Lung and bronchial cancer (LBC) is one of the most common causes of cancer deaths worldwide, accounting for 11.6% of all cancer deaths in 2018. In France, as in the United States, it is the leading cause of death from cancer and if smoking plays a major role in the disease, pollution and socio-economic status are also at risk. A team at the University of Buffalo wanted to understand why these factors do not have the same consequences depending on where patients live. They have published the results of their research under the title “. Explicit artificial intelligence to explore the spatial variability of lung and bronchial cancer mortality in the adjoining United States. In the December 2021 scientific report.
The study, which identified the main risk factors for LBC death using artificial intelligence (XAI), brought together an interdisciplinary team:
- Zia U Ahmed, PhD, Database / Visualization Specialist at UB Renew Institute;
- Kang Sun, PhD, Senior Faculty Member at UB Renew Institute and Assistant Professor of Civil, Structural and Environmental Engineering at UB School of Engineering and Applied Sciences;
- Michael Shelley, PhD, Environmental / Environmental Economist at UB Renew Institute;
- Lina Mu, PhD, MD, Associate Professor of Epidemiology and Environmental Health at UB School of Public Health and Health Professionals.
Zia Yu Ahmed says of the research:
“The results are important because the United States is a spatially diverse environment. There is a wide variety of socio-economic factors and levels of education – basically, one size does not fit everyone. Here, the local interpretation of machine learning models is more important than the global interpretation. A
Lina Mu added:
“The study could be a model for integrating artificial intelligence into an epidemiological study. It could also serve as an example of using predictive models when studying cancer. It could help identify high-risk areas where the cancer registry is not available.” A
The UB team applied Explainable Artificial Intelligence (XAI) to a stack set machine learning model framework to explore and visualize the spatial distribution of known risk factors contributing to lung and bronchial cancer deaths in the adjacent United States. To create the Stack Ensemble model, he used Generalized Linear Model (GLM), Random Forest (RF), Gradient Boost Machine (GBM), Extreme Gradient Boost Machine (XGBoost).
Zia U Ahmed explained:
“XAI is still lacking in local interpretation, especially when it comes to the environment and science. A
The risk factors explored by the study represent lifestyle changes including smoking, socio-economic status (poverty rate), population, air pollution, physical environment, bio-physical factors as well as health insurance.
Studies have shown that smoking rates were associated with higher levels of poverty and ethnicity, with Hispanics, for example, smoking less than whites. It also shows a strong correlation between poverty and therefore lack of access to care for AML mortality in the United States.
In the case of air pollution, researchers have examined the pollutant nitrogen dioxide (NO2), sulfur dioxide (SO2), ozone and particulate matter, and their spatial variability in lung cancer mortality and bronchi.
Smoking and poverty have been shown to be the two main risk factors for lung and bronchial cancer. If these results are to be expected, this study demonstrates the potential for the implementation of interpreted artificial intelligence as a complement or replacement of traditional spatial regression models.
Article Sources: Ahmed, ZU, Sun, K., Shelley, M. Etc. Interpretive Artificial Intelligence (XAI) to explore the spatial variability of lung and bronchial cancer (LBC) mortality rates in the adjacent United States. Science-representative 11, 24090 (2021). https://doi.org/10.1038/s41598-021-03198-8