KEY TAKEAWAYS
- The study aimed to use machine learning to predict mortality and relapse in children with ALL and identify key prognostic factors.
- The results showed that neural networks and bagging algorithms predicted mortality while boosting and random forest predicted relapse.
Acute Lymphoblastic leukemia (ALL) is a common and aggressive childhood cancer. Predicting mortality and relapse is crucial for effective treatment and follow-up management.
Zahra Mehrbakhsh and the team aimed to evaluate the effectiveness of machine learning models in predicting these outcomes in children with ALL.
A retrospective cohort study was conducted on 161 children under the age of 16 diagnosed with ALL. They evaluated the use of machine learning models to predict mortality and relapse in pediatric patients with ALL.
About 10 different machine-learning algorithms were used to analyze the data. The algorithms were trained on a dataset containing information on survival status, relapse occurrence, and various clinical and demographic factors. Model performance was evaluated using cross-validation, and metrics such as sensitivity, specificity, accuracy, and area under the curve (AUC) were reported.
The results showed that the average accuracy of the machine learning algorithms in predicting mortality ranged from 64% to 74%, and in predicting relapse, it ranged from 64% to 84%.
They found that age at diagnosis, hemoglobin levels, and platelet count were the most significant predictors of both mortality and relapse. Additionally, splenomegaly, hepatomegaly, and lymphadenopathy were identified as significant prognostic factors for mortality.
The study concluded that Artificial Neural Networks and Bagging algorithms were the most accurate at predicting mortality while Boosting and Random Forest algorithms were the most accurate for predicting relapse in children with ALL.
The study highlighted the potential of machine learning in predicting outcomes for children with ALL and identified key prognostic factors that can inform treatment decisions and potentially improve patient outcomes.
This study was supported and approved by Hamadan University of Medical Sciences.
Source: https://pubmed.ncbi.nlm.nih.gov/39285373/
Mehrbakhsh Z, Hassanzadeh R, Behnampour N, et al. (2024). “Machine learning-based evaluation of prognostic factors for mortality and relapse in patients with acute lymphoblastic leukemia: a comparative simulation study.” BMC Med Inform Decis Mak. 2024;24(1):261. Published 2024 Sep 16. doi:10.1186/s12911-024-02645-6
