Heterogeneous nature and complex rock properties of carbonate reservoirs makes the drilling process challenging. One of these challenges is uncontrolled mud loss. Caves or a system of cavities could be a high-risk zone for drilling as the mud losses cannot always be controlled by conventional methods, such as mud weight (MW) / equivalent mud weight (ECD) optimization, or by increasing concentration of lost circulation material (LCM) in the drilling mud. Seismic-based detection of such karstification objects is inefficient due to relatively small size, various shapes and low contrast environment. In this paper we, based on drilling data from the Barents sea, analyzed possible patterns in real-time drilling data corresponding to drilling through karstification objects. These patterns can serve as real-time indicators of zones with higher risk of karsts and can be used as an online tool for decision support while drilling in karstified carbonates.