-
Tytuł:
-
Enhancing timeliness of drug overdose mortality surveillance: A machine learning approach.
-
Autorzy:
-
Ward PJ; Kentucky Injury Prevention and Research Center, College of Public Health, University of Kentucky, Lexington, Kentucky, United States of America.; Department of Epidemiology, College of Public Health, University of Kentucky, Lexington, Kentucky, United States of America.
Rock PJ; Kentucky Injury Prevention and Research Center, College of Public Health, University of Kentucky, Lexington, Kentucky, United States of America.; Department of Epidemiology, College of Public Health, University of Kentucky, Lexington, Kentucky, United States of America.
Slavova S; Kentucky Injury Prevention and Research Center, College of Public Health, University of Kentucky, Lexington, Kentucky, United States of America.; Department of Biostatistics, College of Public Health, University of Kentucky, Lexington, Kentucky, United States of America.
Young AM; Department of Epidemiology, College of Public Health, University of Kentucky, Lexington, Kentucky, United States of America.; Center on Drug and Alcohol Research, College of Medicine, University of Kentucky, Lexington, Kentucky, United States of America.
Bunn TL; Kentucky Injury Prevention and Research Center, College of Public Health, University of Kentucky, Lexington, Kentucky, United States of America.; Department of Preventive Medicine and Environmental Health, College of Public Health, University of Kentucky, Lexington, Kentucky, United States of America.
Kavuluru R; Department of Computer Science, College of Engineering, University of Kentucky, Lexington, Kentucky, United States of America.; Division of Biomedical Informatics, Department of Internal Medicine, College of Medicine, University of Kentucky, Lexington, Kentucky, United States of America.
-
Źródło:
-
PloS one [PLoS One] 2019 Oct 16; Vol. 14 (10), pp. e0223318. Date of Electronic Publication: 2019 Oct 16 (Print Publication: 2019).
-
Typ publikacji:
-
Journal Article; Research Support, N.I.H., Extramural; Research Support, U.S. Gov't, P.H.S.
-
Język:
-
English
-
Imprint Name(s):
-
Original Publication: San Francisco, CA : Public Library of Science
-
MeSH Terms:
-
Drug Overdose/*mortality
Cause of Death ; Drug Overdose/diagnosis ; Drug Overdose/epidemiology ; Humans ; International Classification of Diseases ; Kentucky/epidemiology ; Machine Learning ; Public Health Surveillance
-
References:
-
MMWR Morb Mortal Wkly Rep. 2017 Nov 03;66(43):1197-1202. (PMID: 29095804)
Am J Public Health. 2018 Dec;108(12):1587-1589. (PMID: 30403503)
Front Pharmacol. 2018 Oct 26;9:1210. (PMID: 30416445)
Int J Med Inform. 2015 Nov;84(11):956-65. (PMID: 26323193)
Natl Vital Stat Rep. 2016 Dec;65(9):1-15. (PMID: 27996933)
J Biomed Inform. 2017 Nov;75S:S85-S93. (PMID: 28506904)
BMC Med Inform Decis Mak. 2015 Jul 15;15:53. (PMID: 26174442)
Acad Forensic Pathol. 2017 Mar;7(1):60-72. (PMID: 31239957)
AMIA Annu Symp Proc. 2010 Nov 13;2010:677-81. (PMID: 21347064)
Australas Med J. 2013 May 30;6(5):292-9. (PMID: 23745151)
MMWR Morb Mortal Wkly Rep. 1989 Jan 13;38(1):1-4. (PMID: 2491906)
AMIA Jt Summits Transl Sci Proc. 2013 Mar 18;2013:112-6. (PMID: 24303247)
J Am Med Inform Assoc. 2016 Nov;23(6):1077-1084. (PMID: 27026618)
J Biomed Inform. 2009 Oct;42(5):760-72. (PMID: 19683066)
J Biomed Inform. 2018 Apr;80:64-77. (PMID: 29496630)
Proc Conf Empir Methods Nat Lang Process. 2018 Oct-Nov;2018:3132-3142. (PMID: 30775726)
Curr Pharm Biotechnol. 2018;19(2):113-123. (PMID: 29621965)
NCHS Data Brief. 2018 Nov;(329):1-8. (PMID: 30500323)
J Forensic Sci. 2018 Jul;63(4):1138-1145. (PMID: 29143322)
-
Grant Information:
-
U17 CE002732 United States CE NCIPC CDC HHS; UG3 DA044798 United States DA NIDA NIH HHS; UH3 DA044798 United States DA NIDA NIH HHS
-
Entry Date(s):
-
Date Created: 20191017 Date Completed: 20200316 Latest Revision: 20200529
-
Update Code:
-
20240105
-
PubMed Central ID:
-
PMC6795484
-
DOI:
-
10.1371/journal.pone.0223318
-
PMID:
-
31618226
-
Background: Timely data is key to effective public health responses to epidemics. Drug overdose deaths are identified in surveillance systems through ICD-10 codes present on death certificates. ICD-10 coding takes time, but free-text information is available on death certificates prior to ICD-10 coding. The objective of this study was to develop a machine learning method to classify free-text death certificates as drug overdoses to provide faster drug overdose mortality surveillance.
Methods: Using 2017-2018 Kentucky death certificate data, free-text fields were tokenized and features were created from these tokens using natural language processing (NLP). Word, bigram, and trigram features were created as well as features indicating the part-of-speech of each word. These features were then used to train machine learning classifiers on 2017 data. The resulting models were tested on 2018 Kentucky data and compared to a simple rule-based classification approach. Documented code for this method is available for reuse and extensions: https://github.com/pjward5656/dcnlp.
Results: The top scoring machine learning model achieved 0.96 positive predictive value (PPV) and 0.98 sensitivity for an F-score of 0.97 in identification of fatal drug overdoses on test data. This machine learning model achieved significantly higher performance for sensitivity (p<0.001) than the rule-based approach. Additional feature engineering may improve the model's prediction. This model can be deployed on death certificates as soon as the free-text is available, eliminating the time needed to code the death certificates.
Conclusion: Machine learning using natural language processing is a relatively new approach in the context of surveillance of health conditions. This method presents an accessible application of machine learning that improves the timeliness of drug overdose mortality surveillance. As such, it can be employed to inform public health responses to the drug overdose epidemic in near-real time as opposed to several weeks following events.
Competing Interests: The authors have declared that no competing interests exist.
Zaloguj się, aby uzyskać dostęp do pełnego tekstu.
