Machine-Learning prediction of comorbid substance use disorders in ADHD youth using Swedish registry data.
Machine learning
attention-deficit hyperactive disorder
comorbidity
risk factor
substance use disorder
Journal
Journal of child psychology and psychiatry, and allied disciplines
ISSN: 1469-7610
Titre abrégé: J Child Psychol Psychiatry
Pays: England
ID NLM: 0375361
Informations de publication
Date de publication:
12 2020
12 2020
Historique:
received:
24
10
2019
revised:
03
01
2020
accepted:
28
01
2020
pubmed:
3
4
2020
medline:
30
9
2021
entrez:
3
4
2020
Statut:
ppublish
Résumé
Children with attention-deficit/hyperactivity disorder (ADHD) have a high risk for substance use disorders (SUDs). Early identification of at-risk youth would help allocate scarce resources for prevention programs. Psychiatric and somatic diagnoses, family history of these disorders, measures of socioeconomic distress, and information about birth complications were obtained from the national registers in Sweden for 19,787 children with ADHD born between 1989 and 1993. We trained (a) a cross-sectional random forest (RF) model using data available by age 17 to predict SUD diagnosis between ages 18 and 19; and (b) a longitudinal recurrent neural network (RNN) model with the Long Short-Term Memory (LSTM) architecture to predict new diagnoses at each age. The area under the receiver operating characteristic curve (AUC) was 0.73(95%CI 0.70-0.76) for the random forest model (RF). Removing prior diagnosis from the predictors, the RF model was still able to achieve significant AUCs when predicting all SUD diagnoses (0.69, 95%CI 0.66-0.72) or new diagnoses (0.67, 95%CI: 0.64, 0.71) during age 18-19. For the model predicting new diagnoses, model calibration was good with a low Brier score of 0.086. Longitudinal LSTM model was able to predict later SUD risks at as early as 2 years age, 10 years before the earliest diagnosis. The average AUC from longitudinal models predicting new diagnoses 1, 2, 5 and 10 years in the future was 0.63. Population registry data can be used to predict at-risk comorbid SUDs in individuals with ADHD. Such predictions can be made many years prior to age of the onset, and their SUD risks can be monitored using longitudinal models over years during child development. Nevertheless, more work is needed to create prediction models based on electronic health records or linked population registers that are sufficiently accurate for use in the clinic.
Sections du résumé
BACKGROUND
Children with attention-deficit/hyperactivity disorder (ADHD) have a high risk for substance use disorders (SUDs). Early identification of at-risk youth would help allocate scarce resources for prevention programs.
METHODS
Psychiatric and somatic diagnoses, family history of these disorders, measures of socioeconomic distress, and information about birth complications were obtained from the national registers in Sweden for 19,787 children with ADHD born between 1989 and 1993. We trained (a) a cross-sectional random forest (RF) model using data available by age 17 to predict SUD diagnosis between ages 18 and 19; and (b) a longitudinal recurrent neural network (RNN) model with the Long Short-Term Memory (LSTM) architecture to predict new diagnoses at each age.
RESULTS
The area under the receiver operating characteristic curve (AUC) was 0.73(95%CI 0.70-0.76) for the random forest model (RF). Removing prior diagnosis from the predictors, the RF model was still able to achieve significant AUCs when predicting all SUD diagnoses (0.69, 95%CI 0.66-0.72) or new diagnoses (0.67, 95%CI: 0.64, 0.71) during age 18-19. For the model predicting new diagnoses, model calibration was good with a low Brier score of 0.086. Longitudinal LSTM model was able to predict later SUD risks at as early as 2 years age, 10 years before the earliest diagnosis. The average AUC from longitudinal models predicting new diagnoses 1, 2, 5 and 10 years in the future was 0.63.
CONCLUSIONS
Population registry data can be used to predict at-risk comorbid SUDs in individuals with ADHD. Such predictions can be made many years prior to age of the onset, and their SUD risks can be monitored using longitudinal models over years during child development. Nevertheless, more work is needed to create prediction models based on electronic health records or linked population registers that are sufficiently accurate for use in the clinic.
Identifiants
pubmed: 32237241
doi: 10.1111/jcpp.13226
pmc: PMC7754321
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
1370-1379Informations de copyright
© 2020 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for Child and Adolescent Mental Health.
