Background: Management of abdominal aortic aneurysms (AAAs) requires serial imaging surveillance to evaluate the aneurysm dimension. Natural language processing (NLP) has been previously developed to retrospectively identify patients with AAA from electronic health records (EHRs). However, there are no reported studies that use NLP to identify patients with AAA in near real-time from radiology reports.
Objective: This study aims to develop and validate a rule-based NLP algorithm for near real-time automatic extraction of AAA diagnosis from radiology reports for case identification.
Methods: The AAA-NLP algorithm was developed and deployed to an EHR big data infrastructure for near real-time processing of radiology reports from May 1, 2019, to September 2020. NLP extracted named entities for AAA case identification and classified subjects as cases and controls. The reference standard to assess algorithm performance was a manual review of processed radiology reports by trained physicians following standardized criteria. Reviewers were blinded to the diagnosis of each subject. The AAA-NLP algorithm was refined in 3 successive iterations. For each iteration, the AAA-NLP algorithm was modified based on performance compared to the reference standard.
Results: A total of 360 reports were reviewed, of which 120 radiology reports were randomly selected for each iteration. At each iteration, the AAA-NLP algorithm performance improved. The algorithm identified AAA cases in near real-time with high positive predictive value (0.98), sensitivity (0.95), specificity (0.98), F1 score (0.97), and accuracy (0.97).
Conclusions: Implementation of NLP for accurate identification of AAA cases from radiology reports with high performance in near real time is feasible. This NLP technique will support automated input for patient care and clinical decision support tools for the management of patients with AAA.
Worldwide prevalence rates of abdominal aortic aneurysms (AAAs) range from 1.6% to 3.3% for men older than 60 years . Assessment of AAA may be performed by a variety of imaging tests, including ultrasound (US), computerized tomography (CT), and magnetic resonance imaging (MRI). In the United States, the prevalence of AAA has been reported as 2.8% among 9457 individuals screened by US [ ]. Moreover, screening for early identification decreases the risk of aneurysm-related death and morbidity [ , ]. A prior study has shown that 4.5 ruptured AAA per 10,000 person-years were likely to have been prevented by screening, with an estimated 54 life-years gained per year of screening in a population of 23,000 men at risk [ ].
The interpretation of imaging examinations is routinely reported in radiology reports as narrative text in electronic health records (EHRs) . The automated extraction of information from narrative text can be accomplished by natural language processing (NLP) [ - ]. Prior studies have demonstrated high accuracy, sensitivity, specificity, and positive predictive value (PPV) of NLP for extraction of clinical concepts from narrative text in radiology reports [ - ]. Moreover, NLP is useful in cohort ascertainment for epidemiologic studies, query-based case retrieval, clinical decision support (CDS), quality assessment of radiologic practices, and diagnostic surveillance [ ].
A previous retrospective cohort study from our institution developed a rule-based NLP algorithm for retrospective retrieval of AAA cases from radiology reports, which performed with high accuracy . However, to the best of our knowledge, no prior study has demonstrated the use of NLP to identify AAA cases from radiology reports processed in near real-time. Hence, we tested the hypothesis that a rule-based NLP algorithm will extract AAA diagnosis from radiology reports in near real-time with high accuracy.
This study used Mayo Clinic radiology reports from May 1, 2019, to September 30, 2020.
A rule-based AAA-NLP algorithm was developed for information extraction of AAA diagnosis automatically from radiology reports, including CT abdomen pelvis without intravenous (IV) contrast, CT chest abdomen pelvis angiogram with IV contrast, US abdomen complete, US aorta iliac arteries bilateral with doppler, MRI abdomen with and without IV contrast, and MRI pelvis with and without IV contrast. The rule-based NLP algorithm was developed using MedTagger and deployed in the institutional near real-time big data infrastructure to process relevant radiology reports. MedTagger is an open-source NLP tool that has been previously used in various clinical NLP applications . MedTagger enables section identification, extraction of concepts, sentences, and word tokenization [ , ]. The AAA-NLP algorithm had 2 main components composed of text processing and report classification. AAA-relevant concepts were used to classify all reports ( ).
