IEPA 11 has ended
Back To Schedule
Monday, October 8 • 2:50pm - 3:10pm
Symposium 6, Talk 1. "A Novel Approach to Developing A Prediction Model of Transition to Psychosis: Dynamic Prediction Using Joint Modelling"

Sign up or log in to save this to your schedule, view media, leave feedback and see who's attending!

Feedback form is now closed.
Hok Pan Yuen1, Andrew Mackinnon2, Patrick McGorry1, G. Paul Amminger1, Jessica Hartmann1, Miriam Schäfer 
1, Connie Markulev1, Suzie Lavoie1, Barnaby Nelson1; 1Orygen, The National Centre of Excellence in Youth Mental Health, University of Melbourne, 2Centre for Mental Health, Melbourne School of Population and Global Health, University of Melbourne
Introduction Joint modelling (JM) is a promising new statistical methodology which can use data from both study entry and subsequent follow-up assessments to develop prediction models of psychosis onset in ultra-high risk (UHR) individuals. The practical implication is that the prediction provided by these models could be updated as new information about patients’ clinical state is obtained and appropriate treatment could be implemented accordingly. This study aimed to test the potential benefits of joint modelling to dynamically predict the onset of psychosis in UHR individuals. 
 Method Data from the NEURAPRO intervention study was used. This study was a multi-centre placebo-controlled randomized trial of the effect of omega-3 polyunsaturated fatty acids on risk of transition to psychotic disorder 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 cases of transition to psychosis. Candidate predictor variables consisted of demographic characteristics assessed at intake as well as repeated measurements of clinical variables. Results Compared with the conventional approach of using only baseline data for prediction of psychosis, JM prediction showed significantly better sensitivity, specificity and likelihood ratios. The JM approach yielded sensitivity of 82.8%/specificity 72.4%, whereas the baseline-data only model yielded sensitivity of 69.0%/specificity 73.8%. 
 Conclusions Incorporating time-dependent variables into predictive models has the potential to improve the prediction of onset of psychosis and hence to help in providing timely and personalized treatment to patients.


Monday October 8, 2018 2:50pm - 3:10pm EDT
American Ballroom-Center Westin Copley Place, fourth floor

Attendees (9)