Detecting Ground Glass Opacity Features in Patients With Lung Cancer: Automated Extraction and Longitudinal Analysis via Deep Learning–Based Natural Language Processing
BackgroundGround-glass opacities (GGOs) appearing in computed tomography (CT) scans may indicate potential lung malignancy. Proper management of GGOs based on their features can prevent the development of lung cancer. Electronic health records are rich sources of information...
Main Authors: | , , , , , , , , , , , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
JMIR Publications
2023-06-01
|
Series: | JMIR AI |
Online Access: | https://ai.jmir.org/2023/1/e44537 |
_version_ | 1797734123643600896 |
---|---|
author | Kyeryoung Lee Zongzhi Liu Urmila Chandran Iftekhar Kalsekar Balaji Laxmanan Mitchell K Higashi Tomi Jun Meng Ma Minghao Li Yun Mai Christopher Gilman Tongyu Wang Lei Ai Parag Aggarwal Qi Pan William Oh Gustavo Stolovitzky Eric Schadt Xiaoyan Wang |
author_facet | Kyeryoung Lee Zongzhi Liu Urmila Chandran Iftekhar Kalsekar Balaji Laxmanan Mitchell K Higashi Tomi Jun Meng Ma Minghao Li Yun Mai Christopher Gilman Tongyu Wang Lei Ai Parag Aggarwal Qi Pan William Oh Gustavo Stolovitzky Eric Schadt Xiaoyan Wang |
author_sort | Kyeryoung Lee |
collection | DOAJ |
description |
BackgroundGround-glass opacities (GGOs) appearing in computed tomography (CT) scans may indicate potential lung malignancy. Proper management of GGOs based on their features can prevent the development of lung cancer. Electronic health records are rich sources of information on GGO nodules and their granular features, but most of the valuable information is embedded in unstructured clinical notes.
ObjectiveWe aimed to develop, test, and validate a deep learning–based natural language processing (NLP) tool that automatically extracts GGO features to inform the longitudinal trajectory of GGO status from large-scale radiology notes.
MethodsWe developed a bidirectional long short-term memory with a conditional random field–based deep-learning NLP pipeline to extract GGO and granular features of GGO retrospectively from radiology notes of 13,216 lung cancer patients. We evaluated the pipeline with quality assessments and analyzed cohort characterization of the distribution of nodule features longitudinally to assess changes in size and solidity over time.
ResultsOur NLP pipeline built on the GGO ontology we developed achieved between 95% and 100% precision, 89% and 100% recall, and 92% and 100% F1-scores on different GGO features. We deployed this GGO NLP model to extract and structure comprehensive characteristics of GGOs from 29,496 radiology notes of 4521 lung cancer patients. Longitudinal analysis revealed that size increased in 16.8% (240/1424) of patients, decreased in 14.6% (208/1424), and remained unchanged in 68.5% (976/1424) in their last note compared to the first note. Among 1127 patients who had longitudinal radiology notes of GGO status, 815 (72.3%) were reported to have stable status, and 259 (23%) had increased/progressed status in the subsequent notes.
ConclusionsOur deep learning–based NLP pipeline can automatically extract granular GGO features at scale from electronic health records when this information is documented in radiology notes and help inform the natural history of GGO. This will open the way for a new paradigm in lung cancer prevention and early detection. |
first_indexed | 2024-03-12T12:38:45Z |
format | Article |
id | doaj.art-c8dc59e8dfdf409d87901daf54c1b1e0 |
institution | Directory Open Access Journal |
issn | 2817-1705 |
language | English |
last_indexed | 2024-03-12T12:38:45Z |
publishDate | 2023-06-01 |
publisher | JMIR Publications |
record_format | Article |
series | JMIR AI |
spelling | doaj.art-c8dc59e8dfdf409d87901daf54c1b1e02023-08-29T00:00:02ZengJMIR PublicationsJMIR AI2817-17052023-06-012e4453710.2196/44537Detecting Ground Glass Opacity Features in Patients With Lung Cancer: Automated Extraction and Longitudinal Analysis via Deep Learning–Based Natural Language ProcessingKyeryoung Leehttps://orcid.org/0000-0002-6937-9931Zongzhi Liuhttps://orcid.org/0000-0003-1578-9967Urmila Chandranhttps://orcid.org/0000-0001-7183-3622Iftekhar Kalsekarhttps://orcid.org/0000-0001-5232-5972Balaji Laxmananhttps://orcid.org/0000-0003-2862-3624Mitchell K Higashihttps://orcid.org/0000-0002-1047-0263Tomi Junhttps://orcid.org/0000-0002-2120-1704Meng Mahttps://orcid.org/0000-0002-0212-298XMinghao Lihttps://orcid.org/0000-0002-3516-3702Yun Maihttps://orcid.org/0000-0002-6011-8874Christopher Gilmanhttps://orcid.org/0009-0003-1753-2046Tongyu Wanghttps://orcid.org/0009-0008-6618-3088Lei Aihttps://orcid.org/0000-0002-2802-5774Parag Aggarwalhttps://orcid.org/0009-0009-0904-0205Qi Panhttps://orcid.org/0009-0009-9381-5826William Ohhttps://orcid.org/0000-0001-5113-8147Gustavo Stolovitzkyhttps://orcid.org/0000-0002-9618-2819Eric Schadthttps://orcid.org/0000-0002-7892-8808Xiaoyan Wanghttps://orcid.org/0000-0002-4193-4120 BackgroundGround-glass opacities (GGOs) appearing in computed tomography (CT) scans may indicate potential lung