A custom lexicon for AAA was identified by the study team through a manual review of radiology reports. Subsequently, this lexicon was mapped to corresponding concepts and their synonyms in the Unified Medical Language System Metathesaurus. The lexicon used for AAA identification included aorta abdominal aneurysm, aortic aneurysm abdominal, AAA, aneurysm abdominal aorta, and infrarenal aortic aneurysm. Each radiology report was then processed in near real-time by NLP. The AAA-NLP algorithm extracted both the lexicon and the contextual information of assertions, including negations or confirmations, from each radiology report.displays the rules used by the NLP algorithm. The AAA-NLP algorithm classified subjects as AAA cases and controls without AAA.
To enable validation, the NLP output generated by near real-time processing of radiology reports was retrieved from the digital infrastructure by the information technology team and converted to a human-readable format for annotation. This annotation was performed by 2 trained physicians following written guidelines for standardization. The annotators were blinded to the diagnosis of each subject and to the results of the other annotator. In the written guidelines, AAA was defined as an aortic aneurysm diameter ≥3 cm by imaging as recommended by clinical practice guidelines .
The annotators reviewed the output from 120 processed radiology reports in 3 different training sets for iterative validation cycles to refine the algorithm. A total of 360 reports were reviewed. After abstracting and classifying the radiology reports, the information was entered and stored in a digital data set. Reports with a diagnosis of AAA were categorized as “case”; if there was no evidence of AAA or if an alternate diagnosis other than AAA was reported, the report was categorized as “control.” A board-certified cardiologist verified the information and resolved discrepancies in patient classification.
Abdominal aortic aneurysm (AAA)–natural language processing rule and examples of text span.
Rule (any token + keyword for AAA + any token)
Examples of confirmatory assertions
- Suprarenal aortic abdominal aneurysm which measures up to 5.2 cm
- Fusiform infrarenal abdominal aortic aneurysm terminating proximal to the aortobiiliac bifurcation, 56 mm, previously 56 mm
- There is a 5.7×5.1 cm infrarenal aortic aneurysm measured on image 175 of series 4
Examples of negated assertions
- Negative for abdominal aortic aneurysm or dissection
- Abdominal aortic aneurysm is absent
- Negative for thoracic or abdominal aortic aneurysm, dissection, penetrating atherosclerotic ulcer or intramural hematoma
The information extracted by the AAA-NLP algorithm from radiology reports in near real-time was compared to the reference standard manual review of radiology reports following written guidelines for standardization to calculate PPV, sensitivity, specificity, and F1 score. The formula to calculate F1 score was given as follows: 2 × ((PPV×sensitivity) / (PPV+sensitivity)) .
This project was approved by the Mayo Clinic Institutional Review Board (approval number 21-006950).
Reports of 295 patients were validated in 3 different iterations. The data set for each iteration contained 120 reports, but 46 (16%) patients had more than one report. The reasons for more than one report for the same patient were imaging tests performed before and after repair procedures or surveillance for serial assessment of AAA (). There were no discrepancies regarding AAA diagnosis between 2 or more imaging reports from the same patient. shows the distribution of demographic characteristics across AAA cases and controls. Cases and controls had similar ages in each of the iterative validation cycles, and most patients were Caucasian. AAA cases were more likely to have a history of smoking.
For evaluation of the AAA-NLP algorithm performance, 120 processed reports from each iteration were randomly selected. A total of 360 processed reports were reviewed by 2 physicians blinded to AAA diagnosis. There was 100% agreement for interactions 1 and 3. For interaction 2, the annotators disagreed on 1 report yielding a kappa coefficient of 92%. The disagreement was resolved by a board-certified cardiologist, creating the reference standard for comparison. The number of reports classified by the reference standard as true positives, false positives, true negatives, and false negatives in each iteration is shown in.
