Development and Validation of a Machine Learning Prediction Model of Posttraumatic Stress Disorder After Military Deployment

JAMA Network Open, Год журнала: 2023, Номер 6(6), С. e2321273 - e2321273

Опубликована: Июнь 30, 2023

Importance Military deployment involves significant risk for life-threatening experiences that can lead to posttraumatic stress disorder (PTSD). Accurate predeployment prediction of PTSD may facilitate the development targeted intervention strategies enhance resilience. Objective To develop and validate a machine learning (ML) model predict postdeployment PTSD. Design, Setting, Participants This diagnostic/prognostic study included 4771 soldiers from 3 US Army brigade combat teams who completed assessments between January 9, 2012, May 1, 2014. Predeployment occurred 1 2 months before Afghanistan, follow-up approximately 9 post deployment. Machine models were developed in first recruited cohorts using as many 801 predictors comprehensive self-report assessments. In phase, cross-validated performance metrics predictor parsimony considered select an optimal model. Next, selected model’s was evaluated with area under receiver operating characteristics curve expected calibration error temporally geographically distinct cohort. Data analyses performed August November 30, 2022. Main Outcomes Measures Posttraumatic diagnosis assessed by clinically calibrated measures. weighted all address potential biases related cohort selection nonresponse. Results participants (mean [SD] age, 26.9 [6.2] years), 4440 (94.7%) whom men. terms race ethnicity, 144 (2.8%) identified American Indian or Alaska Native, 242 (4.8%) Asian, 556 (13.3%) Black African American, 885 (18.3%) Hispanic, 106 (2.1%) Native Hawaiian other Pacific Islander, 3474 (72.2%) White, 430 (8.9%) unknown ethnicity; could identify more than ethnicity. A total 746 (15.4%) met criteria had comparable (log loss range, 0.372-0.375; 0.75-0.76). gradient-boosting 58 core over elastic net 196 stacked ensemble ML predictors. independent test cohort, 0.74 (95% CI, 0.71-0.77) low 0.032 0.020-0.046). Approximately one-third highest accounted 62.4% 56.5%-67.9%) cases. Core cut across 17 domains: stressful experiences, social network, substance use, childhood adolescence, unit health, injuries, irritability anger, personality, emotional problems, resilience, treatment, anxiety, attention concentration, family history, mood, religion. Conclusions Relevance this soldiers, self-reported information collected The showed good validation sample. These results indicate stratification is feasible prevention early strategies.

Язык: Английский

---

Improving explainability of post-separation suicide attempt prediction models for transitioning service members: insights from the Army Study to Assess Risk and Resilience in Servicemembers — Longitudinal Study

Translational Psychiatry, Год журнала: 2025, Номер 15(1)

Опубликована: Янв. 30, 2025

Язык: Английский

---

Study protocol of an open-label proof-of-concept trial examining the safety and clinical efficacy of psilocybin-assisted therapy for veterans with PTSD

BMJ Open, Год журнала: 2023, Номер 13(5), С. e068884 - e068884

Опубликована: Май 1, 2023

Introduction Psilocybin-assisted therapy has shown significant promise in treating the cluster of mood and anxiety symptoms that comprise post-traumatic stress disorder (PTSD) but yet to be tested specifically this condition. Furthermore, current pharmacological psychotherapeutic treatments for PTSD are difficult tolerate limited efficacy, especially US Military Veteran (USMV) population. This open-label pilot study will examine safety efficacy two psilocybin administration sessions (15 mg 25 mg), combined with psychotherapy, among USMVs severe, treatment resistant PTSD. Methods analysis We recruit 15 Participants receive one low dose mg) moderate/high (25 conjunction preparatory post-psilocybin sessions. The primary outcome type, severity frequency adverse events suicidal ideation/behaviour, as measured by Columbia Suicide Severity Rating Scale. measure Clinician Administered Scale-5. endpoint 1 month following second session, total follow-up time 6 months. Ethics dissemination All participants required provide written informed consent. trial been authorised Ohio State University Institutional Review Board (study number: 2022H0280). Dissemination results occur via a peer-reviewed publication other relevant media. Trial registration number NCT05554094 .

