Forecasting individual risk for long-term Posttraumatic Stress Disorder in emergency medical settings using biomedical data: A machine learning multicenter cohort study

Abstract

The necessary requirement of a traumatic event preceding the development of Posttraumatic Stress Disorder, theoretically allows for administering preventive and early interventions in the early aftermath of such events. Machine learning models including biomedical data to forecast PTSD outcome after trauma are highly promising for detection of individuals most in need of such interventions. In the current study, machine learning was applied on biomedical data collected within 48 h post-trauma to forecast individual risk for long-term PTSD, using a multinominal approach including the full spectrum of common PTSD symptom courses within one prognostic model for the first time. N = 417 patients (37.2% females; mean age 46.09 ± 15.88) admitted with (suspected) serious injury to two urban Academic Level-1 Trauma Centers were included. Routinely collected biomedical information (endocrine measures, vital signs, pharmacotherapy, demographics, injury and trauma characteristics) upon ED admission and subsequent 48 h was used. Cross-validated multi-nominal classification of longitudinal self-reported symptom severity (IES-R) over 12 months and bimodal classification of clinician-rated PTSD diagnosis (CAPS-IV) at 12 months post-trauma was performed using extreme Gradient Boosting and evaluated on hold-out sets. SHapley Additive exPlanations (SHAP) values were used to explain the derived models in human-interpretable form. Good prediction of longitudinal PTSD symptom trajectories (multiclass AUC = 0.89) and clinician-rated PTSD at 12 months (AUC = 0.89) was achieved. Most relevant prognostic variables to forecast both multinominal and dichotomous PTSD outcomes included acute endocrine and psychophysiological measures and hospital-prescribed pharmacotherapy. Thus, individual risk for long-term PTSD was accurately forecasted from biomedical information routinely collected within 48 h post-trauma. These results facilitate future targeted preventive interventions by enabling future early risk detection and provide further insights into the complex etiology of PTSD.

Traumatic stress symptoms in family caregivers of patients with acute leukaemia: protocol for a multisite mixed methods, longitudinal, observational study

“…The high dimensionality and likely multicollinearity among predictors and interaction of predictors pose challenges for statistical models and require the application of advanced computational approaches 91. Studies using advanced ML have been developed to examine predictors of psychiatric risk such as PTSD risk and to facilitate the implementation of precision psychiatry into clinical practice 92–97. We will use a supervised ML approach that is based on well-established methodologies in clinical prediction modelling including data pre-processing, such as handling of missing values, guarding against ‘overfitting’, and rigorous model evaluation in terms of established metrics for discrimination and calibration 98–103.…”

“…91 Studies using advanced ML have been developed to examine predictors of psychiatric risk such as PTSD risk and to facilitate the implementation of precision psychiatry into clinical practice. [92][93][94][95][96][97] We will use a supervised ML approach that is based on well-established methodologies in clinical prediction modelling including data pre-processing, such as handling of missing values, guarding against 'overfitting', and rigorous model evaluation in terms of established metrics for discrimination and calibration. [98][99][100][101][102][103] Confidence intervals for all point estimates will be calculated to communicate uncertainty of the model.…”

Psychiatry in the Digital Age: A Blessing or a Curse?

“…Further promising areas of research regarding ML use in psychiatry include the evaluation of the individual risk for long-term posttraumatic stress disorder (PTSD) based on predictor variables from clinical records, questionnaires, biomedical data, and neuroimaging [222][223][224]. This application of ML could have a significant clinical impact in consideration of emerging evidence for early psychological and pharmacological interventions in individuals at risk for long-term PTSD [225,226].…”

Oxytocin vs. placebo effects on intrusive memory consolidation using a trauma film paradigm: a randomized, controlled experimental study in healthy women

“…ML approaches provide the possibility to investigate a variety of variables and their complex interactions [49]. Using ML approaches, it has been possible to identify biomarkers and multiple polygenic risk scores associated with posttraumatic stress [12,[50][51][52].…”