Bridging a translational gap: using machine learning to improve the prediction of PTSD

BMC Psychiatry

Arieh Y Shalev⁴  Zhiguo Li¹  Alexander Statnikov²  Isaac R Galatzer-Levy⁴  Karen-Inge Karstoft³ 

[1] Center for Health Informatics and Bioinformatics, NYU School of Medicine, New York, NY, USA;Department of Medicine, NYU School of Medicine, New York, NY, USA;Research and Knowledge Centre, Danish Veteran Centre, Garnisonen 1, Ringsted, 4100, Denmark;Department of Psychiatry, NYU School of Medicine, New York, NY, USA

关键词: Support vector machines;    Markov boundary feature selection;    Risk factors;    Early prediction;    Machine learning;    Posttraumatic Stress Disorder (PTSD);   

Others  :  1137822
DOI  :  10.1186/s12888-015-0399-8

 received in 2014-04-24, accepted in 2015-01-23,  发布年份 2015

【 摘 要 】

Background

Predicting Posttraumatic Stress Disorder (PTSD) is a pre-requisite for targeted prevention. Current research has identified group-level risk-indicators, many of which (e.g., head trauma, receiving opiates) concern but a subset of survivors. Identifying interchangeable sets of risk indicators may increase the efficiency of early risk assessment. The study goal is to use supervised machine learning (ML) to uncover interchangeable, maximally predictive combinations of early risk indicators.

Methods

Data variables (features) reflecting event characteristics, emergency department (ED) records and early symptoms were collected in 957 trauma survivors within ten days of ED admission, and used to predict PTSD symptom trajectories during the following fifteen months. A Target Information Equivalence Algorithm (TIE*) identified all minimal sets of features (Markov Boundaries; MBs) that maximized the prediction of a non-remitting PTSD symptom trajectory when integrated in a support vector machine (SVM). The predictive accuracy of each set of predictors was evaluated in a repeated 10-fold cross-validation and expressed as average area under the Receiver Operating Characteristics curve (AUC) for all validation trials.

Results

The average number of MBs per cross validation was 800. MBs’ mean AUC was 0.75 (95% range: 0.67-0.80). The average number of features per MB was 18 (range: 12–32) with 13 features present in over 75% of the sets.

Conclusions

Our findings support the hypothesized existence of multiple and interchangeable sets of risk indicators that equally and exhaustively predict non-remitting PTSD. ML’s ability to increase prediction versatility is a promising step towards developing algorithmic, knowledge-based, personalized prediction of post-traumatic psychopathology.

【 授权许可】

   
2015 Karstoft et al.; licensee BioMed Central.

【 图 表 】

Figure 1.

Figure 2.

Figure 3.