Язык: Английский

---

NLP-enriched social determinants of health improve prediction of suicide death among the Veterans

Research Square (Research Square), Год журнала: 2025, Номер unknown

Опубликована: Март 31, 2025

Predictions of suicide death patients discharged from psychiatric hospitals (PDPH) can guide intervention efforts including intensive post-discharge case management programs, designed to reduce risk among high-risk patients. This study aims determine if additions social and behavioral determinants health (SBDH) as predictors could improve the prediction PDPH. We analyzed a cohort 197,581 US Veterans 129 VHA across between January 1, 2017, July 2019 with total 414,043 discharges. Predictive variables included administrative data SBDH, latter derived unstructured clinical notes via natural language processing (NLP) system ICD codes, observed within 365-day window prior discharge. evaluated impact SBDH on predictive performance two advanced models: an ensemble traditional machine learning models transformer-based deep foundation model for electronic records (TransformEHR). measured sensitivity, positive value (PPV), area under receiver operating characteristic curve (AUROC) overall by gender. Calibration analysis was also conducted measure reliability. TransformEHR achieved AUROC 64.04. Specifically, ICD-based improved 3.1% (95% CI, 1.6% – 4.5%) 2.9% 0.5% 5.4%) TransformEHR, compared without SBDH. NLP-extracted further AUROC: 1.7% 0.1%– 3.3%) 1.8% 0.6%– 2.9%) TransformEHR. 0.2%, 0.4%, 0.8%, PPV per 100 PDPH 7, 30, 90, 180 respectively. Moreover, showed superior calibration fairness model, improving both models. In conclusion, performance, calibration, after their

Язык: Английский

---

A practical risk calculator for suicidal behavior among transitioning U.S. Army soldiers: results from the Study to Assess Risk and Resilience in Servicemembers-Longitudinal Study (STARRS-LS)

Psychological Medicine, Год журнала: 2023, Номер 53(15), С. 7096 - 7105

Опубликована: Март 9, 2023

Abstract Background Risk of suicide-related behaviors is elevated among military personnel transitioning to civilian life. An earlier report showed that high-risk U.S. Army soldiers could be identified shortly before this transition with a machine learning model included predictors from administrative systems, self-report surveys, and geospatial data. Based on result, Veterans Affairs initiative was launched evaluate suicide-prevention intervention for soldiers. To make targeting practical, though, streamlined risk calculator were needed used only short series survey questions. Methods We revised the original in sample n = 8335 observations Study Assess Resilience Servicemembers-Longitudinal (STARRS-LS) who participated one three STARRS 2011–2014 baseline surveys while service or more subsequent panel (LS1: 2016–2018, LS2: 2018–2019) after leaving service. trained ensemble models constrained numbers item-level 70% training sample. The outcome self-reported post-transition suicide attempts (SA). validated 30% test Results Twelve-month SA prevalence 1.0% ( s.e. 0.1). best model, 17 predictors, had ROC-AUC 0.85 0.03). 10–30% respondents highest predicted 44.9–92.5% 12-month SAs. Conclusions accurate based can target prevent SA.

Язык: Английский

---

Predicting Homelessness Among Transitioning U.S. Army Soldiers

American Journal of Preventive Medicine, Год журнала: 2024, Номер 66(6), С. 999 - 1007

Опубликована: Фев. 3, 2024

Язык: Английский

---

The use of machine learning on administrative and survey data to predict suicidal thoughts and behaviors: a systematic review

Frontiers in Psychiatry, Год журнала: 2024, Номер 15

Опубликована: Март 4, 2024

Background Machine learning is a promising tool in the area of suicide prevention due to its ability combine effects multiple risk factors and complex interactions. The power machine has led an influx studies on prediction, as well few recent reviews. Our study distinguished between data sources reported most important predictors outcomes identified literature. Objective aimed identify that applied techniques administrative survey data, summarize performance metrics those studies, enumerate suicidal thoughts behaviors identified. Methods A systematic literature search PubMed, Medline, Embase, PsycINFO, Web Science, Cumulative Index Nursing Allied Health Literature (CINAHL), Complementary Medicine Database (AMED) all have used predict using was performed. conducted for articles published January 1, 2019 May 11, 2022. In addition, three recently reviews (the last which included up until 2019) were retained if they met our inclusion criteria. predictive methods predicting explored box plots distribution under receiver operating characteristic curve (AUC) values by method outcome (i.e., thoughts, attempt, death suicide). Mean AUCs with 95% confidence intervals (CIs) computed each design, source, total sample size, size cases, employed. listed. Results strategy 2,200 unique records, 104 algorithms achieved good prediction AUC 0.80 0.89); however, their appears differ across outcomes. boosting suicide, combined, while neural network attempts. differed depending source population study. Conclusion utility largely depends approach used. findings current review should prove helpful preparing future models data. Systematic registration https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42022333454 identifier CRD42022333454.

