Using Speech to Monitor Symptom Severity in Arabic Speaking Patients With Schizophrenia

Speech disorganization is a key feature of schizophrenia. The development of computerized tools to assess speech disorganization is rapidly growing in schizophrenia research. Several early studies showed that changes in speech distinguish schizophrenia patients from healthy controls and assist in differential diagnostics and relapse prevention (1). The Winterlight app can be used for speech collection and assessment and uses speech-based artificial intelligence to identify vocal biomarkers capable of detecting changes in cognitive/clinical symptoms. Symptom rating scales remain the primary mode of assessing the nature and severity of schizophrenia and the magnitude of any change over time. The Positive and Negative Symptom Scale (PANSS) is a 30-item rating scale that was developed to measure the symptom severity of patients with schizophrenia and assess their dimensions (2). It has been widely used in clinical trials of schizophrenia and is considered as the "gold standard" for the assessment of antipsychotic treatment efficacy. The goal of this study is to test the hypothesis that quantitative measures derived from speech samples acquired using the Winterlight application will be associated with positive and negative symptom subscores as assessed by the PANSS. The investigators will use speech-based artificial intelligence methods to identify aspects of voice and language that are related to schizophrenia symptoms in Arabic-speaking patients. Data collected may be used to evaluate: 1. The relationship between speech measures and PANSS subscores at baseline. 2. The relationship between changes in speech measures and changes in positive symptoms over time. 3. The relationship between changes in speech measures and changes in negative symptoms over time. 4. The ability for speech measures to be used to predict psychotic relapse in individuals with schizophrenia. 5. The feasibility of predicting relapse based on speech and sociodemographic variables.