Machine learning as a diagnostic decision aid for patients with transient loss of consciousness.
Journal
Neurology. Clinical practice
ISSN: 2163-0402
Titre abrégé: Neurol Clin Pract
Pays: United States
ID NLM: 101577149
Informations de publication
Date de publication:
Apr 2020
Apr 2020
Historique:
received:
21
12
2018
accepted:
25
07
2019
entrez:
21
4
2020
pubmed:
21
4
2020
medline:
21
4
2020
Statut:
ppublish
Résumé
Transient loss of consciousness (TLOC) is a common reason for presentation to primary/emergency care; over 90% are because of epilepsy, syncope, or psychogenic non-epileptic seizures (PNES). Misdiagnoses are common, and there are currently no validated decision rules to aid diagnosis and management. We seek to explore the utility of machine-learning techniques to develop a short diagnostic instrument by extracting features with optimal discriminatory values from responses to detailed questionnaires about TLOC manifestations and comorbidities (86 questions to patients, 31 to TLOC witnesses). Multi-center retrospective self- and witness-report questionnaire study in secondary care settings. Feature selection was performed by an iterative algorithm based on random forest analysis. Data were randomly divided in a 2:1 ratio into training and validation sets (163:86 for all data; 208:92 for analysis excluding witness reports). Three hundred patients with proven diagnoses (100 each: epilepsy, syncope and PNES) were recruited from epilepsy and syncope services. Two hundred forty-nine completed patient and witness questionnaires: 86 epilepsy (64 female), 84 PNES (61 female), and 79 syncope (59 female). Responses to 36 questions optimally predicted diagnoses. A classifier trained on these features classified 74/86 (86.0% [95% confidence interval 76.9%-92.6%]) of patients correctly in validation (100 [86.7%-100%] syncope, 85.7 [67.3%-96.0%] epilepsy, 75.0 [56.6%-88.5%] PNES). Excluding witness reports, 34 features provided optimal prediction (classifier accuracy of 72/92 [78.3 (68.4%-86.2%)] in validation, 83.8 [68.0%-93.8%] syncope, 81.5 [61.9%-93.7%] epilepsy, 67.9 [47.7%-84.1%] PNES). A tool based on patient symptoms/comorbidities and witness reports separates well between syncope and other common causes of TLOC. It can help to differentiate epilepsy and PNES. Validated decision rules may improve diagnostic processes and reduce misdiagnosis rates. This study provides Class III evidence that for patients with TLOC, patient and witness questionnaires discriminate between syncope, epilepsy and PNES.
Sections du résumé
BACKGROUND
BACKGROUND
Transient loss of consciousness (TLOC) is a common reason for presentation to primary/emergency care; over 90% are because of epilepsy, syncope, or psychogenic non-epileptic seizures (PNES). Misdiagnoses are common, and there are currently no validated decision rules to aid diagnosis and management. We seek to explore the utility of machine-learning techniques to develop a short diagnostic instrument by extracting features with optimal discriminatory values from responses to detailed questionnaires about TLOC manifestations and comorbidities (86 questions to patients, 31 to TLOC witnesses).
METHODS
METHODS
Multi-center retrospective self- and witness-report questionnaire study in secondary care settings. Feature selection was performed by an iterative algorithm based on random forest analysis. Data were randomly divided in a 2:1 ratio into training and validation sets (163:86 for all data; 208:92 for analysis excluding witness reports).
RESULTS
RESULTS
Three hundred patients with proven diagnoses (100 each: epilepsy, syncope and PNES) were recruited from epilepsy and syncope services. Two hundred forty-nine completed patient and witness questionnaires: 86 epilepsy (64 female), 84 PNES (61 female), and 79 syncope (59 female). Responses to 36 questions optimally predicted diagnoses. A classifier trained on these features classified 74/86 (86.0% [95% confidence interval 76.9%-92.6%]) of patients correctly in validation (100 [86.7%-100%] syncope, 85.7 [67.3%-96.0%] epilepsy, 75.0 [56.6%-88.5%] PNES). Excluding witness reports, 34 features provided optimal prediction (classifier accuracy of 72/92 [78.3 (68.4%-86.2%)] in validation, 83.8 [68.0%-93.8%] syncope, 81.5 [61.9%-93.7%] epilepsy, 67.9 [47.7%-84.1%] PNES).
CONCLUSIONS
CONCLUSIONS
A tool based on patient symptoms/comorbidities and witness reports separates well between syncope and other common causes of TLOC. It can help to differentiate epilepsy and PNES. Validated decision rules may improve diagnostic processes and reduce misdiagnosis rates.
CLASSIFICATION OF EVIDENCE
METHODS
This study provides Class III evidence that for patients with TLOC, patient and witness questionnaires discriminate between syncope, epilepsy and PNES.
Identifiants
pubmed: 32309027
doi: 10.1212/CPJ.0000000000000726
pii: NEURCLINPRACT2018036293
pmc: PMC7156196
doi:
Types de publication
Journal Article
Langues
eng
Pagination
96-105Informations de copyright
© 2019 American Academy of Neurology.
Références
BMJ. 2015 Jan 07;350:g7594
pubmed: 25569120
J Am Coll Cardiol. 2000 Jul;36(1):181-4
pubmed: 10898432
Postgrad Med J. 2006 Oct;82(972):630-41
pubmed: 17068273
Front Psychol. 2010 Nov 16;1:192
pubmed: 21833252
Br J Surg. 1998 Sep;85(9):1217-20
pubmed: 9752863
Neurology. 2016 Aug 9;87(6):625-33
pubmed: 27385741
Am J Emerg Med. 2009 Mar;27(3):271-9
pubmed: 19328369
Epilepsia. 1999 Apr;40(4):485-91
pubmed: 10219276
Neurology. 2009 May 12;72(19):1646-52
pubmed: 19433737
Europace. 2010 Jun;12(6):869-76
pubmed: 20215367
Cochrane Database Syst Rev. 2009 Jul 08;(3):MR000008
pubmed: 19588449
Am J Med. 2014 Nov;127(11):1126.e13-25
pubmed: 24862309
J Clin Neurosci. 2016 Dec;34:105-107
pubmed: 27473020
J Neurol Neurosurg Psychiatry. 2010 Jul;81(7):719-25
pubmed: 20581136
Epilepsy Behav. 2018 Sep;86:6-8
pubmed: 30032094
Seizure. 2008 Jul;17(5):431-6
pubmed: 18282726
Arch Neurol. 1998 Nov;55(11):1456-9
pubmed: 9823830
BMJ. 2017 May 23;357:j2099
pubmed: 28536104
J Am Coll Cardiol. 2002 Jul 3;40(1):142-8
pubmed: 12103268
J Neurol Neurosurg Psychiatry. 2004 Jul;75(7):1009-12
pubmed: 15201361
J Neurol. 1991 Feb;238(1):39-43
pubmed: 2030371
Epilepsia. 2013 Nov;54(11):2005-18
pubmed: 24111933
BMJ. 2002 Aug 10;325(7359):314
pubmed: 12169509
BMC Med Res Methodol. 2006 Feb 23;6:5
pubmed: 16504090
Eur Heart J. 2018 Jun 1;39(21):1883-1948
pubmed: 29562304
Epilepsy Behav. 2010 Nov;19(3):478-82
pubmed: 20850387
Seizure. 2005 Oct;14(7):514-20
pubmed: 16188464
BMC Bioinformatics. 2006 Jan 06;7:3
pubmed: 16398926
Epilepsy Behav. 2010 Nov;19(3):522-5
pubmed: 20920893
Epilepsy Behav. 2014 Aug;37:210-4
pubmed: 25084477
Neurology. 2002 Feb 12;58(3):493-5
pubmed: 11839862
Neurology. 2019 Feb 26;92(9):e895-e904
pubmed: 30804064
Seizure. 2018 Oct;61:139-148
pubmed: 30145472
Epilepsia. 2008 May;49(5):898-904
pubmed: 18070093
Acad Emerg Med. 2007 Nov;14(11):955-9
pubmed: 17923717
Neurology. 2004 Mar 9;62(5):834-5
pubmed: 15007151
Epilepsy Res. 2001 Mar;43(3):193-9
pubmed: 11248531
Epilepsy Behav. 2010 Nov;19(3):462-6
pubmed: 20926353
Epilepsy Behav. 2015 Apr;45:151-4
pubmed: 25812935
Seizure. 2017 May;48:22-27
pubmed: 28371670