Incorporation of explainable artificial intelligence in ensemble machine learning-driven pancreatic cancer diagnosis DOI Creative Commons
Faisal Almisned, Natacha Usanase, Declan Ikechukwu Emegano

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 23, 2025

Abstract Despite the strides made in medical science, pancreatic cancer continues to be a threat, highlighting urgent need for creative strategies address this concern. Recently, potential approach that has attracted significant attention is using machine learning clinical decision-making. This research aims analyze six algorithms, and an ensemble voting classifier, develop hybrid models early detection of based on several characteristics interpret their performance with Shapley Additive Explanations (SHAP). A publicly available dataset composed 590 patient urine samples was utilized conventional classification cancerous from non-cancerous cases through analysis specific attributes. An classifier developed best-performed single models, which were later hybridized form novel models. The outperformed all stand-alone accuracy 96.61% precision 98.72%. exhibited higher than random forest model outperforming others AUC 99.05% (95% confidence interval (CI): 0.93-1.00) interpretation given by SHAP showing top influential features diagnosis greatest positive values. Employing rapid sophisticated high holds promise facilitating effective various diseases, including cancer.

Language: Английский

Incorporation of explainable artificial intelligence in ensemble machine learning-driven pancreatic cancer diagnosis DOI Creative Commons
Faisal Almisned, Natacha Usanase, Declan Ikechukwu Emegano

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: April 23, 2025

Abstract Despite the strides made in medical science, pancreatic cancer continues to be a threat, highlighting urgent need for creative strategies address this concern. Recently, potential approach that has attracted significant attention is using machine learning clinical decision-making. This research aims analyze six algorithms, and an ensemble voting classifier, develop hybrid models early detection of based on several characteristics interpret their performance with Shapley Additive Explanations (SHAP). A publicly available dataset composed 590 patient urine samples was utilized conventional classification cancerous from non-cancerous cases through analysis specific attributes. An classifier developed best-performed single models, which were later hybridized form novel models. The outperformed all stand-alone accuracy 96.61% precision 98.72%. exhibited higher than random forest model outperforming others AUC 99.05% (95% confidence interval (CI): 0.93-1.00) interpretation given by SHAP showing top influential features diagnosis greatest positive values. Employing rapid sophisticated high holds promise facilitating effective various diseases, including cancer.

Language: Английский

Citations

0