Язык: Английский

---

Exposure to Bullying or Hazing During Deployment and Mental Health Outcomes Among US Army Soldiers

JAMA Network Open, Год журнала: 2023, Номер 6(1), С. e2252109 - e2252109

Опубликована: Янв. 24, 2023

Importance Workplace bullying is associated with mental disorders and suicidality in civilians, but few studies have examined associations of these outcomes among military personnel. Objective To evaluate being bullied or hazed during deployment major depressive disorder (MDD), intermittent explosive disorder, posttraumatic stress (PTSD), suicidal ideation, substance use (SUD). Design, Setting, Participants This cohort study used data from the Army Study to Assess Risk Resilience Servicemembers (Army STARRS) New Soldier (NSS; April 1, 2011, November 30, 2012) wave 1 STARRS Longitudinal (STARRS-LS1; September 2016, 2018). A computerized survey administered at 3 US installations (NSS) a web/telephone (STARRS-LS1) were collect data. Data analyzed October 11, 2021, 28, 2022. The STARRS-LS1 recruited probability sample active-duty soldiers veterans who had participated baseline surveys while on active duty (weighted response rate, 35.6%). Respondents whose was NSS deployed combat theater least once eligible for this study. Exposures Being deployment. Main Outcomes Measures primary MDD, PTSD, ideation 12 months before SUD 30 days STARRS-LS1, assessed items Composite International Diagnostic Interview Screening Scales, PTSD Checklist Statistical Manual Mental Disorders, Fifth Edition , Columbia-Suicide Severity Rating Scale. Logistic regression estimate hazing exposure outcomes. Results 1463 participants predominantly male percentage [SE], 90.4% [0.9%]) mean (SE) age 21.1 (0.1) years baseline. At 188 respondents 12.2% [1.1%]) reported Weighted outcome prevalences 18.7% (1.3%) 5.2% (0.9%) 21.8% (1.5%) 14.2% (1.2%) 8.7% (1.0%) SUD. In models that adjusted sociodemographic clinical characteristics other potential traumas, significantly MDD (adjusted odds ratio [aOR], 2.92; 95% CI, 1.74-4.88), (aOR, 2.59; 1.20-5.59), 1.86; 1.23-2.83), 1.91; 1.17-3.13), 2.06; 1.15-3.70). Conclusions Relevance combat-deployed soldiers, reports thoughts. Recognition may inform efforts prevent address health problems service members.

Язык: Английский

---

Associations of vulnerability to stressful life events with suicide attempts after active duty among high-risk soldiers: results from the Study to Assess Risk and Resilience in Servicemembers-longitudinal study (STARRS-LS)

Psychological Medicine, Год журнала: 2022, Номер 53(9), С. 4181 - 4191

Опубликована: Май 27, 2022

Abstract Background The transition from military service to civilian life is a high-risk period for suicide attempts (SAs). Although stressful events (SLEs) faced by transitioning soldiers are thought be implicated, systematic prospective evidence lacking. Methods Participants in the Army Study Assess Risk and Resilience Servicemembers (STARRS) completed baseline self-report surveys while on active duty 2011–2014. Two follow-up Longitudinal Surveys (LS1: 2016–2018; LS2: 2018–2019) were subsequently administered probability subsamples of these respondents. As detailed previous report, SA risk index based survey, administrative, geospatial data collected before separation/deactivation identified 15% LS respondents who had separated/deactivated as being self-reported post-separation/deactivation SAs. current report presents an investigation extent which SLEs occurring 12 months each survey might have mediated/modified association between this Results significantly elevated prevalence some SLEs. In addition, associations with SAs stronger among predicted than lower-risk Demographic rate decomposition showed that 59.5% ( s.e. = 10.2) overall subsequent was linked Conclusions It possible prevent substantial proportion providing targeted preventive interventions exposure/vulnerability commonly

Язык: Английский

---

Supporting servicemembers and veterans during their transition to civilian life using certified sponsors: A three-arm randomized controlled trial.

Psychological Services, Год журнала: 2023, Номер 20(Suppl 2), С. 248 - 259

Опубликована: Янв. 1, 2023

Transitioning servicemembers and veterans (TSMVs) face difficulties throughout their reintegration to civilian life, including challenges with employment, poor social connection, elevated risk for suicide. To meet the needs of this high-risk population, national initiatives have leveraged community-based interventions. Authors conducted a three-arm randomized controlled trial (

Язык: Английский

---