RESUMO
Purpose: Some patients with breast cancer treated by surgery and radiation therapy experience clinically significant toxicity, which may adversely affect cosmesis and quality of life. There is a paucity of validated clinical prediction models for radiation toxicity. We used machine learning (ML) algorithms to develop and optimise a clinical prediction model for acute breast desquamation after whole breast external beam radiation therapy in the prospective multicenter REQUITE cohort study. Methods and Materials: Using demographic and treatment-related features (m = 122) from patients (n = 2058) at 26 centers, we trained 8 ML algorithms with 10-fold cross-validation in a 50:50 random-split data set with class stratification to predict acute breast desquamation. Based on performance in the validation data set, the logistic model tree, random forest, and naïve Bayes models were taken forward to cost-sensitive learning optimisation. Results: One hundred and ninety-two patients experienced acute desquamation. Resampling and cost-sensitive learning optimisation facilitated an improvement in classification performance. Based on maximising sensitivity (true positives), the "hero" model was the cost-sensitive random forest algorithm with a false-negative: false-positive misclassification penalty of 90:1 containing m = 114 predictive features. Model sensitivity and specificity were 0.77 and 0.66, respectively, with an area under the curve of 0.77 in the validation cohort. Conclusions: ML algorithms with resampling and cost-sensitive learning generated clinically valid prediction models for acute desquamation using patient demographic and treatment features. Further external validation and inclusion of genomic markers in ML prediction models are worthwhile, to identify patients at increased risk of toxicity who may benefit from supportive intervention or even a change in treatment plan.
RESUMO
The prediction by classification of side effects incidence in a given medical treatment is a common challenge in medical research. Machine Learning (ML) methods are widely used in the areas of risk prediction and classification. The primary objective of such algorithms is to use several features to predict dichotomous responses (e.g., disease positive/negative). Similar to statistical inference modelling, ML modelling is subject to the class imbalance problem and is affected by the majority class, increasing the false-negative rate. In this study, seventy-nine ML models were built and evaluated to classify approximately 2000 participants from 26 hospitals in eight different countries into two groups of radiotherapy (RT) side effects incidence based on recorded observations from the international study of RT related toxicity "REQUITE". We also examined the effect of sampling techniques and cost-sensitive learning methods on the models when dealing with class imbalance. The combinations of such techniques used had a significant impact on the classification. They resulted in an improvement in incidence status prediction by shifting classifiers' attention to the minority group. The best classification model for RT acute toxicity prediction was identified based on domain experts' success criteria. The Area Under Receiver Operator Characteristic curve of the models tested with an isolated dataset ranged from 0.50 to 0.77. The scale of improved results is promising and will guide further development of models to predict RT acute toxicities. One model was optimised and found to be beneficial to identify patients who are at risk of developing acute RT early-stage toxicities as a result of undergoing breast RT ensuring relevant treatment interventions can be appropriately targeted. The design of the approach presented in this paper resulted in producing a preclinical-valid prediction model. The study was developed by a multi-disciplinary collaboration of data scientists, medical physicists, oncologists and surgeons in the UK Radiotherapy Machine Learning Network.
