Identifying immunosuppression

Owner: Vijeeth Guggilla

Guggilla, V. et al. Large language models outperform traditional structured data-based approaches in identifying immunosuppressed patients. 2025.01.16.25320564 Preprint at https://doi.org/10.1101/2025.01.16.25320564 (2025).

Overview

Identifying immunosuppressive conditions and immunosuppressive medications using structured data (diagnosis codes, medication orders) and using LLMs.

Files

• edw_queries/: this folder contains the SQL queries for pulling all SCRIPT patients diagnosis codes and dates for patients with a given immunosuppressive condition. it also contains Meds.sql for pulling medication information.
• ICD_identifier.ipynb: whole workflow for ICD diagnosis code identification of each condition and metrics generation
• meds_identifier.ipynb: whole workflow for medication order identification of each medication and metrics generation
• LLM_identifier.ipynb: whole workflow for LLM identification of each condition/medication and metrics generation
• local_LLM_identifier.ipynb: whole workflow for local LLM (Llama 3.1) identification of each condition/medication and metrics generation