the cancer is advanced.
What the researchers say comes next, and what critics will want to see
The Hopkins team frames this study as building on earlier 2026 work showing that the same genome-wide fragmentomic approach could detect liver fibrosis and cirrhosis, the precursor conditions that often precede liver cancer, suggesting the platform might eventually support a screening pipeline that catches disease even before a tumor forms. Independent oncologists not involved in the study are likely to note that 377 total participants, while a meaningful cross-population validation, remains a modest sample size for a screening test intended for population-wide deployment, and that prospective trials tracking outcomes in newly screened, previously undiagnosed patients — rather than retrospective analysis of known cancer and non-cancer blood samples — will be needed before regulators or clinical guidelines endorse routine use.
Where this fits in the broader AI liquid biopsy race
DELFI is one of several liquid biopsy platforms competing to bring AI-driven, blood-based cancer detection into mainstream screening, alongside efforts targeting colorectal, lung and multi-cancer detection more broadly. Liver cancer has drawn comparatively less commercial liquid biopsy investment than more common cancers, making Hopkins’ cross-population validation notable less for technical novelty and more for demonstrating that an academic-led liquid biopsy program can generate global validation data without the marketing apparatus of a well-funded startup.
What’s next
The Hopkins team has not announced a timeline for prospective clinical trials or FDA engagement for the liver-cancer-specific application of DELFI, though the platform’s underlying technology has moved through earlier regulatory and commercial pathways for other cancer types. Given the global aflatoxin and viral hepatitis risk factors driving liver cancer incidence outside the United States, researchers are likely to pursue additional validation in other high-incidence regions, including parts of Sub-Saharan Africa and East Asia, before any test reaches routine clinical screening programs.
