Using machine learning with passive wearable sensors to pilot the detection of eating disorder behaviors in everyday life DOI
Christina Ralph‐Nearman, Luis E Sandoval-Araujo, Andrew Karem

и другие.

Psychological Medicine, Год журнала: 2023, Номер 54(6), С. 1084 - 1090

Опубликована: Окт. 20, 2023

Abstract Background Eating disorders (ED) are serious psychiatric disorders, taking a life every 52 minutes, with high relapse. There currently no support or effective intervention therapeutics for individuals an ED in their everyday life. The aim of this study is to build idiographic machine learning (ML) models evaluate the performance physiological recordings detect individual behaviors naturalistic settings. Methods From ongoing (Final N = 120), we piloted ability ML individual's behavioral episodes (e.g. purging) from data six diagnosed ED, all whom endorsed purging. Participants wore ambulatory monitor 30 days and tapped button denote episodes. We built ( 1) logistic regression classifiers (LRC) trained identify onset (~600 windows) v. baseline (~571 physiology (Heart Rate, Electrodermal Activity, Temperature). Results Using data, LRC accurately classified on average 91% cases, 92% specificity 90% sensitivity. Conclusions This evidence suggests that indices within level accuracy. novel use wearable sensors patterns behavior pre-onset can lead just-in-time clinical interventions disrupt problematic promote recovery.

Язык: Английский

Using machine learning with passive wearable sensors to pilot the detection of eating disorder behaviors in everyday life DOI
Christina Ralph‐Nearman, Luis E Sandoval-Araujo, Andrew Karem

и другие.

Psychological Medicine, Год журнала: 2023, Номер 54(6), С. 1084 - 1090

Опубликована: Окт. 20, 2023

Abstract Background Eating disorders (ED) are serious psychiatric disorders, taking a life every 52 minutes, with high relapse. There currently no support or effective intervention therapeutics for individuals an ED in their everyday life. The aim of this study is to build idiographic machine learning (ML) models evaluate the performance physiological recordings detect individual behaviors naturalistic settings. Methods From ongoing (Final N = 120), we piloted ability ML individual's behavioral episodes (e.g. purging) from data six diagnosed ED, all whom endorsed purging. Participants wore ambulatory monitor 30 days and tapped button denote episodes. We built ( 1) logistic regression classifiers (LRC) trained identify onset (~600 windows) v. baseline (~571 physiology (Heart Rate, Electrodermal Activity, Temperature). Results Using data, LRC accurately classified on average 91% cases, 92% specificity 90% sensitivity. Conclusions This evidence suggests that indices within level accuracy. novel use wearable sensors patterns behavior pre-onset can lead just-in-time clinical interventions disrupt problematic promote recovery.

Язык: Английский

Процитировано

10