Artificial Intelligence understands Cancer Mortality Rates

When the word “artificial intelligence (AI)” is mentioned, many people immediately think of robots. In a new study on lung and bronchus cancer (LBC) in the United States, however, AI refers to a collection of machine learning models stacked together to create high-level predictions regarding LBC Mortality rates.

The new study, co-authored by University at Buffalo academics Zia U. Ahmed, Kang Sun, Michael Shelly, and Lina Mu, uses explainable artificial intelligence (XAI) to identify key risk variables for LBC Mortality. While smoking prevalence, poverty, and elevation of a community were the most important risk variables in predicting LBC death rates, relationships between risk factors and LBC Mortality rates were found to vary regionally, and the research looked into these variances.

“The results significant because the United States is a spatially varied ecosystem,” Ahmed said of the study and research. There is a vast range of socioeconomic factors and educational levels, so there is no one-size-fits-all solution. Local machine learning model interpretation is more crucial than global interpretation in this case. XAI in local interpretation, particularly in relation to the environment and science, is still inadequate.

He goes on to say that the findings can help with public health management and action by highlighting which areas require assistance. The work used a combination of ensemble machine learning and explainable algorithms to spatially portray connections between LBC Mortality and risk factors in the United States, which is a significant step forward in this field. With more data and numerous models, AI algorithms operate better, which is why the stack-ensemble is more valuable than any single model.