AI Tool Accelerates Lung Screening for 2.2 Million US Dust-Exposed Workers

Michigan State University researchers have developed an artificial intelligence system designed to screen over 2.2 million United States workers routinely exposed to hazardous silica, coal, and rock dust. The tool addresses a critical bottleneck in occupational health: a severe shortage of certified B readers, the physicians specially trained to interpret chest X-rays for early signs of silicosis and black lung disease. With only approximately two hundred certified specialists nationwide, routine screenings for coal miners and other high-risk professionals face significant delays, allowing preventable lung scarring to progress undetected. The AI model was trained on a proprietary dataset of US worker chest radiographs, making it the first screening system optimized specifically for American occupational imaging standards. Published in Occupational and Environmental Medicine, the research demonstrates how machine learning can function as a diagnostic copilot. The program rapidly analyzes radiographic images and successfully clears roughly fifty percent of scans that show no abnormalities, immediately redirecting specialized physician attention to cases exhibiting subtle early-stage pathology. In clinical validation, the system achieved a ninety-one percent accuracy rate in identifying initial lung scarring, substantially outperforming the seventy-seven percent average of human readers. To assist clinicians, the algorithm generates color-mapped overlays that highlight suspected tissue damage, providing an objective second opinion that reduces inter-rater variability among medical specialists. Lead investigators including occupational medicine chief Kenneth Rosenman, computational mathematics professor Adam Alessio, and associate professor Ling Wang emphasized that early detection is paramount. Because dust-induced lung fibrosis develops over decades and remains irreversible, timely identification allows employers to reassign affected personnel, upgrade facility ventilation, and initiate protective medical management before symptoms become severe. The research team, which includes doctoral candidates Meiqi Liu, Zenas Huang, Ian Loveless, and undergraduate Michal Borek, is currently collaborating with the National Institute for Occupational Safety and Health to transition the algorithm into a deployable clinical application. By integrating high-speed computational screening into existing occupational health workflows, the initiative aims to eliminate diagnostic backlogs, standardize early disease detection, and mitigate long-term respiratory morbidity across America’s most dust-exposed industrial workforce.