Patient Journey Clustering

Overview

This repository contains code for creating patient journey clusters based on cases represented by string sequences. Each character in the string corresponds to a specific event or action in a clinical document, encoded by True/False values. Time gaps between events are also represented in these sequences: a gap of more than 24 hours but less than 72 hours is indicated by an 'x', and a gap of more than 72 hours is indicated by 'xx'.

Mappings

The character mappings in the sequences are as follows:

• "a": consult with nothing noted
• "b": hospitalization
• "c": given a new antibiotic
• "d": given a new antibiotic and hospitalized
• "e": surgical intervention performed
• "f": surgical intervention performed with hospitalization
• "g": surgical intervention performed with new antibiotic given
• "h": surgical intervention performed with new antibiotic given with hospitalization
• "i": abscess was noted draining on exam
• "j": abscess was noted draining on exam and patient hospitalized
• "k": abscess was noted draining on exam and new antibiotic given
• "l": abscess was noted draining on exam, a new antibiotic given, and patient was hospitalized
• "m": abscess was noted draining on exam and surgical intervention was performed
• "n": abscess was noted draining on exam, surgical intervention was performed, and patient was hospitalized
• "o": abscess was noted draining on exam, surgical intervention was performed, and new antibiotic was given
• "p": abscess was noted draining on exam, surgical intervention was performed, a new antibiotic was given, and patient was hospitalized
• "x": time gap

Instructions for Use

To use the code in this repository:

1. Refer to the Patient_Journey_Clustering.ipynb notebook for an example of how to use the code.
2. Ensure that all required libraries listed in requirements.txt are installed.

Citation

When referencing the use of this code, please cite the following paper:

Hur, Brian, Karin M. Verspoor, Timothy Baldwin, Laura Y. Hardefeldt, Caitlin Pfeiffer, Caroline Mansfield, Riati Scarborough, and James R. Gilkerson. “Using Natural Language Processing and Patient Journey Clustering for Temporal Phenotyping of Antimicrobial Therapies for Cat Bite Abscesses.” Preventive Veterinary Medicine 223 (2024): 106112. https://doi.org/10.1016/j.prevetmed.2023.106112.