Papers by Kirk Roberts

7 papers
DrugEHRQA: A Question Answering Dataset on Structured and Unstructured Electronic Health Records For Medicine Related Queries (2022.lrec-1)

Copied to clipboard

Challenge: a new question answering dataset is being developed for electronic health records . structured tables and unstructured notes can be duplicated, contradictory or provide additional context .
Approach: They develop a question-answer-matching dataset using structured tables and unstructured notes from an EHR.
Outcome: The proposed model is based on a model with a modality selection network . it uses the prediction of a RAT-SQL to choose between EHR tables and clinical notes .
A Cross-document Coreference Dataset for Longitudinal Tracking across Radiology Reports (2022.lrec-1)

Copied to clipboard

Challenge: Oftentimes, these findings and devices are referred to multiple times in a single report and are also referred across different reports of a patient.
Approach: They propose a new cross-document coreference resolution (CDCR) dataset for identifying co-referring radiological findings and medical devices across a patient's radiology reports.
Outcome: The proposed dataset contains 5872 mentions (findings and devices) spanning 638 MIMIC-III radiology reports across 60 patients, covering multiple imaging modalities and anatomies.
Extracting Adherence Information from Electronic Health Records (2020.coling-main)

Copied to clipboard

Challenge: adherence is a critical factor in health outcomes, and is often modeled as a binary decision . adherence models include intentional and unintentional non-adherence, social support and other patient attributes such as age and time since diagnosis.
Approach: They propose to extract adherence information from electronic health records using de-identified sentences and a corpus of 3,000 de-identified sentences.
Outcome: The proposed framework extracts medication adherence information from electronic health records.
Evaluation of Dataset Selection for Pre-Training and Fine-Tuning Transformer Language Models for Clinical Question Answering (2020.lrec-1)

Copied to clipboard

Challenge: Existing work on EHR QA models that learn to answer questions from structured data has focused on analyzing questions or mapping questions to existing NLP based information extraction models.
Approach: They conduct 48 experiments on two clinical question answering datasets . they use open-domain and domain-specific corpora to fine-tune Transformer language models .
Outcome: The proposed models can learn to answer questions from unstructured notes with accuracies up to 90% on open-domain and domain-specific corpora.
A FrameNet for Cancer Information in Clinical Narratives: Schema and Annotation (L18-1)

Copied to clipboard

Challenge: Existing natural language processing (NLP) systems for cancer-related information are highly task-specific and often produce incompatible annotations and algorithms.
Approach: They propose a general-purpose natural language processing resource for cancer-related information in clinical notes . the project uses a frame semantic method to emphasize the information presented in the notes themselves .
Outcome: The proposed project emphasizes the information presented in the clinical notes and its linguistic structure.
Rad-SpatialNet: A Frame-based Resource for Fine-Grained Spatial Relations in Radiology Reports (2020.lrec-1)

Copied to clipboard

Challenge: Existing framework for encoding spatial language in radiology is based on frame semantics .
Approach: They propose a framework for encoding spatial language in radiology based on frame semantics and a corpus of 400 radiology reports annotated with spatial trigger expressions and contextual information.
Outcome: The proposed framework is based on the existing SpatialNet representation in the general domain and is able to generate more accurate representations of spatial language in radiology.
RadQA: A Question Answering Dataset to Improve Comprehension of Radiology Reports (2022.lrec-1)

Copied to clipboard

Challenge: Question answering (QA) is an intuitive means to query text data.
Approach: They propose a radiology question-answer-evidence-pair dataset with 3074 questions posed against radiology reports and annotated with their corresponding answer spans by physicians.
Outcome: The proposed dataset has 3074 questions posed against radiology reports and annotated with their corresponding answer spans by physicians.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations