In a pilot project, Sber's artificial intelligence helped detect almost twice as many cases of malignant neoplasms with the same number of examinations. The Can-SAVE system allowed detecting 2148 cases versus 1123 with the traditional approach. According to the company, the increase was 91%.
The model was developed by specialists from Sberbank's Center for Practical Artificial Intelligence and the Sber AI team. The algorithm studies information from electronic medical records — data on diagnoses and medical services — and assesses the risk of detecting an oncological disease in the next 12 months.
The system does not diagnose cancer instead of a doctor and does not issue a final diagnosis. It helps medical professionals decide which patients should be examined first.
In a study involving about 2 million patients from five regions of Russia, the developers tested how well the algorithm identifies people at increased risk. Among groups of 1000 people selected by Can-SAVE, tumors were detected 3.8–9.9 times more often than in groups formed according to standard medical protocols.
According to Sber, the model also outperformed 17 other machine learning and AI-based solutions. Can-SAVE, according to the developers, is capable of processing data for a city of a million people in less than three hours on a regular computer. The research results were presented at the international data analysis conference KDD 2026.
Read more on the topic:
Never miss our newsAdd this site to your preferred sources to see us more oftenAdd on Google
