New AI Detects 75 Chest Pathologies with 99.8 Perecent Precision: Study Finds
The autonomous artificial intelligence (AI) system can achieve an extraordinary 99.8% precision while successfully identifying 75 distinct pathologies across a massive dataset, a recent study that appeared online on arXiv in March 2025 has shown.
Chest X-ray imaging remains an essential diagnostic tool in India for detecting thoracic diseases, yet previous research indicates that traditional manual interpretation methods can lead to up to 30% of abnormalities going undetected. Due to a critical clinical gap caused by a shortage of fewer than 15,000 radiologists for 1.4 billion people, Bargava Subramanian and colleagues at 5C Network conducted the study to develop a scalable AI-driven approach for enhancing multi-pathology detection capabilities within the Indian healthcare system.
Therefore, the multi-site clinical trial utilized a colossal dataset of 5,003,742 chest radiographs gathered from diverse settings, including 17 major Indian healthcare systems, to validate a computer-aided detection (CAD) system utilizing Vision Transformers (ViT), Faster Regional Convolutional Neural Networks (Faster R-CNN), and U-Net architectures. The study focused on the comprehensive identification and segmentation of 75 thoracic pathologies across all age groups while excluding non-chest images through automated verification to achieve the primary endpoints of high-precision classification and accurate pathology localization.
Key Findings of the Study Include:
• Superior Triage Performance: The autonomous system achieved a remarkable 99.8% precision and 99.6% recall for binary normal versus abnormal classification, ensuring clinicians can definitively prioritize critical patient needs.
• Extensive Pathology Coverage: The study demonstrated that the technology accurately identifies and localizes 75 distinct pathologies, maintaining precision rates up to 97% and recall exceeding 95% for conditions like pneumonia and tuberculosis.
• Enhanced Workflow Efficiency: Integrating this AI solution reduced radiology reporting times by as much as 50% compared to traditional methods, facilitating significantly faster clinical decision-making.
• Consistent Demographic Reliability: Subgroup analysis confirmed that the model maintains stable diagnostic accuracy across all age groups and imaging equipment types, including both computed radiography (CR) and digital radiography (DR) machines.
• Proven Real-World Utility: During live clinical application, the technology successfully processed over 150,000 radiographs with an average daily volume of 2,000 scans, validating its scalability across 17 major healthcare systems.
The results suggest that the scalable AI model provides a highly reliable tool for autonomous CXR classification, achieving up to 98% precision and over 95% recall for multi-pathology detection. These findings confirm the system can effectively address diagnostic gaps in underserved regions while optimizing healthcare workflows through standardized, high-quality reporting.
The study concludes that implementing such AI-driven diagnostic solutions could assist clinicians in managing heavy patient volumes and improving the consistency of thoracic imaging evaluations across various clinical settings.
Reference
Subramanian, B., Jaikumar, S., Shastry, P., Kumarasami, N., Sivasailam, K., Keerthana R., A., Mounigasri M., & Venkatesh, K. P. (2024). Autonomous AI for Multi-Pathology Detection in Chest X-Rays: A Multi-Site Study in Indian Healthcare System.
