Improving psychosis risk detection through artificial intelligence

Research project description

PhD will investigate intervention in youth at clinical high risk (CHR) for psychosis.  

Early intervention in youth at clinical high risk (CHR) for psychosis is highly effective in preventing, delaying, or mitigating the impact of a chronic psychotic disorder. It is important to note that not all youth at CHR (approximately ⅓) will progress to psychosis, which poses a challenge for clinicians in determining appropriate interventions for each individual and regularly results in treatment delays of well over one year.  

Help-seeking youth in the prodromal phase of a psychotic disorder can only receive tailored treatment once a diagnosis has been confirmed. If there is a delay in diagnosis, clinicians lose their ability to intervene early to ameliorate social, functional, and cognitive outcomes. Clinicians at mental health evaluation centers are concerned because the only tool they have to attempt to make such a confirmation prior to the first episode of psychosis has a specificity of only 30% (Miller et al., 2003). Researchers have tried to improve this number by demonstrating that The Story Game (Caplan and Sherman, 1989), which measures illogical thinking and language cohesion, can predict conversion to psychosis with a specificity of 70% (Bearden et al., 2011). However, that approach never made it into clinical practice. One root cause for that failure is that The Story Game takes approximately four hours of clinician time to administer and score. Accordingly, there is a need for a faster way to deliver and score The Story Game. If solved, this would improve specificity of detection of psychotic disorders from 30% to 70%. Likewise, it is expected that adequate detection of the risk of psychosis will allow achieving better indicators of well-being. 

This proposal intends to build a scalable version of The Story Game that accomplishes automated measurement of illogical thinking and language cohesion. Responses to automated prompts will be collected via a simple microphone, a speech-to-text algorithm will then translate this into written text, and the resulting text will be processed using natural language processing (NLP) methods. This will result in a rapid diagnostic aid for clinicians at early psychosis centers who diagnose symptoms of adolescent psychological distress. 

Academic background / Skills

Candidates must hold a degree that allows admission to the official doctoral programme at UAB.  

Additional requirements for a stronger application are: 

• Bachelor’s Health Sciences Degree
• Demonstrable experience in research methodology (clinical, epidemiological trials, cohorts, etc.) 
• Intermediate-High level of English, in reading comprehension level and with the ability to write and communicate in a formal register. 
• Mastery skills in Microsoft Office tools, especially Word, Excel and Power Point 
• Mastery skills in SPSS statistical package. 
• Knowledge of databases elaboration and management (elaborating and defining variables, debugging databases, etc.) 
• Prior experience with programming languages relevant to data analysis and research (e.g. Python) will be considered. 
• Capacity for teamwork  

Research group/s description

Research Group of Cognition and gender: implications for mental health is a multidisciplinary research group, which is made up of 7 psychologists, 3 psychiatrists and 1 metodologist/statistician. 

In this proposal we will participate with subline COGnition & gENder: implications in PSYchotic spectrum disorders (COGen-PSY subgroup), which focuses on the characterization of psychotic spectrum disorders and their subclinical forms.

Recently, the principal investigator of this subline (Dr. Ana Barajas) has started a collaboration with the research group lead to Dr. Enrique Gutiérrez, the Data Science group at MATIC in the Polytechnic University of Madrid in a project based on the improve of prediction of risk of psychosis using Artificial Intelligence Techniques such as Natural Language Processing.

This collaboration is both robust and dynamic, with aspirations to broaden the research to encompass national and international projects. This expansion includes a commitment to enhancing the application of these technologies and methodologies to advance mental health diagnostics and therapeutic approaches.

Both Dr. Gutiérrez and Dr. Barajas are integral to the progress of discussions within the research team. Their roles in this interdisciplinary collaboration are significant, merging clinical expertise with technological advancements. Their involvement in ongoing training sessions led by Dr. Rochelle Caplan, who is recognized for her linguistic assessment protocol for youth at risk of psychosis, highlights the depth of this partnership.