Identifying Pregnancy in Clinical Notes

Identifying medical conditions in clinical notes can add business value to health care organizations. Pregnancy is one example of a condition that can be identified.

This project used the HuggingFace transformers library to fine-tune a large language model to identify pregnancy in clinical notes, and used Gradio to publish an app interface.

Key steps in this project:

• Used the kaggle API to download a publicly available dataset of clincial notes
• Augmented the dataset with synthetic data, and used upsampling to address class imbalance
• Used the HuggingFace transformers module to fine-tune an existing large language model to the classification task
• Used sklearn metrics to evaluate model performance on a test set
• Used the gradio module to develop an app to enable testing of use cases
• Published the fine-tuned model and Gradio app on HuggingFace