S136. A NOVEL APPROACH FOR DEVELOPING PREDICTION MODEL OF TRANSITION TO PSYCHOSIS: DYNAMIC PREDICTION USING JOINT MODELLING. (1st April 2018)

• Record Type: Journal Article
• Authors: Yuen, Hok Pan; Mackinnon, Andrew; Hartmann, Jessica; Amminger, Paul; Markulev, Connie; Lavoie, Suzie; Schafer, Miriam; Polari, Andrea; Mossaheb, Nilufar; Schlogelhofer, Monika; Smesny, Stefan; Hickie, Ian; Berger, Gregor; Chen, Eric; de Hann, Lieuwe; Nieman, Dorien; Nordentoft, Merete; Riecher-Rössler, Anita; Verma, Swapna; Thompson, Andrew; Yung, Alison; McGorry, Patrick; Nelson, Barnaby
• Is Part Of: Schizophrenia bulletin. Volume 44(2018)Supplement 1
• Page Start: S378
• Page End: S379
• Publication Date: 2018-04-01
• Subjects: Schizophrenia -- Periodicals; Schizophrenia -- Research -- Periodicals; 616.898005
• DOI: 10.1093/schbul/sby018.923
• Languages: English
• ISSNs: 0586-7614

Abstract

Abstract: Background: Ever since the establishment of strategies for identifying people at ultra-high risk (UHR) of developing psychosis about twenty years ago, much research has been conducted in seeking risk factors and in developing prediction models for predicting which UHR individuals will actually make a transition to psychosis. The goal is to provide specific interventions to those of high susceptibility. Such research almost invariably uses fixed predictor variables, typically variables assessed at baseline, i.e. service entry. Interest has now emerged to investigate whether the dynamic nature of psychopathology can be used to improve prediction of the onset of psychosis. As studies on UHR individuals usually require follow-up of participants over time, the longitudinal nature of these studies provides the opportunity to capture the dynamic characteristics of psychopathology by conducting multiple assessments across the study period. The idea is that prediction can be updated continuously as more information about changes in patients' conditions are obtained. Over the past two decades, statistical methodology that can combine the time-to-transition aspect and the longitudinal aspect of UHR studies into one model has emerged. The methodology is called joint modelling. Methods: The aim is to describe the joint modelling methodology and to demonstrate how joint modelling can be used to develop a prediction model for transition to psychosis. The data from the NEURAPRO Study was used for the demonstration. This study was a multi-centre placebo-controlled randomized trial of the effect of omega-3 polyunsaturated fatty acids on transition risk in UHR individuals. The sample size was 304. Study assessments were conducted monthly during the first 6 months and then at months 9 and 12. There were in total 40 known transitions. Results: Compared with the conventional approach of using only fixed predictors, joint modelling prediction models showed significantly better sensitivity, specificity and likelihood ratios. Discussion: Joint modelling is a useful statistical tool which can improve the prediction of the onset of psychosis and has the potential in guiding the provision of timely and personalized treatment to patients concerned.