Breast Cancer Prognosis Using a Machine Learning Approach

Abstract

Machine learning (ML) has been recently introduced to develop prognostic classification models that can be used to predict outcomes in individual cancer patients. Here, we report the significance of an ML-based decision support system (DSS), combined with random optimization (RO), to extract prognostic information from routinely collected demographic, clinical and biochemical data of breast cancer (BC) patients. A DSS model was developed in a training set (n = 318), whose performance analysis in the testing set (n = 136) resulted in a C-index for progression-free survival of 0.84, with an accuracy of 86%. Furthermore, the model was capable of stratifying the testing set into two groups of patients with low- or high-risk of progression with a hazard ratio (HR) of 10.9 (p < 0.0001). Validation in multicenter prospective studies and appropriate management of privacy issues in relation to digital electronic health records (EHR) data are presently needed. Nonetheless, we may conclude that the implementation of ML algorithms and RO models into EHR data might help to achieve prognostic information, and has the potential to revolutionize the practice of personalized medicine.

DeepBTS: Prediction of Recurrence-free Survival of Non-small Cell Lung Cancer Using a Time-binned Deep Neural Network

“…Previous survival analysis studies using machine learning involved the use of support vector machine-based prediction models (sensitivity = 0.89; specificity = 0.73) or decision support systems (C-index = 0.84 with an accuracy of 86%) to predict breast cancer recurrence 9,10 , and probabilistic neural networks to predict cervical cancer recurrence (sensitivity = 0.975; accuracy = 0.892) 11 . In their research on lung cancer, Lynch et al compared various supervised machine learning classification techniques using the Surveillance, Epidemiology, and End Results (SEER) database and showed that the models in which the gradient boosting machine was utilized with the root-mean-squared error (RMSE) were the most accurate 12 .…”

Artificial Intelligence and Patient-Centered Decision-Making

“…Honoring his epistemic obligation to rely on experts in medical decision 20 While our example is purely hypothetical, there are several studies that investigate the prospects of using machine learning systems for breast cancer stratification and treatment options. See, for instance, Ferroni et al (2019) and Xiao et al (2018). 21 Note: even if we assume that a post-hoc interpretation of such a network was possible-for instance through clustering "significant" units in the network to yield information that those specific 1237 patient variables with those specific interconnected range values result in that specific risk estimate-it is far from clear how such an interpretation could be of practical use to the scientific community.…”

Prediction of Colon Cancer Stages and Survival Period with Machine Learning Approach

“…It is difficult to correlate the features manually and predict the outcome of a patient in terms of tumor staging or DFS period. Therefore, we used different ML classifiers such as Random Forest, Support Vector Machine (SVM), Logistic Regression (LR), Multilayer Perceptrons (MLP), K-Nearest Neighbors (KNN), and Adaptive Boosting (AdaBoost) [22][23][24]28], which are popularly used in medical data analysis. Implementing different classifiers in Scikit-learn for the multi-class classification, the scheme of one-against-one was used for the SVM and the scheme OvR was used for Logistic Regression (LR) and other models, which gave the average of all metrics used in our analysis.…”

“…In recent years, the use of machine learning has gained much importance in the field of clinical study [14,15]. Recently, many researchers and clinicians have proposed the use of machine learning in the field of colorectal cancer [16][17][18] and other types of cancers [19][20][21][22][23][24]. For instance, in [25] authors used ML for prostate cancer screening and have determined the impact of few variables such as rate of change of prostate-specific antigen (PSA), age, BMI, and race on the model's accuracy.…”