Deep Vortex analyzed over 78,000 Facebook profiles and 660,000 fan pages from Serbia to study political tendencies. Using graph networks and clustering algorithms, they detected four dominant political communities, reflecting real-world party divisions. This approach provides fast, cost-efficient insights comparable to traditional political research.

Deep Vortex applied U-Net deep learning to segment high-resolution lung images, focusing on pulmonary acinus structures like alveoli and ducts. This enabled precise 3D reconstruction of the lung parenchyma, supporting simulations of airflow and targeted drug delivery. The approach improves patient-specific diagnosis and therapy while handling limited image datasets efficiently.

Using data from 215 NGOs across the EU and Western Balkans, Deep Vortex examined how social capital influences knowledge management maturity. Through predictive modeling and neural networks, they classified organizations by maturity levels, identifying gaps and opportunities for optimization. The study helps NGOs leverage social resources for more effective project planning and collaboration.

Deep Vortex developed a credit risk scoring solution for a major logistics company by combining financial datasets, OCR, and feature engineering. Machine learning models clustered clients by risk and optimized credit line allocation, supporting automated decision-making. The solution balanced internal and external stakeholder needs while improving efficiency and reducing financial exposure.

CityAI created a machine learning solution, HealthBeat, to predict cardiovascular risk factors and sick-leave outcomes using historical patient data. Multiple models assessed risk levels, sick-leave duration, cost, relapse, and cure causes. This approach enables proactive health management and data-driven interventions, providing scalable support for cardiovascular care.