Application of machine learning techniques in the prediction of excess lifetime cancer risks of agricultural byproducts used as building and construction materials
Solomon Oyebisi, H. I. Owamah
Abstract
Recycling improves the circular economy and resource sustainability by using agricultural waste to create new products. However, agricultural byproducts resulting from the recycling of agricultural waste materials contain naturally occurring radionuclides with potential risks to human health and the environment. Therefore, this article provides an overview of relevant literature on radiological properties of agricultural byproducts (rice husk ash, mussel shell, palm oil clinker, and palm oil fuel ash), prioritizing their specific activities ( 226 Ra, 232 Th and 40 K). Consequently, absorbed gamma dose rates (AGDR), annual effective dose rates (AEDR) and excess lifetime cancer risks (ELCR) of the agricultural byproducts studied were determined. The specific concentrations, AGDR, AEDR, and ECLR were trained, validated, and tested using various machine learning algorithms. An evaluation of the radiological properties of all agricultural byproducts examined revealed that they pose no risk of cancer. Additionally, compared to support vector machine, regression trees, ensemble trees, Gaussian process regression, and neural networks, linear regression yielded the best performance metrics, making it the most suitable technique for predicting excess lifetime cancer risks of the surveyed agricultural byproducts. • A comprehensive review of the radiological properties of agricultural byproducts is presented. • Agricultural byproducts and how they pose excess lifetime cancer risks are highlighted. • Excess lifetime cancer risks are modeled using six major Machine Learning algorithms (MLAs). • This study provides valuable references and guidance for scientists and researchers. • More comprehensive datasets can improve the performance of the developed models.