The field of cancer diagnosis categorization is moving towards leveraging large language models and natural language processing techniques to improve the accuracy and efficiency of clinical text analysis. Recent studies have focused on evaluating the performance of various large language models, such as BioBERT and GPT-4o, in classifying cancer diagnoses from electronic health records. The results have shown promising performance, with some models achieving high weighted macro F1-scores and accuracy. However, common misclassification patterns and the need for standardized documentation practices and robust human oversight remain challenges. Noteworthy papers include: Cancer Diagnosis Categorization in Electronic Health Records Using Large Language Models and BioBERT, which evaluated the performance of four large language models and BioBERT in classifying cancer diagnoses. Unlocking Public Catalogues: Instruction-Tuning LLMs for ICD Coding of German Tumor Diagnoses, which investigated the use of instruction-based fine-tuning to improve the coding accuracy of open-weight LLMs for German tumor diagnosis texts.