The use of artificial intelligence in critical applications can be limited due to a lack of model understanding. Genetic programming (GP) offers an alternative to black-box methods like deep neural networks by evolving interpretable function representations. In this chapter, we present recent advances in the application of GP to the analysis of biomedical dataBiomedical data, focusing on the challenges of evolving interpretable models for real-world datasets. We first detail Kartezio, a Cartesian Genetic Programming (CGP) approach that generates interpretable pipelines for biomedical image segmentation. We demonstrate that Kartezio can achieve competitive performance with state-of-the-art deep learning methods while requiring significantly smaller training datasets. This is particularly important in the biomedical domain, where fully annotated data is scarce. We further explore the use of small labeled datasets by proposing a data sampling mechanism based on active learningActive learning methods. Finally, real biomedical dataBiomedical data often involves multiple data types, from numerical integers to complex multichannel images. We present the Multimodal Active Genetic Evolution (MAGE) algorithm, which extends CGP’s representation to include functions of different types, opening new possibilities beyond segmentation. Through this work, we demonstrate that GP can be a powerful tool for high-dimensional data analysis, providing interpretable models that can be usedBiomedical data inImage processing critical applications.