Bladder cancer remains a significant global health challenge, with rising incidence rates and persistent challenges in early diagnosis and personalized treatment. This study proposes machine learning and explainable artificial intelligence (XAI) to predict Microsatellite Instability (MSI) thresholds in bladder cancer, a biomarker with emerging prognostic and therapeutic implications. Using a dataset from cBioPortal.org, we conducted comprehensive data preprocessing, feature engineering, and exploratory data analysis (EDA) to identify key clinical, demographic, and genetic predictors. Categorical variables were processed using one-hot encoding, and correlation-based filtering was employed to address multicollinearity, retaining clinically relevant features such as Tumor Mutational Burden (TMB) and Fraction Genome Altered. This work evaluates multiple machine learning models, including Logistic Regression, Random Forest, and XGBoost, with hyperparameter tuning through grid search to optimize classification performance. Model interpretability was enhanced through SHAP values, which revealed the influence of critical features like Age at Diagnosis and Mutation Count. The XGBoost model achieved the highest accuracy (0.822) and demonstrated robust predictive capabilities, though Logistic Regression and Random Forest also performed competitively. Our findings underscore the importance of genetic factors in bladder cancer prognosis and highlight the potential of machine learning to improve risk stratification and personalized treatment strategies. However, limitations such as dataset heterogeneity, computational constraints, and the need for external validation underscore the necessity for further research. This study contributes to the growing body of evidence supporting the integration of machine learning and XAI in oncology, leading to a better understanding of the development of the disease.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Explainable AI for Clinical Decision-Making: Unlocking the Potential of MSI Thresholds in Bladder Cancer

  • Amro Sayed Ahmad,
  • Firas Alghanim,
  • Bashier Elkarami,
  • Mousa AbuGhosh,
  • Hazem Qattous,
  • Abedalrhman Alkhateeb

摘要

Bladder cancer remains a significant global health challenge, with rising incidence rates and persistent challenges in early diagnosis and personalized treatment. This study proposes machine learning and explainable artificial intelligence (XAI) to predict Microsatellite Instability (MSI) thresholds in bladder cancer, a biomarker with emerging prognostic and therapeutic implications. Using a dataset from cBioPortal.org, we conducted comprehensive data preprocessing, feature engineering, and exploratory data analysis (EDA) to identify key clinical, demographic, and genetic predictors. Categorical variables were processed using one-hot encoding, and correlation-based filtering was employed to address multicollinearity, retaining clinically relevant features such as Tumor Mutational Burden (TMB) and Fraction Genome Altered. This work evaluates multiple machine learning models, including Logistic Regression, Random Forest, and XGBoost, with hyperparameter tuning through grid search to optimize classification performance. Model interpretability was enhanced through SHAP values, which revealed the influence of critical features like Age at Diagnosis and Mutation Count. The XGBoost model achieved the highest accuracy (0.822) and demonstrated robust predictive capabilities, though Logistic Regression and Random Forest also performed competitively. Our findings underscore the importance of genetic factors in bladder cancer prognosis and highlight the potential of machine learning to improve risk stratification and personalized treatment strategies. However, limitations such as dataset heterogeneity, computational constraints, and the need for external validation underscore the necessity for further research. This study contributes to the growing body of evidence supporting the integration of machine learning and XAI in oncology, leading to a better understanding of the development of the disease.