Canadian uranium explorer Labrador Uranium says the training of an artificial intelligence algorithm on its central Labrador project geological data has shown promise.
“Significantly, the validation process confirmed that the algorithm had a high success rate in predicting known mineralisation, providing confidence that other priority target areas may yet host unknown mineralisation,” the company said this week.
Management said completion of an initial phase of regional exploration targeting at its Central Mineral Belt (CMB) project, integrating a “mineral systems approach” with machine learning, defined specific areas for further work. They included areas in the vicinity of the Moran Lake trend, where recent drilling reportedly intersected mineralisation beyond the boundaries of a historical resource estimate.
Interim Labrador Uranium CEO Philip Williams said uranium deposits found at Moran Lake and Anna Lake were among several discoveries in the large CMB district based on vast amounts of exploration date generated since the 1950s.
“With most of these deposits found through simple surface prospecting, we believe the potential to find additional mineral deposits under cover remains strong,” he said.
“Our regional exploration, machine learning program was designed to compile and process the vast data over our large, 150,000-plus-hectare land position, and provide direction for future exploration programs.
“As part of the validation process it was confirmed that the algorithm had a high success rate in predicting known mineralization. This gives us confidence in prioritising new target areas by extending the predictions to areas of potentially yet to be discovered mineralisation.
“We look forward to ground-truthing several of these areas during the upcoming 2023 field season.”
Labrador Uranium’s data-driven methodology was primarily aimed at reducing targeting risk at the CMB project at an early stage, preparing more target areas for direct detection methods such as drilling, the company said.
“Currently the ML model is being interrogated and refined using shapely additive explanations [SHAP] to identify and explain features that best predict deposit locations,” it said.
“This analysis is expected to inform conceptual geological interpretations and guide future data acquisition for improved ML results in future iterations of the ML process and its use on more refined target areas.”