I am curious to read this to see if I can learn anything. If they have a model to predict CHR patients, then maybe that model can be used to predict Psychosis relapse? RELATED TO #hope-s : I am thinking in the future, maybe the population with the most use will be patients who CHR, to prevent conversion to Psychosis? With TECC ? [[Technology Enable Clinical Care (TECC)]] ---- Aim of the study is to gather large number of Clinical High Risk patients, follow up with them for two years. Then come out with a prediction model for conversion. 1/3 of the population convert to psychosis. **What is "Dynamic Prediction" - i.e Time-dependent or time-variable.** - They also use "Time Series Analysis" - Which is the use of EMA. - *"Time Series analysis involves analysis of a series of data points indexed in time order. Patterns in the data that emerge over time can be used for the purpose of forecasting future values. This approach has been more widely deployed in other areas of psychiatry, e.g depression and transdiagnostic prediction, than in psychosis prediction. In these other areas, particular temporal patterns in the data, including increase in emotional state correlating with itself over time (temporal autocorrelation) and increased variance and change in the association between emotions over time, have proven valuable as 'early warning signs' of an imminent change in mental state (e.g 'Tipping point' into a depressive episode.) In the area of psychosis and psychosis risk, the focus has instead been on the proximal relationship between everyday events/ contexts and momentary fluctuations in mental state, such as stress sensitivity assessed using ecological momentary assessment (EMA), rather than on prediction of clinical outcomes."* - Page 24 in Protocol V2. **Joint Modelling - Repeat the test at monthly as a way to test?** **Network Theory -- This is relevant to the HOPES study.** *Network Theory posits that symptoms are not all explained by a shared underlying cause as in the traditional latent disease model (e.g lunch cancer being the underlying common cause of various symptoms such as shortness of breath, chest pain, and coughing up blood). Rather, mental disorders are seen as complex dynamic systems in which symptoms and psychological, biological and social components have autonomous causal power to influence and trigger each other. If symptoms form patterns of mutual reinforcement and feedback loops, the system as a whole may become trapped or 'locked' in a state of extended symptom activation, a point at which a mental disorder may be diagnosed (e.g conversion to psychosis). Network analysis models the pairwise relationship ('edge') between symptoms ('nodes) cross-sectionally and dynamically over time (using time series), showing interactions between symptoms. It is plausible that this 'locked' state is more likely to occur in biologically vulnerable people; however this process may be interrupted by psychosocial and biological early intervention strategies. These possibilities can be assessed in the current dataset by examining interactions between data modalities over time (e.g Digital Momentaary Assessments and Biomarkers) as well as modelling impact of treatment exposure.* - Page 25 Multimodal Prediction - - whats that? - Language production (e.g indices of semantic coherence and syntactic complexity) has shown promises as a marker of psychosis risk.) This study uses [[Technology Enable Clinical Care (TECC)]] mindLAMP, but it's only an optional component. ([[What is the mindLAMP platform]])