Text this: Post-processing methods for mitigating algorithmic bias in healthcare classification models