Predicting breast cancer risk using personal health data and machine learning models

Abstract

Among women, breast cancer is a leading cause of death. Breast cancer risk predictions can inform screening and preventative actions. Previous works found that adding inputs to the widely-used Gail model improved its ability to predict breast cancer risk. However, these models used simple statistical architectures and the additional inputs were derived from costly and / or invasive procedures. By contrast, we developed machine learning models that used highly accessible personal health data to predict five-year breast cancer risk. We created machine learning models using only the Gail model inputs and models using both Gail model inputs and additional personal health data relevant to breast cancer risk. For both sets of inputs, six machine learning models were trained and evaluated on the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial data set. The area under the receiver operating characteristic curve metric quantified each model’s performance. Since this data set has a small percentage of positive breast cancer cases, we also reported sensitivity, specificity, and precision. We used Delong tests (p < 0.05) to compare the testing data set performance of each machine learning model to that of the Breast Cancer Risk Prediction Tool (BCRAT), an implementation of the Gail model. None of the machine learning models with only BCRAT inputs were significantly stronger than the BCRAT. However, the logistic regression, linear discriminant analysis, and neural network models with the broader set of inputs were all significantly stronger than the BCRAT. These results suggest that relative to the BCRAT, additional easy-to-obtain personal health inputs can improve five-year breast cancer risk prediction. Our models could be used as non-invasive and cost-effective risk stratification tools to increase early breast cancer detection and prevention, motivating both immediate actions like screening and long-term preventative measures such as hormone replacement therapy and chemoprevention.

A machine learning approach to personalized dose adjustment of lamotrigine using noninvasive clinical parameters

“…Third, our models were trained and tested on the derivation cohort, and validated on another validation cohort randomly split from the entire dataset. Such external validation verifies the generalizability of our models to new data 36 . Finally, the superiority of ML in terms of developing predictive models becomes clearer on larger populations with greater numbers of predictors.…”

The Use of Modified K-Means Algorithm to Enhance the Performance of Support Vector Machine in Classifying Breast Cancer

“…Stark et al [21] developed six models of machine learning by applying Gaussian naïve Bayes, decision tree, discriminant analysis, logistic regression analysis, support vector machine, and feed-forward ANN. They yielded five-year breast cancer risk estimates that are more accurate than those resulted using only BCRAT tools.…”

Developing a Diagnostic Decision Support Tool With Machine Learning Classification Algorithms to Improve Breast Cancer Screening: A Cross-Sectional Study on Iranian Women

“…Following the progress in the computational power and data storage systems, such techniques are applied extensively in the healthcare eld 18,19 . Researchers have implemented machine learning to address several aspects of breast cancer management [20][21][22][23] and some works have previously used statistical or machine learning algorithms on various categories of predictors to provide breast cancer risk scoring systems [24][25][26] . However, each speci c clinical problem requires a scale that receives its inputs from the data that is available to its intended end-users, and performs well in the relevant patient population and healthcare setting 27,28 .…”