Abstract: The northeastern Yunnan - northwestern Guizhou region represents a significant lead-zinc polymetallic mineral province in China, where multiple large and super-large lead-zinc polymetallic deposits have been discovered, demonstrating considerable potential for deep mineral prospecting. However, deep mineral prospecting faces challenges, such as high costs, significant risks, and technical difficulties. Hence, by integrating the geological-geophysical-geochemical exploration data, this study conducted implicit 3D modeling of a deposit in the northwestern Guizhou region, followed by deep mineralization prediction of this deposit. The specific methodology is as follows. First, the 3D resistivity and polarizability models are generated through geophysical inversion and spatial interpolation of the audio-frequency magnetotelluric (AMT) and induced polarization (IP) sounding data. Second, based on borehole data and mineralization-related geological characteristics, a stratigraphic model is established using the K-means clustering algorithm in machine learning; a fault model is developed by integrating surface fault occurrences and AMT interpretations. Third, based on the resistivity, stratigraphic, and fault models, 3D mineralization prediction is conducted using the mineralization condition analysis method. Finally, the deep mineral prospecting targets can be delineated to guide the deep borehole arrangement. Drilling verification confirms the high reliability of the targets delineated using the above methodology. Overall, the implicit 3D modeling based on the geological-geophysical-geochemical exploration data can effectively predict unknown areas using known information. This approach provides sufficient samples for implicit 3D modeling, thereby assisting in constructing more accurate geological models under available conditions and providing a reference for mineral exploration in covered areas and in deep and marginal zones of old mines.