Loading...

Analysis of Status, Treatment, and Complications in Pediatric Patients with Neuroblastoma

Hesami, Parnian | 2025

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58737 (01)
  4. University: Sharif University of Technology
  5. Department: Industrial Engineering
  6. Advisor(s): Rafiee, Majid
  7. Abstract:
  8. Neuroblastoma is one of the most common extracranial solid tumors in children, accounting for a significant proportion of childhood cancer cases and associated with high mortality. This study aimed to classify and predict survival and relapse in patients with neuroblastoma based on clinicopathological data collected from patients at the Mahak Pediatric Cancer Hospital. In this context, machine learning algorithms were employed, which, by enabling the analysis of complex data, allowed for more accurate prediction of clinical outcomes. To establish a comprehensive platform for analyzing and evaluating different models, a heterogeneous ensemble framework was designed in this study. By combining diverse machine learning algorithms, this framework enabled simultaneous comparison and evaluation of model performance, providing a more complete picture of their ability to classify patients. The developed framework played a key role in creating a structured and systematic analytical environment, facilitating methodical comparisons across multiple models. To enhance interpretability and better understand the performance of machine learning models, the SHAP method was employed. This approach allowed for the identification and quantification of the impact of each clinicopathological feature on the classification of patient survival and relapse. Analysis of SHAP values revealed that certain features played a critical role in model decision-making, and complex interactions between variables could also be observed. This analysis not only provided a deeper understanding of the models’ decision-making processes but also aided in identifying high-risk factors and prioritizing them for clinical management of neuroblastoma patients. Overall, the results of this study demonstrated that clinicopathological data can serve as valuable resources for predicting treatment outcomes in neuroblastoma patients, a domain previously dominated by gene expression data. Additionally, the XGBoost algorithm was identified as the most accurate model in classifying survival and relapse, the heterogeneous ensemble framework provided a more comprehensive assessment of model performance, and SHAP analysis highlighted key variables and high-risk factors
  9. Keywords:
  10. Neuroblastoma ; Survival Prediction ; Machine Learning ; Imbalanced Classification ; Shapley Additive Explanations (SHAP)Analysis ; Imbalanced Classification ; Survival and Relapse Prediction

 Digital Object List

 Bookmark

...see more