JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Predicting doxorubicin-induced cardiotoxicity in breast cancer: leveraging machine learning with synthetic data

  • 0Huna, São Paulo, SP, Brazil. dani@huna-ai.com.
Medical & Biological Engineering & Computing +

|

January 19, 2025

|

January 19, 2025

View abstract on PubMed

Summary

This summary is machine-generated.

A new model predicts Doxorubicin (DOXO) cardiotoxicity in breast cancer patients using routine blood tests and clinical data. This approach aids early identification of at-risk individuals, enabling timely interventions.

Area Of Science

  • Oncology
  • Cardiology
  • Biostatistics

Background

  • Doxorubicin (DOXO) chemotherapy is vital for breast cancer treatment.
  • DOXO-induced cardiotoxicity affects over 25% of patients, necessitating risk prediction.
  • Early identification of susceptible patients allows for alternative treatments or cardioprotective strategies.

Purpose Of The Study

  • To develop and validate a predictive model for DOXO-induced cardiotoxicity.
  • To assess the efficacy of the Data Augmentation and Smoothing (DAS) method for enhancing small medical datasets.
  • To identify key clinical and biomarker predictors of cardiotoxicity.

Main Methods

  • Analysis of data from 78 Brazilian breast cancer patients.
  • Application of the Data Augmentation and Smoothing (DAS) method to generate 4892 synthetic samples.
  • Integration of routine blood biomarkers (CRP, cholesterol profile, hematocrit, hemoglobin) and clinical data (BMI, smoking status).
  • Comparison of DAS with ADASYN, SMOTE, and SDV for synthetic data generation.

Main Results

  • The predictive model achieved an AUROC of 0.85±0.10, sensitivity of 0.89, and specificity of 0.69.
  • The DAS method demonstrated superior performance in synthetic data generation compared to ADASYN, SMOTE, and SDV.
  • The model effectively integrated routine blood biomarkers and clinical measures for risk assessment.

Conclusions

  • The developed model shows potential as a screening tool for DOXO-induced cardiotoxicity in breast cancer patients.
  • The DAS method is a promising technique for generating high-fidelity synthetic medical data, particularly for small sample sizes.
  • This study pioneers a specific cardiotoxicity prediction model for DOXO, improving patient management and outcomes.

Keywords: