Psychosis presents many challenges in determining the diagnosis and identifying specific biomarkers and treatments, especially in its early stages. Specifically, diagnoses are often blurred at presentation, and subgroups of individuals do not achieve a symptomatic and/or functional recovery, despite receiving specialized care. A precise definition of outcomes and response to treatments assessed readily on presentation would certainly impact patient management. Within this context, machine learning techniques and automatic classification methods are promising tools in data analysis and clinical practice. These techniques have the potential to learn subtle data attributes, analyze massive data sets and combine different modalities (e.g., neuroimaging, clinical data, genetics, etc.) with the final aim to inform clinical decisions associated with diagnosis, prognosis, and treatment in psychiatry.
Our goal is to examine how modern computational techniques could contribute to clinical practice in addressing the unmet needs of psychosis patients in aspects such as differential diagnoses, treatment, and outcome prediction. The outcome is uncertain, especially at the presentation or in the early stages of the disease. Moreover, psychosis can present different subtypes (e.g., affective/non-affective psychosis) that necessitate different treatments. Unfortunately, tools that help clinicians define diagnoses and, more importantly, prognoses are still lacking. Besides, a tool to inform clinicians in choosing the right treatment would be of massive help, given that a substantial percentage of patients do not respond to treatment and often try different types of drugs.
Automated analytical model-building techniques such as machine learning could improve psychotic patients’ management. However, the critical aspect of the feasibility of clinical deployment must be considered: these algorithms might prove useful but need to be implementable in a real-life clinical environment to have a real impact on patients’ health. A translational approach is therefore mandatory. Moreover, to be translated into a clinical setting, an algorithm should be precise and trusted by clinicians. Thus, understanding how well machine learning algorithms translate from a research framework to a real-world healthcare system is essential.
We will welcome original research articles, methodological articles and systematic review/meta analyses conducted with machine learning techniques in samples of adolescents, adults, elderly patients with psychosis or at high clinical risk for psychosis, covering the following topics:
• Challenges in determining standard and differential diagnosis
• Clinical and biological predictors of prognostic trajectories
• Biomarkers and predictors of differential treatment response
• Neuroimaging studies using brain features to predict critical outcomes (e.g., suicide)
• Genetic studies proposing novel predictors of disease progression and conversion
• Neuropsychological investigations, including cognitive markers and neurocognitive deficits as predictors of response to diverse treatment modalities or prognostic trajectories
Psychosis presents many challenges in determining the diagnosis and identifying specific biomarkers and treatments, especially in its early stages. Specifically, diagnoses are often blurred at presentation, and subgroups of individuals do not achieve a symptomatic and/or functional recovery, despite receiving specialized care. A precise definition of outcomes and response to treatments assessed readily on presentation would certainly impact patient management. Within this context, machine learning techniques and automatic classification methods are promising tools in data analysis and clinical practice. These techniques have the potential to learn subtle data attributes, analyze massive data sets and combine different modalities (e.g., neuroimaging, clinical data, genetics, etc.) with the final aim to inform clinical decisions associated with diagnosis, prognosis, and treatment in psychiatry.
Our goal is to examine how modern computational techniques could contribute to clinical practice in addressing the unmet needs of psychosis patients in aspects such as differential diagnoses, treatment, and outcome prediction. The outcome is uncertain, especially at the presentation or in the early stages of the disease. Moreover, psychosis can present different subtypes (e.g., affective/non-affective psychosis) that necessitate different treatments. Unfortunately, tools that help clinicians define diagnoses and, more importantly, prognoses are still lacking. Besides, a tool to inform clinicians in choosing the right treatment would be of massive help, given that a substantial percentage of patients do not respond to treatment and often try different types of drugs.
Automated analytical model-building techniques such as machine learning could improve psychotic patients’ management. However, the critical aspect of the feasibility of clinical deployment must be considered: these algorithms might prove useful but need to be implementable in a real-life clinical environment to have a real impact on patients’ health. A translational approach is therefore mandatory. Moreover, to be translated into a clinical setting, an algorithm should be precise and trusted by clinicians. Thus, understanding how well machine learning algorithms translate from a research framework to a real-world healthcare system is essential.
We will welcome original research articles, methodological articles and systematic review/meta analyses conducted with machine learning techniques in samples of adolescents, adults, elderly patients with psychosis or at high clinical risk for psychosis, covering the following topics:
• Challenges in determining standard and differential diagnosis
• Clinical and biological predictors of prognostic trajectories
• Biomarkers and predictors of differential treatment response
• Neuroimaging studies using brain features to predict critical outcomes (e.g., suicide)
• Genetic studies proposing novel predictors of disease progression and conversion
• Neuropsychological investigations, including cognitive markers and neurocognitive deficits as predictors of response to diverse treatment modalities or prognostic trajectories