Références
Psychol Health. 2017 Aug;32(8):976-1017
pubmed: 28513195
J Abnorm Psychol. 2003 Aug;112(3):497-507
pubmed: 12943028
J Child Psychol Psychiatry. 2014 Aug;55(8):878-85
pubmed: 25158998
Alcohol Clin Exp Res. 2018 Jun;42(6):1084-1095
pubmed: 29693716
Neural Comput. 1997 Nov 15;9(8):1735-80
pubmed: 9377276
Pediatrics. 2003 Jan;111(1):179-85
pubmed: 12509574
Psychol Med. 2006 Feb;36(2):167-79
pubmed: 16420713
Alcohol Alcohol. 2012 Mar-Apr;47(2):178-86
pubmed: 22262507
Biol Psychiatry Cogn Neurosci Neuroimaging. 2018 Jul;3(7):635-643
pubmed: 30047479
Biometrics. 1988 Sep;44(3):837-45
pubmed: 3203132
Pharmacoepidemiol Drug Saf. 2007 Jul;16(7):726-35
pubmed: 16897791
Addiction. 2015 Jan;110(1):19-35
pubmed: 25287883
Subst Use Misuse. 2015 Feb;50(3):328-39
pubmed: 25493643
Pediatrics. 2014 Jun;133(6):1070-80
pubmed: 24819571
Arch Gen Psychiatry. 2012 Jul;69(7):690-7
pubmed: 22393206
Lancet Psychiatry. 2015 Oct;2(10):891-900
pubmed: 26342957
Lancet. 2013 Nov 9;382(9904):1575-86
pubmed: 23993280
N Engl J Med. 2012 Nov 22;367(21):2006-14
pubmed: 23171097
J Am Acad Child Adolesc Psychiatry. 2007 Aug;46(8):1028-1040
pubmed: 17667481
Eur J Epidemiol. 2009;24(11):659-67
pubmed: 19504049
Drug Alcohol Depend. 2013 Nov 1;133(1):212-21
pubmed: 23726978
J Learn Disabil. 1998 Nov-Dec;31(6):533-44
pubmed: 9813951
Drug Test Anal. 2014 Jan-Feb;6(1-2):39-45
pubmed: 23836598
Ethn Dis. 2000 Winter;10(1):39-52
pubmed: 10764129
Drug Alcohol Depend. 2008 Dec 1;98(3):264-9
pubmed: 18757141
Drug Alcohol Depend. 2016 Aug 01;165:71-8
pubmed: 27242288
Methods Mol Biol. 2011;675:215-20
pubmed: 20949391
Br J Psychiatry. 2013 Aug;203(2):103-6
pubmed: 23703314
Soc Sci Med. 2018 Jul;209:51-58
pubmed: 29793164
Curr Med Chem. 2012;19(33):5612-8
pubmed: 22856661
Acta Obstet Gynecol Scand. 2000 Jun;79(6):440-9
pubmed: 10857867
Biol Psychiatry. 2015 May 15;77(10):880-6
pubmed: 25457227
J Clin Psychiatry. 2017 May;78(5):572-583
pubmed: 27337225
JAMA Psychiatry. 2014 Mar;71(3):319-25
pubmed: 24477798
Harm Reduct J. 2014 Aug 20;11:24
pubmed: 25142321
Hum Genet. 2012 Jun;131(6):779-89
pubmed: 22543841
Am J Psychiatry. 2017 Sep 1;174(9):877-885
pubmed: 28659039
Am J Psychiatry. 2017 Feb 1;174(2):154-162
pubmed: 27609239
Soc Work Public Health. 2013;28(3-4):175-7
pubmed: 23731412
Clin Obstet Gynecol. 2013 Mar;56(1):124-32
pubmed: 23314714
Int J Epidemiol. 2013 Aug;42(4):1057-66
pubmed: 24062294
Am J Psychiatry. 2018 Oct 1;175(10):951-960
pubmed: 29792051
JAMA. 2009 May 20;301(19):2016-23
pubmed: 19454640
Scand J Public Health. 2004;32(4):250-6
pubmed: 15370764
Eur J Epidemiol. 2016 Feb;31(2):125-36
pubmed: 26769609
Harv Rev Psychiatry. 2001 Nov-Dec;9(6):267-79
pubmed: 11600486
Lancet. 2016 Oct 8;388(10053):1459-1544
pubmed: 27733281
Drug Alcohol Rev. 2018 Sep;37(6):743-751
pubmed: 29998467
BMC Public Health. 2011 Jun 09;11:450
pubmed: 21658213