|Characteristic||Iteration 1||Iteration 2||Iteration 3|
|Case (n=31)||Control (n=52)||Case (n=44)||Control (n=59)||Case (n=59)||Control (n=50)|
|Age (years), mean (SD)||78.6 (11.1)||74.4 (12.4)||70.3 (8.4)||69.5 (14.1)||81.2 (8.5)||72.8 (10.4)|
|Male sex, n (%)||26 (84)||21 (40)||34 (77)||33 (56)||46 (78)||25 (50)|
|Caucasian, n (%)||31 (100)||52 (100)||42 (95)||54 (92)||58 (98)||48 (96)|
|Comorbidities, n (%)|
|Hypertension||24 (77)||39 (75)||31 (70)||27 (46)||46 (78)||37 (74)|
|Hyperlipidemia||21 (68)||22 (42)||29 (66)||29 (49)||42 (71)||27 (54)|
|Smoking history||29 (94)||24 (46)||35 (80)||23 (39)||50 (85)||25 (50)|
|DMa||9 (29)||7 (13)||10 (23)||12 (20)||19 (32)||13 (26)|
|PADb||4 (13)||4 (8)||5 (11)||4 (7)||9 (15)||4 (8)|
|CADc||16 (52)||7 (13)||18 (41)||10 (17)||32 (54)||15 (30)|
|Patients with ≥2 reports, n||18||7||13||1||3||4|
|AAAd diameter (cm), mean (SD)||4.6 (1.08)||N/Ae||4.8 (1.3)||N/A||4.9 (1.2)||N/A|
|Reports after AAA repair, n||2||N/A||9||N/A||8||N/A|
aDM: diabetes mellitus.
bPAD: peripheral artery disease.
cCAD: coronary artery disease.
dAAA: abdominal aortic aneurysm.
eN/A: not applicable.
|Iteration 1||Iteration 2||Iteration 3|
|Predicted case||Predicted |
|Total||Predicted case||Predicted |
|Total||Predicted case||Predicted |
|Actual case||TPa 59||FNb 6||65||TP 56||FN 2||58||TP 59||FN 3||62|
|Actual control||FPc 1||TNd 54||55||FP 4||TN 58||62||FP 1||TN 57||58|
aTP: true positive.
bFN: false negative.
cFP: false positive.
dTN: true negative.
Radiology reports are composed of multiple sections.shows an example of a deidentified radiology report with all sections.
During the first iteration implementation, section ID number was used and section detection was challenging. For the second iteration, the algorithm was revised to include section header names for the filter criteria and solve sentence boundary issues. For the third iteration, section detection was implemented based on section names from our complete corpus using the frequency of normalized text with the tool lexical variant generation of the National Library of Medicine . In a separate experiment, 203 additional radiology reports were reviewed by the annotators for evaluation of report section extraction, which resulted in accuracy of 0.96.
During this iterative refinement process, the report sections termed “reason for exam,” “referral diagnosis,” “exam type,” and “signed by” () were excluded, resulting in enhanced NLP algorithm performance. The report sections selected for processing were findings and impressions. During each iteration, the algorithm performance further improved. The performance metrics of the iterations are summarized in .
During the last iteration, 3 false negatives and 1 false positive contributed to the error analysis. False negatives were due to the complex nature of narrative text in these reports (ie, no significant interval changes in appearances of a partially thrombosed infrarenal AAA measuring 42×40 mm, extending to the level of aortic bifurcation and proximal common iliac arteries; no signs of rupture or impending rupture of the known infrarenal AAA; and no slightly increased size of fusiform infrarenal AAA). Additionally, the false positive was due to a typographical error, which was the report of a patient with an aorta diameter of 2.7 cm labeled as AAA, which does not meet the criteria for AAA (≥3.0 cm).
|Performance metric||Iteration 1 (n=120)||Iteration 2 (n=120)||Iteration 3 (n=120)|
aPPV: positive predictive value.
In this study, a novel rule-based NLP algorithm was developed for the extraction of AAA diagnosis from radiology reports and prospectively deployed in the institutional big data infrastructure for near real-time processing. Compared to the reference standard of manual review of radiology reports, the AAA-NLP algorithm extracted AAA diagnosis in near real time with high sensitivity, PPV, F1 score, specificity, and accuracy.
To the best of our knowledge, this study is the first to describe the use of NLP algorithms prospectively to extract AAA diagnosis in near real time from radiology reports. Clinicians, information technologists, and informaticians collaborated to refine the algorithm to improve performance. In previous studies, billing codes were used to find AAA cases [, ]. However, in those studies, the cohorts were limited to patients with AAA who underwent procedures for aneurysm repair or had a history of ruptured AAA [ , ]. No prior studies using billing codes algorithms retrieved a broader spectrum of AAA diagnosis while also including patients presenting with uncomplicated AAA (ie, patients who did not undergo prior repair or who had not previously presented with ruptured AAA). In contrast, in this study, NLP automatically extracted AAA diagnosis from radiology reports prospectively and regardless of prior repair or rupture, thereby expanding the scope of computational approaches to include the detection of AAA cases prior to rupture or repair.
A radiology report consists of free text, organized into standard sections . The American College of Radiology has published guidelines with recommendations for the use of sections for narrative (free text) entry in radiology reports [ ]. NLP techniques enable the automatic extraction of information from narrative text [ - ]. Moreover, information extracted by NLP can be used to populate CDS systems automatically without the need for manual data entry and be better aligned with existing workflows such that radiologists can spend time interpreting images rather than filling out forms.
NLP is a computational methodology used for electronic phenotyping to extract meaningful clinical information from text fields [, , ]. In this study, we used NLP to process radiology text reports. The previous NLP algorithm used to find cases of AAA from radiology reports [ ] was designed for retrospective cohort identification, whereas this report describes the prospective implementation of an NLP algorithm for input to a patient-specific CDS system for near real-time processing of radiology reports. Near real-time processing requires <3 milliseconds to process a document after a radiologist releases a report to the EHR [ ]. The AAA-NLP implementation described in this study was developed within the existing digital infrastructure and can be used in clinical practice immediately without the need to retrain the algorithm. Additionally, the previously described algorithm [ ] did not identify document sections in the radiology reports. By selecting specific sections for NLP information extraction, improvement in NLP performance was observed, as shown in the Results section. In the future, transformer-based NLP models [ , ] may be trained to interpret nuanced language, and ablation experiments [ ] could be used to further evaluate these models.
The use of NLP algorithms has advantages compared to other methods. In comparison, the use of check box forms in radiology reports may require the development of new workflows [, ]. The use of check box forms also requires the radiologist to direct attention away from the imaging interpretation process [ , ]. Manual entry of summaries of radiology findings in a check box can increase reporting time with decreased radiologist productivity [ , ]. Check box use could also result in the loss of important and clinically relevant descriptive information available only in the radiology narrative reports.
The rule-based AAA-NLP algorithm described in this study shows accurate detection of a broad spectrum of AAA cases prospectively in near real time from radiology reports, regardless of the presence of prior rupture or repair. This methodology will also potentially generate input for CDS to assist providers in managing patients with AAA by displaying the relevant information automatically at the point of care and in near real time for CDS tools. It will also support the automatic identification of cohorts for research purposes (eg, cohorts for clinical trials) and quality projects, and will support a learning health care system. NLP has been previously used for the identification of peripheral arterial disease and critical limb ischemia from narrative clinical notes of EHRs [, ]. Therefore, it will also be possible to develop NLP algorithms for the identification of AAA cases from clinical notes in near real time.
In efforts to develop a learning health care system, Mayo Clinic has developed a robust big data–empowered clinical NLP infrastructure that enables near real-time NLP processing for the delivery of relevant information to the point of care via CDS . Accordingly, we have deployed the AAA-NLP algorithm described herein to this digital infrastructure for translation to clinical practice. Importantly, the near real-time identification of patients with AAA by NLP responds to the American Heart Association scientific statement, which recommends the implementation of technologies to extract clinical information in real time that will promptly provide synopses of the information extracted [ ].
