Bioinformatics Core Tools

NLP Tools

The Chart Review Tool Powered by NLP (CHANL) is designed to facilitate chart review of narrative text notes from electronic medical records (EMR). It can be used to perform intelligent searches, not only for multiple keywords simultaneously, but can automatically identify other terms and concepts related to the keyword in the notes. The interface highlights identified information in the notes, allowing users to easily locate the information needed. Current users have found this to significantly improve the efficiency and accuracy of chart reviews. This tool was developed by Tianrun Cai, MD, Brigham and Women’s Hospital.

EXTraction of EMR Numerical Data (EXTEND) was developed by Tianrun Cai, MD, Brigham and Women’s Hospital, Katherine Liao, MD, MPH, Brigham and Women’s Hospital, Frank Rybicki, MD, PhD, University of Ottawa, and Tianxi Cai, ScD, Harvard T.H. Chan School of Public Health. EXTEND is a natural language processing (NLP) tool that can efficiently extract numerical clinical data from different types of narrative notes with high accuracy. By expanding the dictionary and developing new rules, the usage of EXTEND can be easily expanded to extract additional numerical data important in clinical outcomes research.

To gain access to EXTEND, please register for VERITY Membership by selecting the “Become a Member” option in the upper right corner. VERITY Membership is free. If you are already a member or have received your membership credentials, please click here.

Reference/Acknowledgments: Cai, T. et al. BMC Med Inform Decis Mak. 2019

Narrative Information Linear Extraction (NILE) is an efficient and effective software for natural language processing (NLP) of clinical narrative texts. It uses a prefix tree algorithm for named entity recognition, and finite-state machines for semantic analysis, both of which were inspired by the natural reading behavior of humans. The design aims to directly translate linguistic and clinical knowledge to code, allowing for the development of functions to parse complex language patterns.

The software was developed by Sheng Yu and Tianxi Cai at Harvard T.H. Chan School of Public Health and Tianrun Cai at The Brigham and Women’s Hospital. It is distributed free of charge for academic and non-commercial research use by the President and Fellows of Harvard College.

Reference/Acknowledgments: Yu, S. et al. arXiv. 2013

Phenotyping Tools & Resources

Online Narrative and Codified feature Search Engine is a search engine that identifies related narrative and codified EHR features based on a target phenotype or condition; can be used as initial step to identify potential EHR features for an algorithm.

Reference/Acknowledgments: Xiong, X. et al. medRxiv. 2023

Centralized Interactive Phenomics Resource is a knowledgebase of computable EHR-phenotypes hosted at VA to standardize phenotypes across various patient populations, EHRs, research projects, and operations.

Reference/Acknowledgments: Honerlaw, J. et al. JAMIA. 2024

This is a hierarchical mapping tool for PheCodes to ICD-9 and ICD-10 billing codes to represent clinically relevant phenotypes. 

Reference/Acknowledgments: Wei, WQ. et al. PLoS One. 2017

Phenotyping Algorithm Pipelines

KOMAP
Knowledge-Driven Online Multimodal Automated Phenotyping (KOMAP) System is an automated method to train a phenotyping algorithm using codified, e.g., ICD and NLP data; uses information from ONCE tool.

Reference/Acknowledgments: Xiong, X. et al. medRxiv. 2023

High-throughput PHenotyping with EHR using a Common Automated Pipeline
The PheCAP package is a Semi-supervised method using codified and NLP data to classify patients with a specific phenotype that implements surrogate-assisted feature extraction (SAFE) and common machine learning approaches to train and validate phenotyping models. PheCAP begins with data from the EMR, including structured data and information extracted from the narrative notes using natural language processing (NLP). The standardized steps integrate automated procedures, which reduce the level of manual input, and machine learning approaches for algorithm training.

Reference/Acknowledgments: Zhang, Y. et al. Nat Protoc. 2019

Multimodal Automated Phenotyping (MAP) is an algorithm that yields a predicted probability of phenotype for each patient and a threshold for classifying subjects with phenotype yes/no using ICD, NLP counts, and healthcare utilization data.

Reference/Acknowledgments: Liao, KP. et al. JAMIA. 2019

Back To Top