A major breakthrough has been made that a simple blood test analyzed by artificial intelligence could enable early detection of primary vitreoretinal lymphoma (PVRL), a rare and aggressive cancer often considered misleading in terms of inflammation.
A learning model designed to analyze patterns was trained by researchers using routine complete blood count data from 255 PVRL patients and 292 controls. This technique is a non-invasive diagnostic technique that aims to replace the current dominant invasive position.
The six-trait random forest model achieved an area under the curve (AUC) of 0.85 in the discovery cohort, with continuous verification between groups at an AUC of 0.80–0.83.
This result surpasses conventional biomarkers, the interleukin-10/interleukin-6 ratio, which scored only 0.65-0.78.
PVRL usually presents as uveitis with blurred vision and blurred vision, and diagnosis is often delayed for months due to unspecified symptoms.
The model specifically identified 38 PVRL cases among 66 high-risk individuals and another 2 among 83,610 low-risk patients.
The data showed a sensitivity of 95.0%, a specificity of 99.97, a positive predictive value of 57.6% and a negative predictive value of 99.9%.
The recent study underlines the fatal outcome of PVRL when it spreads to the brain, often bilaterally and in older adults with less pain.
Nevertheless, a free web application allows therapists to enter blood results based on direct risk scores so they can seek immediate care for vision loss.
Furthermore, this blood-based strategy could help save eyesight and lives, bypassing expensive imaging or lumbar punctures, while experts called it an important discovery for this misdiagnosed malignancy.
