Predictive modeling of methadone poisoning outcomes in children ≤ 5 years: utilizing machine learning and the National Poison Data System for improved clinical decision-making.

Journal: European journal of pediatrics
PMID:

Abstract

UNLABELLED: The escalating therapeutic use of methadone has coincided with an increase in accidental ingestions, particularly among children ≤ 5 years. This study utilized machine learning (ML) methodologies on data from the National Poison Data System (NPDS) to predict pediatric methadone poisoning outcomes to enhance clinical decision-making. We analyzed 140 medical parameters from pediatric patient records. Pre-processing steps, including synthetic oversampling, addressed the imbalanced distribution of the outcome variable. We evaluated various ML models in multiclass classification tasks. Random forest showed versatility with an accuracy of 0.96 and a strong receiver operating characteristic area under the curve (ROC AUC) (0.98). Meanwhile, the support vector machine (SVM) had the highest negative predictive value (NPV) (0.64). Shapley Additive exPlanation (SHAP) analysis identified key predictors such as coma, cyanosis, respiratory arrest, and respiratory depression for predicting serious outcomes.

Authors

  • Omid Mehrpour
    Data Science Institute, Southern Methodist University, Dallas, TX, USA. omehrpour@smu.edu.
  • Samaneh Nakhaee
    Medical Toxicology and Drug Abuse Research Center (MTDRC), Birjand University of Medical Sciences (BUMS), Birjand, Iran.
  • Jafar Abdollahi
    Department of Computer Engineering, Ardabil Branch, Islamic Azad University, Ardabil, Iran.
  • Varun Vohra
    Department of Emergency Medicine, Michigan Poison & Drug Information Center, Wayne State University School of Medicine, Detroit, MI, USA.