【 参考文献 】

• [1]Brewin CR, Andrews B, Valentine JD: Meta-analysis of risk factors for posttraumatic stress disorder in trauma-exposed adults. J Consult Clin Psychol 2000, 68(5):748-766.
• [2]Ozer EJ, Best SR, Lipsey TL, Weiss DS: Predictors of posttraumatic stress disorder and symptoms in adults: A meta-analysis. Psychol Bull 2003, 129(1):52-73.
• [3]Kessler RC, Sonnega A, Bromet E, Hughes M, Nelson C: Post-traumatic stress disorder in the National Comorbidity Survey. Arch Gen Psychiatry 1995, 52:1048-1060.
• [4]Marmar CR, Weiss DS, Schlenger WE, et al.: Peritraumatic dissociation and posttraumatic stress in male Vietnam theater veterans. Am J Psychiatry 1994, 151(6):902-907.
• [5]Shalev AY, Peri T, Canetti L, Schreiber S: Predictors of PTSD in injured trauma survivors: a prospective study. Am J Psychiatry 1996, 153(2):219-225.
• [6]Forbes D, Creamer M, Biddle D: The validity of the PTSD checklist as a measure of symptomatic change in combat-related PTSD. Behav Res Ther 2001, 39(8):977-986.
• [7]Bryant RA, Harvey AG, Guthrie RM, Moulds ML: A prospective study of psychophysiological arousal, acute stress disorder, and posttraumatic stress disorder. J Abnorm Psychol 2000, 109(2):341-344.
• [8]Difede J, Barocas D: Acute intrusive and avoidant PTSD symptoms as predictors of chronic PTSD following burn injury. J Trauma Stress 1999, 12(2):363-369.
• [9]Kleim B, Ehlers A, Glucksman E: Early predictors of chronic post-traumatic stress disorder in assault survivors. Psychol Med 2007, 37(10):1457-1467.
• [10]Mellman TA, David D, Bustamante V, Fins AI, Esposito K: Predictors of post-traumatic stress disorder following severe injury. Depress Anxiety 2001, 14(4):226-231.
• [11]Shalev AY, Peri T, Brandes D, Freedman S, Orr SP, Pitman RK: Auditory startle response in trauma survivors with posttraumatic stress disorder: a prospective study. Am J Psychiatry 2000, 157(2):255-261.
• [12]Segman RH, Shefi N, Goltser-Dubner T, Friedman N, Kaminski N, Shalev AY: Peripheral blood mononuclear cell gene expression profiles identify emergent post-traumatic stress disorder among trauma survivors. Mol Psychiatry 2005, 10(5):500. –513, 425
• [13]Koenen KC, Stellman JM, Stellman SD, Sommer JF Jr: Risk factors for course of posttraumatic stress disorder among Vietnam veterans: a 14-year follow-up of American Legionnaires. J Consult Clin Psych 2003, 71(6):980-986.
• [14]Shalev AY, Segman RH: Commentary: biological findings in PTSD – too much or too little? Prog Brain Res 2008, 167:187-199.
• [15]Orrù G, Pettersson-Yeo W, Marquand AF, Sartori G, Mechelli A: Using support vector machine to identify imaging biomarkers of neurological and psychiatric disease: a critical review. Neurosci Biobehav Rev 2012, 36(4):1140-1152.
• [16]Norman SB, Stein MB, Dimsdale JE, Hoyt DB: Pain in the aftermath of trauma is a risk factor for post-traumatic stress disorder. Psychol Med 2008, 38(4):533-542.
• [17]Bryan CJ, Clemans TA: Repetitive traumatic brain injury, psychological symptoms, and suicide risk in a clinical sample of deployed military personnel. JAMA Psychiatry (Chicago, Ill) 2013, 70(7):686-691.
• [18]Haagsma JA, Ringburg AN, van Lieshout EM, van Beeck EF, Patka P, Schipper IB: Prevalence rate, predictors and long-term course of probable posttraumatic stress disorder after major trauma: a prospective cohort study. BMC Psychiatry 2012, 12:236. BioMed Central Full Text
• [19]Bienvenu OJ, Gellar J, Althouse BM, Colantuoni E, Sricharoenchai T, Mendez-Tellez PA, et al.: Post-traumatic stress disorder symptoms after acute lung injury: a 2-year prospective longitudinal study. Psychol Med 2013, 43(12):2657-2671.
• [20]Statnikov A, Lytkin NI, Lemeire J, Aliferis CF: Algorithms for Discovery of Multiple Markov Boundaries. J Mach Learn Res 2013, 14:499-566.
• [21]Breiman L: Statistical modeling: the two cultures. Stat Sci 2001, 16(3):199-215.
• [22]Statnikov A, Aliferis CF: Analysis and computational dissection of molecular signature multiplicity. PLoS Comput Biol 2010, 6(5):e1000790.
• [23]Galatzer-Levy IR, Karstoft KI, Statnikov A, Shalev AY: Quantitative forecasting of PTSD from early trauma responses: a Machine Learning application. J Psychiatr Res 2014, 59:68-76.
