A Predictive Model for Pathologic Complete Response in Breast Cancer Patients Treated with Neoadjuvant Chemotherapy Using Machine Learning

Abstract

Background: In patients with breast cancer after Neoadjuvant Chemotherapy (NAC), pathological Complete Response (pCR) was associated with better long-term outcomes. We here attempted to predict pCR using machine learning. Patients and Methods: From 2008 to 2017, 1308 breast cancer patients underwent NAC before surgery, of whom 377 patients underwent Cancer SCAN TM for gene data. Of 377, 238 were analyzed here, with 139 excluded due to incomplete medical data. Results: The pCR (−) vs. (+) group had 200 vs. 38 patients. In our predictive model with gene data, the Area Under the Curve (AUC) of the Receiver Operating Characteristic (ROC) curve was 0.909 and accuracy was 0.875. In another model without gene data, the AUC of ROC curve was 0.743 and accuracy was 0.800. We also conducted internal validation with 72 patients undergoing NAC and Cancer SCAN TM during July 2017 and April 2018. When we applied a 0.4 threshold value, accuracy was 0.806 and 0.778 in the predictive model with vs. without gene profiles, respectively. Conclusion: The present predictive model may be a useful and easy-toaccess tool for pCR-prediction in breast cancer patients treated with NAC.

Hyperparameter Tuning Based Machine Learning Classifier for Breast Cancer Prediction

“…In this investigation, Kim et al [4] presented a simple to use machine learning prediction tool for pathological Complete Response (pCR) in breast cancer survivors medicated with Neoadjuvant Chemotherapy (NAC) and generated their training set by using Two-class Bayes point machine technique. They made use of information from clinical traits and gene xpression patterns.…”