Development of a Natural Language Processing Engine to Generate Bladder Cancer Pathology Data for Health Services Research - 22/11/17
, Olga V. Patterson e, Patrick R. Alba e, Erik A. Pattison a, b, John D. Seigne b, c, Scott L. DuVall e, Douglas J. Robertson a, d, Brenda Sirovich a, d, Philip P. Goodney a, dAbstract |
Objective |
To take the first step toward assembling population-based cohorts of patients with bladder cancer with longitudinal pathology data, we developed and validated a natural language processing (NLP) engine that abstracts pathology data from full-text pathology reports.
Methods |
Using 600 bladder pathology reports randomly selected from the Department of Veterans Affairs, we developed and validated an NLP engine to abstract data on histology, invasion (presence vs absence and depth), grade, the presence of muscularis propria, and the presence of carcinoma in situ. Our gold standard was based on an independent review of reports by 2 urologists, followed by adjudication. We assessed the NLP performance by calculating the accuracy, the positive predictive value, and the sensitivity. We subsequently applied the NLP engine to pathology reports from 10,725 patients with bladder cancer.
Results |
When comparing the NLP output to the gold standard, NLP achieved the highest accuracy (0.98) for the presence vs the absence of carcinoma in situ. Accuracy for histology, invasion (presence vs absence), grade, and the presence of muscularis propria ranged from 0.83 to 0.96. The most challenging variable was depth of invasion (accuracy 0.68), with an acceptable positive predictive value for lamina propria (0.82) and for muscularis propria (0.87) invasion. The validated engine was capable of abstracting pathologic characteristics for 99% of the patients with bladder cancer.
Conclusion |
NLP had high accuracy for 5 of 6 variables and abstracted data for the vast majority of the patients. This now allows for the assembly of population-based cohorts with longitudinal pathology data.
Le texte complet de cet article est disponible en PDF.Plan
| Disclaimer: Opinions expressed in this manuscript are those of the authors and do not constitute official positions of the U.S. Federal Government or the Department of Veterans Affairs. |
|
| Financial Disclosure: Florian R. Schroeck was site principal investigator (without any compensation) for the Phase III Trial of Vicinium (Eleven Biotherapeutics) and received research funding from the U.S. Department of Veterans Affairs, Conquer Cancer Foundation, and American Cancer Society. John D. Seigne owned 100 shares of common stock of Johnson & Johnson. Scott L. DuVall received research grants from the following for-profit organizations: AbbVie Inc., Amgen Inc., Anolinx LLC, Astellas Pharma Inc., AstraZeneca Pharmaceuticals LP, Boehringer Ingelheim International GmbH, Eli Lilly and Company, F. Hoffmann-La Roche Ltd, Genentech Inc., Genomic Health, Inc., Gilead Sciences Inc., GlaxoSmithKline PLC, HITEKS Solutions Inc., Innocrin Pharmaceuticals Inc., Kantar Health, LexisNexis Risk Solutions, Merck & Co., Inc., Mylan Specialty LP, Myriad Genetics, Inc., Northrop Grumman Information Systems, Novartis International AG, PAREXEL International Corporation, and Shire PLC through the University of Utah or Western Institute for Biomedical Research. The remaining authors declare that they have no relevant financial interests. |
|
| Funding Support: Florian R. Schroeck is supported by the Department of Veterans Affairs, Veterans Health Administration, VISN1 Career Development Award; by a pilot grant from the American Cancer Society (IRG-82-003-30); by a Career Development Award from the Conquer Cancer Foundation; and by the Department of Surgery at the Dartmouth-Hitchcock Medical Center (Dow-Crichlow Award). Philip P. Goodney is supported by a grant from the Food and Drug Administration (U01FD005478-01, Sedrakyan = PI). This project was supported using resources and facilities at the White River Junction VA Medical Center, the VA Salt Lake City Health Care System, and the VA Informatics and Computing Infrastructure (VINCI), VA HSR RES 13-457. |
Vol 110
P. 84-91 - décembre 2017 Retour au numéro
Article précédent
- Cancer and All-cause Mortality in Bladder Cancer Patients Undergoing Radical Cystectomy: Development and Validation of a Nomogram for Treatment Decision-making
- Stephen B. Williams, Jinhai Huo, Yiyi Chu, Jacques G. Baillargeon, Timothy Daskivich, Yong-Fang Kuo, Christopher D. Kosarek, Simon P. Kim, Eduardo Orihuela, Douglas S. Tyler, Stephen J. Freedland, Ashish M. Kamat
- Editorial Comment
- Jack Zeineh, Michael J. Donovan
Bienvenue sur EM-consulte, la référence des professionnels de santé.
L’accès au texte intégral de cet article nécessite un abonnement.
Déjà abonné à cette revue ?