Predicting mechanical restraint of psychiatric inpatients by applying machine learning on electronic health data

A. A. Danielsen*, M. H.J. Fenger, S. D. Østergaard, K. L. Nielbo, O. Mors

*Kontaktforfatter for dette arbejde

Publikation: Bidrag til tidsskriftTidsskriftartikelForskningpeer review

Resumé

Objective: Mechanical restraint (MR) is used to prevent patients from harming themselves or others during inpatient treatment. The objective of this study was to investigate whether incident MR occurring in the first 3 days following admission could be predicted based on analysis of electronic health data available after the first hour of admission. Methods: The dataset consisted of clinical notes from electronic health records from the Central Denmark Region and data from the Danish Health Registers from patients admitted to a psychiatric department in the period from 2011 to 2015. Supervised machine learning algorithms were trained on a randomly selected subset of the data and validated using an independent test dataset. Results: A total of 5050 patients with 8869 admissions were included in the study. One hundred patients were mechanically restrained in the period between one hour and 3 days after the admission. A Random Forest algorithm predicted MR with an area under the curve of 0.87 (95% CI 0.79–0.93). At 94% specificity, the sensitivity was 56%. Among the ten strongest predictors, nine were derived from the clinical notes. Conclusions: These findings open for the development of an early warning system that may guide interventions to reduce the use of MR.

OriginalsprogEngelsk
TidsskriftActa Psychiatrica Scandinavica
Vol/bind140
Udgave nummer2
Sider (fra-til)147-157
ISSN0001-690X
DOI
StatusUdgivet - aug. 2019

Fingeraftryk

Inpatients
Health
Electronic Health Records
Denmark
Area Under Curve
Datasets

Citer dette

Danielsen, A. A. ; Fenger, M. H.J. ; Østergaard, S. D. ; Nielbo, K. L. ; Mors, O. / Predicting mechanical restraint of psychiatric inpatients by applying machine learning on electronic health data. I: Acta Psychiatrica Scandinavica. 2019 ; Bind 140, Nr. 2. s. 147-157.
@article{c70742922ad1427497bf03a436dcd2d1,
title = "Predicting mechanical restraint of psychiatric inpatients by applying machine learning on electronic health data",
abstract = "Objective: Mechanical restraint (MR) is used to prevent patients from harming themselves or others during inpatient treatment. The objective of this study was to investigate whether incident MR occurring in the first 3 days following admission could be predicted based on analysis of electronic health data available after the first hour of admission. Methods: The dataset consisted of clinical notes from electronic health records from the Central Denmark Region and data from the Danish Health Registers from patients admitted to a psychiatric department in the period from 2011 to 2015. Supervised machine learning algorithms were trained on a randomly selected subset of the data and validated using an independent test dataset. Results: A total of 5050 patients with 8869 admissions were included in the study. One hundred patients were mechanically restrained in the period between one hour and 3 days after the admission. A Random Forest algorithm predicted MR with an area under the curve of 0.87 (95{\%} CI 0.79–0.93). At 94{\%} specificity, the sensitivity was 56{\%}. Among the ten strongest predictors, nine were derived from the clinical notes. Conclusions: These findings open for the development of an early warning system that may guide interventions to reduce the use of MR.",
keywords = "coercion, electronic medical records, mental disorders, natural language processing, supervised machine learning",
author = "Danielsen, {A. A.} and Fenger, {M. H.J.} and {\O}stergaard, {S. D.} and Nielbo, {K. L.} and O. Mors",
year = "2019",
month = "8",
doi = "10.1111/acps.13061",
language = "English",
volume = "140",
pages = "147--157",
journal = "Acta Psychiatrica Scandinavica",
issn = "0001-690X",
publisher = "John Wiley & Sons Ltd",
number = "2",

}

Predicting mechanical restraint of psychiatric inpatients by applying machine learning on electronic health data. / Danielsen, A. A.; Fenger, M. H.J.; Østergaard, S. D.; Nielbo, K. L.; Mors, O.

I: Acta Psychiatrica Scandinavica, Bind 140, Nr. 2, 08.2019, s. 147-157.

Publikation: Bidrag til tidsskriftTidsskriftartikelForskningpeer review

TY - JOUR

T1 - Predicting mechanical restraint of psychiatric inpatients by applying machine learning on electronic health data

AU - Danielsen, A. A.

AU - Fenger, M. H.J.

AU - Østergaard, S. D.

AU - Nielbo, K. L.

AU - Mors, O.

PY - 2019/8

Y1 - 2019/8

N2 - Objective: Mechanical restraint (MR) is used to prevent patients from harming themselves or others during inpatient treatment. The objective of this study was to investigate whether incident MR occurring in the first 3 days following admission could be predicted based on analysis of electronic health data available after the first hour of admission. Methods: The dataset consisted of clinical notes from electronic health records from the Central Denmark Region and data from the Danish Health Registers from patients admitted to a psychiatric department in the period from 2011 to 2015. Supervised machine learning algorithms were trained on a randomly selected subset of the data and validated using an independent test dataset. Results: A total of 5050 patients with 8869 admissions were included in the study. One hundred patients were mechanically restrained in the period between one hour and 3 days after the admission. A Random Forest algorithm predicted MR with an area under the curve of 0.87 (95% CI 0.79–0.93). At 94% specificity, the sensitivity was 56%. Among the ten strongest predictors, nine were derived from the clinical notes. Conclusions: These findings open for the development of an early warning system that may guide interventions to reduce the use of MR.

AB - Objective: Mechanical restraint (MR) is used to prevent patients from harming themselves or others during inpatient treatment. The objective of this study was to investigate whether incident MR occurring in the first 3 days following admission could be predicted based on analysis of electronic health data available after the first hour of admission. Methods: The dataset consisted of clinical notes from electronic health records from the Central Denmark Region and data from the Danish Health Registers from patients admitted to a psychiatric department in the period from 2011 to 2015. Supervised machine learning algorithms were trained on a randomly selected subset of the data and validated using an independent test dataset. Results: A total of 5050 patients with 8869 admissions were included in the study. One hundred patients were mechanically restrained in the period between one hour and 3 days after the admission. A Random Forest algorithm predicted MR with an area under the curve of 0.87 (95% CI 0.79–0.93). At 94% specificity, the sensitivity was 56%. Among the ten strongest predictors, nine were derived from the clinical notes. Conclusions: These findings open for the development of an early warning system that may guide interventions to reduce the use of MR.

KW - coercion

KW - electronic medical records

KW - mental disorders

KW - natural language processing

KW - supervised machine learning

U2 - 10.1111/acps.13061

DO - 10.1111/acps.13061

M3 - Journal article

VL - 140

SP - 147

EP - 157

JO - Acta Psychiatrica Scandinavica

JF - Acta Psychiatrica Scandinavica

SN - 0001-690X

IS - 2

ER -