malignancy. Proper management of GGOs based on their features can prevent the development of lung cancer. Electronic health records are rich sources of information on GGO nodules and their granular features, but most of the valuable information is embedded in unstructured clinical notes. ObjectiveWe aimed to develop, test, and validate a deep learning–based natural language processing (NLP) tool that automatically extracts GGO features to inform the longitudinal trajectory of GGO status from large-scale radiology notes. MethodsWe developed a bidirectional long short-term memory with a conditional random field–based deep-learning NLP pipeline to extract GGO and granular features of GGO retrospectively from radiology notes of 13,216 lung cancer patients. We evaluated the pipeline with quality assessments and analyzed cohort characterization of the distribution of nodule features longitudinally to assess changes in size and solidity over time. ResultsOur NLP pipeline built on the GGO ontology we developed achieved between 95% and 100% precision, 89% and 100% recall, and 92% and 100% F1-scores on different GGO features. We deployed this GGO NLP model to extract and structure comprehensive characteristics of GGOs from 29,496 radiology notes of 4521 lung cancer patients. Longitudinal analysis revealed that size increased in 16.8% (240/1424) of patients, decreased in 14.6% (208/1424), and remained unchanged in 68.5% (976/1424) in their last note compared to the first note. Among 1127 patients who had longitudinal radiology notes of GGO status, 815 (72.3%) were reported to have stable status, and 259 (23%) had increased/progressed status in the subsequent notes. ConclusionsOur deep learning–based NLP pipeline can automatically extract granular GGO features at scale from electronic health records when this information is documented in radiology notes and help inform the natural history of GGO. This will open the way for a new paradigm in lung cancer prevention and early detection.https://ai.jmir.org/2023/1/e44537 |
spellingShingle | Kyeryoung Lee Zongzhi Liu Urmila Chandran Iftekhar Kalsekar Balaji Laxmanan Mitchell K Higashi Tomi Jun Meng Ma Minghao Li Yun Mai Christopher Gilman Tongyu Wang Lei Ai Parag Aggarwal Qi Pan William Oh Gustavo Stolovitzky Eric Schadt Xiaoyan Wang Detecting Ground Glass Opacity Features in Patients With Lung Cancer: Automated Extraction and Longitudinal Analysis via Deep Learning–Based Natural Language Processing JMIR AI |
title | Detecting Ground Glass Opacity Features in Patients With Lung Cancer: Automated Extraction and Longitudinal Analysis via Deep Learning–Based Natural Language Processing |
title_full | Detecting Ground Glass Opacity Features in Patients With Lung Cancer: Automated Extraction and Longitudinal Analysis via Deep Learning–Based Natural Language Processing |
title_fullStr | Detecting Ground Glass Opacity Features in Patients With Lung Cancer: Automated Extraction and Longitudinal Analysis via Deep Learning–Based Natural Language Processing |
title_full_unstemmed | Detecting Ground Glass Opacity Features in Patients With Lung Cancer: Automated Extraction and Longitudinal Analysis via Deep Learning–Based Natural Language Processing |
title_short | Detecting Ground Glass Opacity Features in Patients With Lung Cancer: Automated Extraction and Longitudinal Analysis via Deep Learning–Based Natural Language Processing |
title_sort | detecting ground glass opacity features in patients with lung cancer automated extraction and longitudinal analysis via deep learning based natural language processing |
url | https://ai.jmir.org/2023/1/e44537 |
work_keys_str_mv | AT kyeryounglee detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT zongzhiliu detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT urmilachandran detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT iftekharkalsekar detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT balajilaxmanan detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT mitchellkhigashi detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT tomijun detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT mengma detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT minghaoli detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT yunmai detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT christophergilman detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT tongyuwang detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT leiai detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT paragaggarwal detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT qipan detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT williamoh detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT gustavostolovitzky detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT ericschadt detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing AT xiaoyanwang detectinggroundglassopacityfeaturesinpatientswithlungcancerautomatedextractionandlongitudinalanalysisviadeeplearningbasednaturallanguageprocessing |