Classification of patient multicategory survival outcomes is important for personalized cancer treatments. Machine learning (ML) algorithms have increasingly been used to inform healthcare decisions, but these models are vulnerable to biases in data collection and algorithm creation. ML models have previously been shown to exhibit racial bias, but their fairness towards patients from different age and sex groups have yet to be studied. Therefore, we compared the multimetric performances of five ML models (random forests, multinomial logistic regression, linear support vector classifier, linear discriminant analysis, and multilayer perceptron) when classifying colorectal cancer patients (n = 589) of various age, sex, and racial groups using The Cancer Genome Atlas data. All five models exhibited biases for these sociodemographic groups. We then repeated the same process on lung adenocarcinoma (n = 515) to validate our findings. Surprisingly, most models tended to perform more poorly overall for the largest sociodemographic groups. Methods to optimize model performance, including testing the model on merged age, sex, or racial groups, and creating a model trained on and used for an individual or merged sociodemographic group, show potential to reduce disparities in model performance for different groups. This is supported by our regression analysis showing associations between model choice and methodology used with reduced performance disparities across demographic subgroups. Notably, these methods may be used to improve ML fairness while avoiding penalizing the model for exhibiting bias and thus sacrificing overall performance.
Feng et al. (Tue,) studied this question.