JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Innovative approach towards early prediction of ovarian cancer: Machine learning- enabled XAI techniques

  • 0School of Computer Science and Engineering, Vellore Institute of Technology, Chennai, India.
Heliyon +

|

December 13, 2024

|

December 13, 2024

View abstract on PubMed

Summary

This summary is machine-generated.

Machine learning and explainable AI enhance early ovarian cancer detection. Ensemble methods improved Support Vector Machines accuracy to 89%, aiding diagnosis and patient outcomes.

Area Of Science

  • Oncology
  • Artificial Intelligence
  • Women's Health

Background

  • Ovarian cancer poses significant global health challenges, necessitating improved early detection strategies.
  • Early diagnosis is crucial for improving patient outcomes and survival rates in ovarian cancer.
  • Machine learning (ML) and eXplainable Artificial Intelligence (XAI) offer promising tools for enhancing diagnostic accuracy.

Purpose Of The Study

  • To evaluate the effectiveness of various ML techniques for early ovarian cancer detection.
  • To assess the role of XAI in making ML predictions transparent and understandable for clinical application.
  • To investigate the integration of ML and AI in biomarker evaluation for ovarian cancer.

Main Methods

  • Utilized a dataset of 349 patients with known ovarian cancer status from Kaggle.
  • Applied preprocessing techniques including feature scaling and dimensionality reduction.
  • Employed Minimum Redundancy Maximum Relevance (MRMR) for feature selection and evaluated K Nearest Neighbors, Support Vector Machines, Decision Trees, and ensemble methods (Max Voting, Boosting, Bagging, Stacking).
  • Integrated XAI, specifically Shapley values, to interpret ML model decisions.

Main Results

  • Support Vector Machines achieved an 85% base model accuracy.
  • Ensemble learning techniques, particularly stacking, improved accuracy to 89%.
  • XAI provided deeper insights into ML model decision-making processes.

Conclusions

  • ML and XAI integration presents a viable and effective approach for early ovarian cancer detection.
  • Ensemble learning methods significantly enhance predictive accuracy.
  • This research contributes a promising strategy for improving oncology diagnostics and women's health.

Keywords: