AI tool for screening Silicosis

Fighting The Occupational Hazard Silicosis In The Age of AI

-Ameya Srivastava

01 July, 2024

Introduction

Every breath sustains life, yet for millions of miners worldwide, each inhalation introduces a silent assailant: silica dust, the causative agent of the deadly lung disease known as silicosis. This chronic occupational hazard poses a significant threat to the respiratory health of workers across various industries, including masonry, construction, and manufacturing. In India alone, an estimated 11.5 million workers are exposed to toxic silica particles, a figure projected to surge to 52 million by 2025-26. Silicosis has a long history, with references dating back to 400 BC by Hippocrates and notable industrial tragedies such as the Hawk’s Nest Tunnel disaster in 1930, underscoring its enduring impact on public health (Baum and Arnold 2023).

Epidemiology

Asia, particularly China and India, emerges as the primary hotspot for silicosis, with prevalence rates ranging from 3.5% among ordnance factory workers in Delhi to 79% among stone workers in Rajasthan (Yi et al. 2023; Rupani 2023). The informal sector, comprising 92% of India's working population, bears the brunt of silica exposure, further exacerbating the disease burden (Rupani 2023). The co-occurrence of silicosis and tuberculosis, termed silico-tuberculosis, compounds the challenge, particularly in states like Rajasthan. The pervasiveness of silicosis underscores the urgent need for effective prevention and intervention strategies.

Health Effects and Challenges in Diagnosis

Silicosis-affected individuals may experience stiffness in the lungs, difficulty in breathing, chest pain, night sweats, inflammation, and severe coughing. The crystalline form of silica (RCS) pierces the lung tissue like shards of glass. It is engulfed in the alveolar macrophages causing swelling of lymph nodes, ultimately leading to pulmonary fibrosis or lung cancer. A few weeks of high exposure causes acute silicosis while 10 to 30 years of low exposure results in chronic silicosis. Workers generally believe they have recurrent TB and do not seek treatment, and return to dusty settings exacerbating silica exposure.

Silica dust (SiO2), despite being odorless and non-irritant, poses significant health risks, often leading to a delayed diagnosis due to inadequate awareness of early symptoms. The remote locations of mining sites exacerbate diagnostic challenges, with nodules in chest X-rays often misinterpreted as tuberculosis, delaying appropriate intervention. The long latency period of the disease further complicates early detection efforts, allowing preventable conditions to escalate into life-threatening illnesses. The adverse health effects of silicosis, including pulmonary fibrosis and lung cancer, highlight the urgent need for improved diagnostic techniques and preventive measures.

Emergence of AI-Based Screening Tools

The emergence of AI-based disease detection models presents a promising solution to the challenges posed by silicosis diagnosis. Deep learning algorithms have emerged that can analyze chest X-rays with unparalleled accuracy for lung pathologies. By automating the screening process, these models alleviate the burden on radiologists and facilitate timely intervention, ultimately improving patient outcomes. One such field to benefit immensely from computer-aided detection is Tuberculosis. There are now many commercially available systems that have achieved promising metrics of agreement (area under the curve of receiver operator characteristic (AUROC), sensitivity, and specificity). There is a large resource of CAD validation studies in TB detection, such that it is now conditionally recommended by the WHO for the detection of active TB in the adult population.

Artificial Intelligence Related Challenges

AI-aided detection systems have also been designed for pneumoconiosis, including silicosis, but have not been validated on as large datasets as TB. Silicosis manifests as a fibroid disorder with a nodular appearance on CXR, mimicking symptoms of TB and conversely, TB can present features as bilateral nodulations and massive fibrosis mimicking silicosis. Few CAD systems claim to distinguish between silicosis and tuberculosis but are unable to perform well in the case of silico-tuberculosis (Ehrlich et al. 2022). Most of these models are proprietary knowledge hence their mechanisms of detection are not disclosed but Zhang et al show that a convolutional neural classification network model performs as well as human readers if the CXRs are segmented into multiple subregions for analysis (Zhang et al. 2021). Limited and heterogeneous datasets, coupled with the need for continuous model refinement, pose significant hurdles to the development of accurate screening tools. Distinguishing between nodules indicative of silicosis, tuberculosis, and silico-tuberculosis remains a formidable task, requiring innovative approaches and rigorous validation studies. Furthermore, ethical considerations, including patient privacy and automation bias, necessitate scrutiny to ensure responsible AI deployment in public health settings (Spiegel et al. 2021).

Global Efforts to Curb Silicosis

Efforts to combat silicosis span across various countries, with initiatives ranging from AI-based screening programs to biomarker research and policy interventions. 

In South Africa, the Medical Bureau of Occupational Diseases has implemented AI-based screening procedures, streamlining the diagnosis process for gold miners (Spiegel et al. 2021). Similarly, countries like Peru and Chile are leveraging pulse oximetry for early screening, highlighting the diverse approaches to addressing this global health challenge (Donroe et al. 2008). Several biomarkers like serum Heme Oxygenase-1 (HO-1), Club/Clara cell protein, 16 (CC16),  and Nephronectin (Npnt) are used for assessing early signs of silicosis, but with limited success. ICMR-NIOH, Ahmedabad has reportedly devised an early detection test for silicosis centered around Clara Cell Protein 16 (CC-16). However, the test has not reached the stage of clinical trials yet (Sishodiya 2022). Rajasthan was one of the first states in India to introduce a policy in 2019 to curb the silicosis disease burden and implement a large-scale machine-learning model. AI-embedded systems are the need of the hour to handle the disease burden effectively.

Measures to be Taken - What the Future Holds for Indian Miners

In India, addressing the silicosis burden requires a multifaceted approach encompassing standardized exposure limits, routine surveillance systems, and robust preventive measures. The deployment of AI-powered surveillance tools holds promise for early detection and improved monitoring of high-risk areas. Empowering site supervisors with incentives for implementing preventive protocols and ensuring timely diagnosis and compensation for affected workers are critical steps toward mitigating the impact of silicosis. As artificial intelligence continues to advance, it offers hope for a future where silicosis is detected earlier, diagnosed more precisely, and managed effectively, ultimately safeguarding the health and well-being of miners worldwide.

References:

  1. Baum, Lauren, and Thomas C. Arnold. 2023. “Silicosis.” In StatPearls. Treasure Island (FL): StatPearls Publishing.
  2. Donroe, Joseph A., Paola J. Maurtua-Neumann, Robert H. Gilman, Ana Teresa Acosta, Gene Caine, John E. Parker, Jaime Carlos Alvarez Carhuaricra, et al. 2008. “Surveillance for Early Silicosis in High Altitude Miners Using Pulse Oximetry (International Journal of Occupational and Environmental Health 14, 3).” International Journal of Occupational and Environmental Health 14 (4). https://pure.johnshopkins.edu/en/publications/surveillance-for-early-silicosis-in-high-altitude-miners-using-pu-6.
  3. Ehrlich, Rodney, Stephen Barker, Jim Te Water Naude, David Rees, Barry Kistnasamy, Julian Naidoo, and Annalee Yassi. 2022. “Accuracy of Computer-Aided Detection of Occupational Lung Disease: Silicosis and Pulmonary Tuberculosis in Ex-Miners from the South African Gold Mines.” International Journal of Environmental Research and Public Health 19 (19). https://doi.org/10.3390/ijerph191912402.
  4. Rupani, Mihir P. 2023. “Challenges and Opportunities for Silicosis Prevention and Control: Need for a National Health Program on Silicosis in India.” Journal of Occupational Medicine and Toxicology  18 (1): 11.
  5. Sishodiya, Prahlad K. 2022. “Silicosis–An Ancient Disease: Providing Succour to Silicosis Victims, Lessons from Rajasthan Model.” Indian Journal of Occupational and Environmental Medicine 26 (2): 57.
  6. Spiegel, Jerry M., Rodney Ehrlich, Annalee Yassi, Francisco Riera, James Wilkinson, Karen Lockhart, Stephen Barker, and Barry Kistnasamy. 2021. “Using Artificial Intelligence for High-Volume Identification of Silicosis and Tuberculosis: A Bio-Ethics Approach.” Annals of Global Health 87 (1): 58.
  7. Yi, Xinglin, Yi He, Yu Zhang, Qiuyue Luo, Caixia Deng, Guihua Tang, Jiongye Zhang, Xiangdong Zhou, and Hu Luo. 2023. “Current Status, Trends, and Predictions in the Burden of Silicosis in 204 Countries and Territories from 1990 to 2019.” Frontiers in Public Health 11 (July): 1216924.
  8. Zhang, Liuzhuo, Ruichen Rong, Qiwei Li, Donghan M. Yang, Bo Yao, Danni Luo, Xiong Zhang, et al. 2021. “A Deep Learning-Based Model for Screening and Staging Pneumoconiosis.” Scientific Reports 11 (1): 2201.
Share this


Related Posts


Cookie Management