This NLP algorithm was developed, tested, and implemented in a single tertiary medical center. Future studies should evaluate this algorithm at other institutions to demonstrate portability. A robust institutional digital infrastructure is required for the execution of near real-time processing of radiology reports . Hence, the absence of adequate digital infrastructure may limit porting of this algorithm. For implementation, the analysis of radiology report architecture to enable the selection of document types and document sections may also be necessary for portability. Another potential challenge for porting this algorithm to other EHRs is differences in lexicons used for the extraction of the AAA concept across institutions. In mitigation, for this NLP algorithm, each lexicon was mapped to corresponding concepts and synonyms in the publicly available Unified Medical Language System Metathesaurus for standardization.
The algorithm was developed for the extraction of AAA diagnosis but not for the extraction of iliac artery or thoracic aortic aneurysms. Future studies should create and validate NLP algorithms for the extraction of thoracic and iliac artery aneurysms. The clinical criteria for AAA diagnosis involve a minimum diameter, but this NLP algorithm did not interpret the reported diameter. This is an area for future improvement in the algorithm, as clinical criteria for AAA may change over time. In this study, most patients were Caucasian. This was likely related to the ethnic distribution of communities in the Midwest, where this study was conducted [, ]. Additionally, prior studies have reported a higher prevalence of AAA among Caucasians compared to other races [ , ]. There were differences in comorbidities of patients included in the 3 iterations. However, the NLP was developed for the extraction of the diagnosis of AAA and not developed for the extraction of associated patient comorbidities. The differences in patient comorbidities did not influence NLP performance for the extraction of AAA from radiology reports.
Implementation of NLP for prospective identification of AAA cases from radiology reports in near real time with high performance is feasible. This near real-time NLP technique described will potentially be helpful for the generation of automated input for CDS tools to assist clinicians in the management of patients with AAA, quality improvement projects, and research (automated identification of cohorts).
The authors would like to thank Kara M Firzlaff for secretarial support and Christopher G Scott for statistical analysis. This study was funded by the Mayo Clinic K2R award.
The data sets generated and analyzed during this study are not publicly available because participants in this study did not agree for their data to be shared publicly but are available from the corresponding author on reasonable request.
Conflicts of Interest
- US Preventive Services Task Force, Owens DK, Davidson KW, Krist AH, Barry MJ, Cabana M, et al. Screening for abdominal aortic aneurysm: US preventive services task force recommendation statement. JAMA 2019;322(22):2211-2218. [CrossRef] [Medline]
- Summers KL, Kerut EK, Sheahan CM, Sheahan MG. Evaluating the prevalence of abdominal aortic aneurysms in the United States through a national screening database. J Vasc Surg 2021;73(1):61-68. [CrossRef] [Medline]
- Chaikof EL, Dalman RL, Eskandari MK, Jackson BM, Lee WA, Mansour MA, et al. The society for vascular surgery practice guidelines on the care of patients with an abdominal aortic aneurysm. J Vasc Surg 2018;67(1):2-77.e2 [FREE Full text] [CrossRef] [Medline]
- Wilmink ABM, Quick CRG, Hubbard CS, Day NE. Effectiveness and cost of screening for abdominal aortic aneurysm: results of a population screening program. J Vasc Surg 2003;38(1):72-77 [FREE Full text] [CrossRef] [Medline]
- Pons E, Braun LMM, Hunink MGM, Kors JA. Natural language processing in radiology: a systematic review. Radiology 2016;279(2):329-343. [CrossRef] [Medline]
- Jensen PB, Jensen LJ, Brunak S. Mining electronic health records: towards better research applications and clinical care. Nat Rev Genet 2012;13(6):395-405. [CrossRef] [Medline]
- Demner-Fushman D, Chapman WW, McDonald CJ. What can natural language processing do for clinical decision support? J Biomed Inform 2009;42(5):760-772 [FREE Full text] [CrossRef] [Medline]
- Lopez-Jimenez F, Attia Z, Arruda-Olson AM, Carter R, Chareonthaitawee P, Jouni H, et al. Artificial intelligence in cardiology: present and future. Mayo Clin Proc 2020;95(5):1015-1039. [CrossRef] [Medline]
- Wang Y, Mehrabi S, Sohn S, Atkinson EJ, Amin S, Liu H. Natural language processing of radiology reports for identification of skeletal site-specific fractures. BMC Med Inform Decis Mak 2019;19(suppl 3):73 [FREE Full text] [CrossRef] [Medline]
- Fiszman M, Chapman WW, Aronsky D, Evans RS, Haug PJ. Automatic detection of acute bacterial pneumonia from chest X-ray reports. J Am Med Inform Assoc 2000;7(6):593-604 [FREE Full text] [CrossRef] [Medline]
- Solti I, Cooke CR, Xia F, Wurfel MM. Automated classification of radiology reports for acute lung injury: comparison of keyword and machine learning based natural language processing approaches. Proceedings (IEEE Int Conf Bioinformatics Biomed) 2009;2009:314-319 [FREE Full text] [CrossRef] [Medline]
- Sohn S, Ye Z, Liu H, Chute CG, Kullo IJ. Identifying abdominal aortic aneurysm cases and controls using natural language processing of radiology reports. AMIA Jt Summits Transl Sci Proc 2013;2013:249-253 [FREE Full text] [Medline]
- Wang Y, Wang L, Rastegar-Mojarad M, Moon S, Shen F, Afzal N, et al. Clinical information extraction applications: a literature review. J Biomed Inform 2018;77:34-49 [FREE Full text] [CrossRef] [Medline]
- Liu H, Bielinski SJ, Sohn S, Murphy S, Wagholikar KB, Jonnalagadda SR, et al. An information extraction framework for cohort identification using electronic health records. AMIA Jt Summits Transl Sci Proc 2013;2013:149-153 [FREE Full text] [Medline]
- Wagholikar K, Torii M, Jonnalagadda S, Liu H. Feasibility of pooling annotated corpora for clinical concept extraction. AMIA Jt Summits Transl Sci Proc 2012;2012:38 [FREE Full text] [Medline]
- Hirsch AT, Haskal ZJ, Hertzer NR, Bakal CW, Creager MA, Halperin JL, American Association for Vascular Surgery/Society for Vascular Surgery, Society for Cardiovascular Angiography and Interventions, Society for Vascular Medicine and Biology, Society of Interventional Radiology, ACC/AHA Task Force on Practice Guidelines. ACC/AHA guidelines for the management of patients with peripheral arterial disease (lower extremity, renal, mesenteric, and abdominal aortic): a collaborative report from the American Associations for Vascular Surgery/Society for Vascular Surgery, Society for Cardiovascular Angiography and Interventions, Society for Vascular Medicine and Biology, Society of Interventional Radiology, and the ACC/AHA Task Force on practice guidelines (writing committee to develop guidelines for the management of patients with peripheral arterial disease)--summary of recommendations. J Vasc Interv Radiol 2006;17(9):1383-1397. [CrossRef] [Medline]
- Lexical Tools: Lvg (Lexical Variants Generation). National Library of Medicine. 2022. URL: https://tinyurl.com/2ej3v4p [accessed 2023-01-24]
- Borthwick KM, Smelser DT, Bock JA, Elmore JR, Ryer EJ, Ye Z, et al. ePhenotyping for abdominal aortic aneurysm in the Electronic Medical Records and Genomics (eMERGE) network: algorithm development and konstanz information miner workflow. Int J Biomed Data Min 2015;4(1):113 [FREE Full text] [Medline]
- Jetty P, van Walraven C. Coding accuracy of abdominal aortic aneurysm repair procedures in administrative databases: a note of caution. J Eval Clin Pract 2011;17(1):91-96. [CrossRef] [Medline]
- ACR practice parameter for communication of diagnostic imaging findings. American College of Radiology. 2020. URL: https://www.acr.org/-/media/acr/files/practice-parameters/communicationdiag.pdf [accessed 2023-01-24]
- Afzal N, Mallipeddi VP, Sohn S, Liu H, Chaudhry R, Scott CG, et al. Natural language processing of clinical notes for identification of critical limb ischemia. Int J Med Inform 2018;111:83-89 [FREE Full text] [CrossRef] [Medline]
- Kaggal VC, Elayavilli RK, Mehrabi S, Pankratz JJ, Sohn S, Wang Y, et al. Toward a learning health-care system: knowledge delivery at the point of care empowered by big data and NLP. Biomed Inform Insights 2016;8(suppl 1):13-22 [FREE Full text] [CrossRef] [Medline]
- Gillioz A, Casas J, Mugellini E, Khaled OA. Overview of the transformer-based models for NLP tasks. 2020 Presented at: 2020 15th Conference on Computer Science and Information Systems (FedCSIS); September 6-9, 2020; Sofia, Bulgaria p. 179-183. [CrossRef]
- Gillioz A, Casas J, Mugellini E, Khaled OA. Overview of the transformer-based models for NLP tasks. 2020 Presented at: 2020 15th Conference on Computer Science and Information Systems (FedCSIS); September 6-9, 2020; Sofia, Bulgaria p. 179-183. [CrossRef]
- Zhang D, Yin C, Zeng J, Yuan X, Zhang P. Combining structured and unstructured data for predictive models: a deep learning approach. BMC Med Inform Decis Mak 2020;20(1):280. [CrossRef]
- Weiss DL, Langlotz CP. Structured reporting: patient care enhancement or productivity nightmare? Radiology 2008;249(3):739-747. [CrossRef] [Medline]
- Mityul MI, Gilcrease-Garcia B, Mangano MD, Demertzis JL, Gunn AJ. Radiology reporting: current practices and an introduction to patient-centered opportunities for improvement. AJR Am J Roentgenol 2018;210(2):376-385. [CrossRef] [Medline]
- Afzal N, Sohn S, Abram S, Scott CG, Chaudhry R, Liu H, et al. Mining peripheral arterial disease cases from narrative clinical notes using natural language processing. J Vasc Surg 2017;65(6):1753-1761 [FREE Full text] [CrossRef] [Medline]
- Maddox TM, Albert NM, Borden WB, Curtis LH, Ferguson TB, Kao DP, American Heart Association Council on Quality of Care and Outcomes Research, Council on Cardiovascular Disease in the Young, Council on Clinical Cardiology, Council on Functional Genomics and Translational Biology, Stroke Council. The learning healthcare system and cardiovascular care: a scientific statement from the American Heart Association. Circulation 2017;135(14):e826-e857 [FREE Full text] [CrossRef] [Medline]
- St Sauver JL, Grossardt BR, Leibson CL, Yawn BP, Melton LJ, Rocca WA. Generalizability of epidemiological findings and public health decisions: an illustration from the Rochester Epidemiology Project. Mayo Clin Proc 2012;87(2):151-160 [FREE Full text] [CrossRef] [Medline]
- Johnson G, Avery A, McDougal EG, Burnham SJ, Keagy BA. Aneurysms of the abdominal aorta. Incidence in blacks and whites in North Carolina. Arch Surg 1985;120(10):1138-1140. [CrossRef] [Medline]
- Kent KC, Zwolak RM, Egorova NN, Riles TS, Manganaro A, Moskowitz AJ, et al. Analysis of risk factors for abdominal aortic aneurysm in a cohort of more than 3 million individuals. J Vasc Surg 2010;52(3):539-548 [FREE Full text] [CrossRef] [Medline]
|AAA: abdominal aortic aneurysm|
|CDS: clinical decision support|
|CT: computerized tomography|
|EHR: electronic health record|
|MRI: magnetic resonance imaging|
|NLP: natural language processing|
|PPV: positive predictive value|
Edited by C Lovis; submitted 11.07.22; peer-reviewed by N Afzal, C Reeder; comments to author 16.08.22; revised version received 29.12.22; accepted 19.01.23; published 24.02.23Copyright
©Simon Gaviria-Valencia, Sean P Murphy, Vinod C Kaggal, Robert D McBane II, Thom W Rooke, Rajeev Chaudhry, Mateo Alzate-Aguirre, Adelaide M Arruda-Olson. Originally published in JMIR Medical Informatics (https://medinform.jmir.org), 24.02.2023.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Medical Informatics, is properly cited. The complete bibliographic information, a link to the original publication on https://medinform.jmir.org/, as well as this copyright and license information must be included.