Predicting opioid dependence from electronic health records with machine learning

Randall J Ellis, Zichen Wang, Nicholas Genes, Avi Ma'ayan, Randall J Ellis, Zichen Wang, Nicholas Genes, Avi Ma'ayan

Abstract

Background: The opioid epidemic in the United States is averaging over 100 deaths per day due to overdose. The effectiveness of opioids as pain treatments, and the drug-seeking behavior of opioid addicts, leads physicians in the United States to issue over 200 million opioid prescriptions every year. To better understand the biomedical profile of opioid-dependent patients, we analyzed information from electronic health records (EHR) including lab tests, vital signs, medical procedures, prescriptions, and other data from millions of patients to predict opioid substance dependence.

Results: We trained a machine learning model to classify patients by likelihood of having a diagnosis of substance dependence using EHR data from patients diagnosed with substance dependence, along with control patients with no history of substance-related conditions, matched by age, gender, and status of HIV, hepatitis C, and sickle cell disease. The top machine learning classifier using all features achieved a mean area under the receiver operating characteristic (AUROC) curve of ~ 92%, and analysis of the model uncovered associations between basic clinical factors and substance dependence. Additionally, diagnoses, prescriptions, and procedures prior to the diagnoses of substance dependence were analyzed to elucidate the clinical profile of substance-dependent patients, relative to controls.

Conclusions: The predictive model may hold utility for identifying patients at risk of developing dependence, risk of overdose, and opioid-seeking patients that report other symptoms in their visits to the emergency room.

Keywords: Artificial intelligence; Electronic health records; Electronic medical records; Machine learning; Opioid dependence; Opioid epidemic.

Conflict of interest statement

This study has been granted exemption from human-subject research by the Program for the Protection of Human Subjects (PPHS) at the Institutional Review Boards (IRB), Mount Sinai Health System. The project number is HS#:18–00993.The authors declare that they have no competing interests.Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Figures

Fig. 1
Fig. 1
a Distribution of tests (labs and vitals) per case before filtering. b Distribution of tests (labs and vitals) per case after removing patients with less than 17 tests and tests with 90% or greater missing values. c Distribution of tests (labs and vitals) per control after removing tests with 90% or greater missing values
Fig. 2
Fig. 2
Distribution of ages for 11,573 cases given their first substance dependence diagnoses at 20 years of age or older
Fig. 3
Fig. 3
Histograms of (a) opioid prescriptions per patient, and (b) prescriptions per age in years
Fig. 4
Fig. 4
Gini importance values for the top 20 features by mean Gini importance
Fig. 5
Fig. 5
Receiver operating characteristic curves, normalized, and non-normalized confusion matrices for classifiers using no imputation (a, b), imputation by the mean (c, d), and imputation by the median (e, f), retaining only patients with more than 17 labs and vitals
Fig. 6
Fig. 6
Scatter plot of 483 diagnoses with statistically significant over- or underrepresentation in the cases compared to the controls measured using the Fisher Exact test. Each point is a diagnosis that was statistically significant. 31 diagnoses have odds ratios of less than 1

