Silicosis is an incurable lung disease caused by inhaling dust containing free crystalline silica. Silica is a chemical compound found in crystalline and amorphous (non-crystalline) forms.
The Occupational Exposure to Silica in Chile (Situación de Exposición Laboral a Sílice en Chile) report (Ministry of Health, 2015) estimated that 5.4 per cent of the workforce is at high risk of exposure to silica.
Given the specific characteristics and complex nature of the standards training course provided by the International Labour Organization (ILO), there is a very low percentage of specialists (currently 10 certified doctors) to meet the high daily demand for image reading, which averages some 332 images per day. Accordingly, there is a high risk of error, hindering early detection of this disease in workers.
In this context, the Mutual for Safety CChC (Mutual de Seguridad – CChC) identified a need to effectively differentiate between healthy patients and those showing signs of the pathology, in order to optimize diagnosis and medical care. To this end, an artificial intelligence (AI) system was developed to assist with medical decision-making. The model for detecting pneumoconiosis (silicosis and asbestosis) lets medical teams take evidence-based decisions, facilitating timely treatment of cases showing symptoms of the disease thanks to the model’s high sensitivity level.
The current model is able to detect healthy images with an accuracy of 99 per cent, thereby optimizing differentiation between healthy patients and patients with pneumoconiosis.