Published 20 Sep 2023

AI's Rising Importance in Combating Wildfire-Related Lung Health Issues

Author: Katrina Matti, Daniel Lehewych



Wildfires and Lung Health Complications
The destruction wrought by wildfires is increasingly evident to the modern eye – one that observes the fire, its distant smoke, and the harmful microscopic particulate matter that is harmful to lungs when inhaled.  
According to research from Current Pulmonology Reports, smoke from wildfires contains a harmful mixture of fine particles (PM2.5) and hazardous gases that invade the respiratory and circulatory systems. Exposure to such particulate matter can result in a range of consequences. These consequences span from short-term irritation and wheezing to long-term declines in lung function. Diseases like asthma, bronchitis, and pneumonia become aggravated, and connections to heart disease and mental health add layers to this complex problem. 
Wildfires, a natural occurrence often intensified by human activities and climate change, are a growing concern in North America, prompting government officials and health leaders to deliberate and propose ideas for adapting to and combatting such weather conditions.  
In June 2023, North American cities were blanketed by thick smoke from Canadian wildfires. As reported in The Lancet, officials were quick to recognize and communicate the health risks. New York City's Health Commissioner, Ashwin Vasan, detailed how particulate matter from the smoke can affect those with existing health conditions, emphasizing the need to stay indoors and to wear masks if outdoor activity couldn’t be avoided.  
Last August, the U.S. Environmental Protection Agency (EPA) published guidelines urging health workers to anticipate smoke events and tailor disease management plans accordingly.  Measures like stocking up on government-approved N95 respirators and creating personal contingency plans indicate our growing understanding of the issue.  
The Lancet states that since 2001, exposure to wildfire smoke has increased, with projections showing a growing trend. But the response from governments, researchers, and community leaders demonstrates a determination to address the problem head-on. Kristie Ebi of the University of Washington insists on an "all-of-society approach," promoting better management across various sectors and social groups –one that is open to the adoption of new tools and strategies for progress.
AI as an aid in fighting wildfire-induced health risks  
An expanding body of research demonstrates AI’s ability to competently aid doctors in detecting lung malignancies early – a feature which could be a game-changer in the fight against wildfire-induced lung health risks.  
AI’s potential in medicine has seen a rise, offering early detection and interventions in lung illnesses. For example, the deep learning algorithm, qCT-Lung from, demonstrates an example of AI’s capabilities in identifying pulmonary nodules, among other critical health markers.  
According to research from the journal Chest, during the COVID-19 crisis in India, qCT-Lung was used on chest CT scans, successfully detecting nodules and even identifying lung cancer cases. The synergy of human expertise and AI-driven insights marks a significant advancement in healthcare. 
Practitioners do not confine the integration of AI in healthcare to cancer detection. There is a growing interest in tailoring AI to address wildfire-induced lung diseases, specifically involving assisting healthcare professionals in early diagnosis and providing personalized care. Using real-time data from wildfire occurrences can further enhance the management of health impacts. 
Research from PLOS ONE probed the capacity of a deep learning algorithm in analyzing routine frontal chest radiographs (CXR) from adult patients. Researchers applied the algorithm to 874 frontal CXRs from 724 individuals, targeting pulmonary opacities, pleural effusions, and other chest irregularities. 
Around 42% of the analyzed CXRs exhibited no abnormalities, with the rest divided between single and multiple irregularities. Notably, the analysis by the deep learning model was on par with human evaluations, showing no substantial statistical difference in detecting all abnormalities, thus accentuating the technology’s adeptness in mirroring human precision. 
The research also yielded insights into the effectiveness of DL, as measured by the area under the curve (AUC). Deep learning’s effectiveness ranged between 83% and 93%, closely paralleling the 69% to 92% range for human radiologists. Even more fascinating was that the lowest point noted in the study did not diminish the overall robustness of deep learning’s capabilities but highlighted specific areas for further refinement and research. 
Moreover, the finding that the presence of chest wall implanted devices could influence the accuracy of the DL algorithm offers an opportunity for enhancing the technology, tailoring it to accommodate even these complex scenarios. 
Integrating AI into medical practice is no longer confined to theory, particularly in the critical domain of lung cancer screening. Further research from Chest sheds light on a promising collaboration between traditional radiological techniques and AI, offering an exciting pathway toward enhanced diagnostic accuracy. 
This research sought to compare how radiologists use the standardized Lung-RADS classification to categorize lung nodules with how an AI algorithm calculates the malignancy risk for those same nodules. Chest researchers studied 210 adult patients, employing human expertise and machine learning to distinguish between malignant and benign nodules. 
The AI algorithm ventured beyond mere imitation, assigning malignancy risk scores based on a diverse range of factors, such as the size and shape of the nodules, family history of lung cancer, and the patient's smoking history. The results were telling; AI-generated risk scores were notably higher for malignant nodules, clearly differentiating them from benign ones. 
This comparison between traditional Lung-RADS classification and AI scores was not merely an academic exercise. The AI's ability to predict malignancy was comparable to that of the Lung-RADS system.  These close results highlight the potential synergy between human intuition and algorithmic precision. The implications of this study are profound, suggesting that AI doesn't merely mimic human expertise but augments it. AI-generated malignancy risk scores demonstrate significant sensitivity in identifying cancerous nodules through low-dose CT scans for lung cancer screening.  
This collaborative approach may signal a new era in early cancer detection, where technological innovation enhances human insight. In this merging of minds and machines, both are essential contributors to a potentially life-saving process. 
Chest X-rays, renowned for their ability to delineate the anatomical structures of the chest, can identify abnormalities, but they might not be the ideal tool for detecting small particulate matter, especially if it's dispersed throughout the lungs. This technology primarily visualizes structures by differences in density, such as bones appearing white and air appearing black. Particulate matter may not present a significant density difference to be distinguished easily, especially if it's microscopic. 
However, alternative imaging modalities might offer greater insight. A CT scan, for example, provides more detailed images and can detect smaller changes in tissue density. In cases of pneumoconiosis—a disease frequently observed in those who have inhaled wildfire particulates—the inhaled particles can lead to fibrotic changes in the lung tissue, which can be visualized on a CT scan. 
In these instances, the inhaled particles themselves might not be directly visible. However, the secondary effects they cause become the markers for their presence. And these include fibrosis, nodules, or other structural alterations. A skilled radiologist could infer the presence of particulate matter based on these characteristics with the augmentation of AI software such as qCT-Lung, enabling them to see more patients and prioritize critical cases.  
Wildfires represent more than mere destruction; they highlight a nuanced health crisis that extends far beyond the visible landscape. The promising advances of AI provide a compelling answer, detecting, diagnosing, and potentially mitigating the respiratory impacts of wildfires. 
The combination of human insight and technological innovation offers a promising future. AI's role in addressing the challenges posed by wildfires and lung health is a compelling testament to progress, showcasing an effective response and a vision of what might be. 
The potential of AI in transforming the landscape of medicine and health is immense. This new chapter in diagnostics and treatment is not just a medical footnote but a significant stride toward improved care and understanding. As we grapple with the multifaceted effects of wildfires, technology offers a vision of hope and a path to a more responsive and empathetic healthcare system. 


wildfire, lung health, lung cancer

Share this story