References

    1. Rudd RA. Increases in drug and opioid-involved overdose deaths—United States, 2010–2015. MMWR Morb Mortal Wkly Rep. 2016;65(50-51):1445–52.
    1. Florence C, Luo F, Xu L, Zhou C. The economic burden of prescription opioid overdose, abuse and dependence in the United States, 2013. Med Care. 2016;54(10):901. doi: 10.1097/MLR.0000000000000625.
    1. Weiner SG, Griggs CA, Mitchell PM, Langlois BK, Friedman FD, Moore RL, Lin SC, Nelson KP, Feldman JA. Clinician impression versus prescription drug monitoring program criteria in the assessment of drug-seeking behavior in the emergency department. Ann Emerg Med. 2013;62(4):281–289. doi: 10.1016/j.annemergmed.2013.05.025.
    1. Longo LP, Parran JT, Johnson B, Kinsey W. Addiction: part II. Identification and management of the drug-seeking patient. Am Fam Physician. 2000;61(8):2401–2408.
    1. Pretorius RW, Zurick GM. A systematic approach to identifying drug-seeking patients. Fam Pract Manag. 2008;15(4):3.
    1. Moeller KE, Lee KC, Kissack JC. Urine drug screening: practical guide for clinicians. Mayo Clin Proc. 2008;92(5);774–96.
    1. Ciesielski T, Iyengar R, Bothra A, Tomala D, Cislo G, Gage BF. A tool to assess risk of de novo opioid abuse or dependence. Am J Med. 2016;129(7):699–705. doi: 10.1016/j.amjmed.2016.02.014.
    1. Rice JB, White AG, Birnbaum HG, Schiller M, Brown DA, Roland CL. A model to identify patients at risk for prescription opioid abuse, dependence, and misuse. Pain Med. 2012;13(9):1162–1173. doi: 10.1111/j.1526-4637.2012.01450.x.
    1. Cochran BN, Flentje A, Heck NC, Van Den Bos J, Perlman D, Torres J, Valuck R, Carter J. Factors predicting development of opioid use disorders among individuals who receive an initial opioid prescription: mathematical modeling using a database of commercially-insured individuals. Drug Alcohol Depend. 2014;138:202–208. doi: 10.1016/j.drugalcdep.2014.02.701.
    1. Dufour R, Mardekian J, Pasquale MK, Schaaf D, Andrews GA, Patel NC. Understanding predictors of opioid abuse: predictive model development and validation. Am J Pharm Benefits. 2014;6(5):208–216.
    1. Hylan TR, Von Korff M, Saunders K, Masters E, Palmer RE, Carrell D, Cronkite D, Mardekian J, Gross D. Automated prediction of risk for problem opioid use in a primary care setting. J Pain. 2015;16(4):380–387. doi: 10.1016/j.jpain.2015.01.011.
    1. Casey JA, Schwartz BS, Stewart WF, Adler NE. Using electronic health records for population health research: a review of methods and applications. Annu Rev Public Health. 2016;37:61–81. doi: 10.1146/annurev-publhealth-032315-021353.
    1. Miotto R, Li L, Kidd BA, Dudley JT. Deep patient: an unsupervised representation to predict the future of patients from the electronic health records. Sci Rep. 2016;6:26094. doi: 10.1038/srep26094.
    1. Denny JC. Mining electronic health records in the genomics era. PLoS Comput Biol. 2012;8(12):e1002823. doi: 10.1371/journal.pcbi.1002823.
    1. Osborn DP, Hardoon S, Omar RZ, Holt RI, King M, Larsen J, Marston L, Morris RW, Nazareth I, Walters K. Cardiovascular risk prediction models for people with severe mental illness: results from the prediction and management of cardiovascular risk in people with severe mental illnesses (PRIMROSE) research program. JAMA psychiatry. 2015;72(2):143–151. doi: 10.1001/jamapsychiatry.2014.2133.
    1. McCoy TH, Castro VM, Rosenfield HR, Cagan A, Kohane IS, Perlis RH. A clinical perspective on the relevance of research domain criteria in electronic health records. Am J Psychiatr. 2015;172(4):316–320. doi: 10.1176/appi.ajp.2014.14091177.
    1. Nguyen P, Tran T, Wickramasinghe N, Venkatesh S. $\mathtt {Deepr} $: a convolutional net for medical records. IEEE J Biomed Health Inform. 2017;21(1):22–30. doi: 10.1109/JBHI.2016.2633963.
    1. Rajkomar A, Oren E, Chen K, Dai AM, Hajaj N, Hardt M, Liu PJ, Liu X, Marcus J, Sun M. Scalable and accurate deep learning with electronic health records. npj Digit Med. 2018;1(1):18. doi: 10.1038/s41746-018-0029-1.
    1. Wang Z, Li L, Glicksberg BS, Israel A, Dudley JT, Ma'ayan A. Predicting age by mining electronic medical records with deep learning characterizes differences between chronological and physiological age. J Biomed Inform. 2017;76:59–68. doi: 10.1016/j.jbi.2017.11.003.
    1. Slee VN. The international classification of diseases: ninth revision (ICD-9) Ann Intern Med. 1978;88(3):424–426. doi: 10.7326/0003-4819-88-3-424.
    1. Iglewicz B, Hoaglin DC. How to detect and handle outliers. Milwaukee: Asq Press; 1993. vol. 16.
    1. Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, Blondel M, Prettenhofer P, Weiss R, Dubourg V. Scikit-learn: machine learning in python. J Mach Learn Res. 2011;12(Oct):2825–2830.
    1. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143(1):29–36. doi: 10.1148/radiology.143.1.7063747.
    1. Mimica Ž, Pogorelić Z, Perko Z, Sršen D, Stipić R, Dujmović D. Effect of surgical incision on pain and respiratory function after abdominal surgery: a randomized clinical trial. Hepato-Gastroenterology. 2007;54(80):2216–2220.
    1. Nishino T, Shimoyama N, Ide T, Isono S. Experimental pain augments experimental dyspnea, but not vice versa in human volunteers. Anesthesiology. 1999;91(6):1633. doi: 10.1097/00000542-199912000-00014.
    1. KIROV SM, Shepherd J, Donald K. Intraoperative and postoperative changes in peripheral white blood cell counts: the contribution of stress. Aust N Z J Surg. 1979;49(6):738–742. doi: 10.1111/j.1445-2197.1979.tb06505.x.
    1. Toft P, Svendsen P, Tonnesen E, Jw R, Christensen N. Redistribution of lymphocytes after major surgical stress. Acta Anaesthesiol Scand. 1993;37(3):245–249. doi: 10.1111/j.1399-6576.1993.tb03708.x.
    1. Cook C, Murphy F, Phelan M, Sander G, Stone K. The case for a harm reduction decade: progress, potential and paradigm shifts: harm reduction international. 2016.
    1. Hellard M, Sacks-Davis R, Gold J. Hepatitis C treatment for injection drug users: a review of the available evidence. Clin Infect Dis. 2009;49(4):561–573. doi: 10.1086/600304.
    1. Grinspan ZM, Abramson EL, Banerjee S, Kern LM, Kaushal R, Shapiro JS. Potential value of health information exchange for people with epilepsy: crossover patterns and missing clinical data. In: AMIA Annual Symposium Proceedings (Washington, DC): 2013. American Medical Informatics Association: 527.
    1. Richter KP, Gibson CA, Ahluwalia JS, Schmelzle KH. Tobacco use and quit attempts among methadone maintenance clients. Am J Public Health. 2001;91(2):296. doi: 10.2105/AJPH.91.2.296.
    1. Flannery AH, Adkins DA, Cook AM. Unpeeling the evidence for the banana bag: evidence-based recommendations for the management of alcohol-associated vitamin and electrolyte deficiencies in the ICU. Crit Care Med. 2016;44(8):1545–1552. doi: 10.1097/CCM.0000000000001659.

Source: PubMed

Sponsorit ja yhteistyökumppanit

Sairaudet

Huumeiden interventiot

3
Tilaa