• [24]Boser BE, Guyon IM, Vapnik VN: A training algorithm for optimal margin classifiers. In Proceedings of the fifth annual workshop on Computational learning theory (COLT '92). ACM, New York, NY, USA; 1992. 144-152. doi:10.1145/130385.130401
• [25]Galatzer-Levy IR, Ankri Y, Freedman S, Israeli-Shalev Y, Roitman P, Gilad M, et al. Early PTSD Symptom Trajectories: Persistence, Recovery, and Response to Treatment: Results from the Jerusalem Trauma Outreach and Prevention Study (J-TOPS). PLoS One. 2013;8(8):e70084.
• [26]Shalev AY, Ankri Y, Israeli-Shalev Y, Peleg T, Adessky R, Freedman S: Prevention of posttraumatic stress disorder by early treatment: results from the Jerusalem Trauma Outreach And Prevention study. Arch Gen Psychiatry 2012, 69(2):166-176.
• [27]Shalev AY, Ankri YL, Peleg T, Israeli-Shalev Y, Freedman S: Barriers to receiving early care for PTSD: results from the Jerusalem trauma outreach and prevention study. Psychiatr Serv 2011, 62(7):765-773.
• [28]Foa EB, Tolin DF: Comparison of the PTSD Symptom Scale-Interview Version and the Clinician-Administered PTSD scale. J Trauma Stress 2000, 13(2):181-191.
• [29]Kessler RC, Andrews G, Colpe LJ, Hiripi E, Mroczek DK, Normand SL, et al.: Short screening scales to monitor population prevalences and trends in non-specific psychological distress. Psychol Med 2002, 32(6):959-76.
• [30]Bryant RA, Moulds ML, Guthrie RM: Acute Stress Disorder Scale: a self-report measure of acute stress disorder. Psychol Assess 2000, 12(1):61-68.
• [31]Guy W: Clinical global impression scale. The ECDEU Assessment Manual for Psychopharmacology-Revised Volume DHEW Publ No ADM 76 1976, 338:218-222.
• [32]Foa EB, Ehlers A, Clark DM, Tolin DF, Orsillo SM: The Posttraumatic Cognitions Inventory (PTCI): Development and validation. Psychol Assess 1999, 11(3):303-314.
• [33]Lazarus RS, Folkman S: Stress, Appraisal, and Coping. Springer, New York; 1984.
• [34]Bryant RA, O'Donnell ML, Creamer M, McFarlane AC, Silove D: A Multisite Analysis of the Fluctuating Course of Posttraumatic Stress Disorder. JAMA psychiatry (Chicago, Ill.). Jun 19 2013:1–8
• [35]Statnikov A, Tsamardinos I, Brown LE, Aliferis CF: Causal Explorer: A Matlab Library of Algorithms for Causal Discovery and Variable Selection for Classification. In Challenges in Machine Learning Volume 2: Causation and Prediction Challenge. Edited by Guyon IM, Aliferis CF, Cooper GF. Microtome Publishing, Bookline, Massachusetts; 2010:267-278.
• [36]Chang C-C, Lin C-J: LIBSVM: A library for support vector machines. ACM Trans Intell Syst Technol 2011, 2(3):1-27.
• [37]Bradley AP: The use of the area under the ROC curve in the evaluation of machine learning algorithms. Pattern Recogn 1997, 30(7):15.
• [38]Shalev AY, Freedman S: PTSD following terrorist attacks: a prospective evaluation. Am J Psychiatry 2005, 162(6):1188-1191.
• [39]Freedman SA, Brandes D, Peri T, Shalev A: Predictors of chronic post-traumatic stress disorder. A prospective study. Br J Psychiatry 1999, 174:353-359.
• [40]Breslau N, Chilcoat HD, Kessler RC, Davis GC: Previous exposure to trauma and PTSD effects of subsequent trauma: results from the Detroit Area Survey of Trauma. Am J Psychiatry 1999, 156(6):902-907.
• [41]Koenen KC, Harley R, Lyons MJ, Wolfe J, Simpson JC, Goldberg J, et al.: A twin registry study of familial and individual risk factors for trauma exposure and posttraumatic stress disorder. J Nerv Ment Dis 2002, 190(4):209-218.
• [42]Shalev AY, Videlock EJ, Peleg T, Segman R, Pitman RK, Yehuda R: Stress hormones and post-traumatic stress disorder in civilian trauma victims: a longitudinal study. Part I: HPA axis responses. Int J Neuropsychopharmacol 2008, 11(03):365-372.
• [43]Binder EB, Bradley RG, Liu W, Epstein MP, Deveau TC, Mercer KB, et al.: Association of fkbp5 polymorphisms and childhood abuse with risk of posttraumatic stress disorder symptoms in adults. JAMA 2008, 299(11):1291-1305.
• [44]Boscarino JA, Erlich PM, Hoffman SN, Zhang X: Higher FKBP5, COMT, CHRNA5, and CRHR1 allele burdens are associated with PTSD and interact with trauma exposure: implications for neuropsychiatric research and treatment. Neuropsychiatr Dis Treat 2012, 8:131-139.
• [45]Visweswaran S, Angus DC, Hsieh M, Weissfeld L, Yealy D, Cooper GF: Learning patient-specific predictive models from clinical data. J Biomed Inform 2010, 43(5):669-685.