Cognitive Impairments in Schizophrenia: A Study in a Large Clinical Sample Using Natural Language Processing.

Area of Science:

• Psychiatry
• Computational Linguistics
• Medical Informatics

Background:

• Cognitive impairments significantly impact schizophrenia outcomes but are often overlooked.
• Traditional assessment methods are resource-intensive and not routinely used.
• Natural language processing (NLP) offers a scalable solution for extracting this data.

Purpose of the Study:

• To identify cognitive problems in schizophrenia patients using NLP on clinical records.
• To assess the association between these cognitive issues and clinical outcomes.
• To evaluate the NLP tool's performance against a standard rating scale.

Main Methods:

• Developed an NLP application to detect cognitive dysfunctions from free-text medical records.
• Validated the NLP tool against the Health of the Nation Outcome Scales (HoNOS) cognitive subscale.
• Analyzed cognitive trajectories, socio-demographic factors, and clinical outcomes.

Main Results:

• High prevalence of cognitive impairments in schizophrenia patients was observed.
• NLP-derived findings correlated strongly with socio-demographic factors and adverse outcomes.
• NLP results aligned with HoNOS, revealing additional associations with attention and social cognition.

Conclusions:

• Cognitive problems in schizophrenia are prevalent and reliably detectable via NLP.
• NLP extraction from clinical text offers broader coverage and richer data than traditional methods.
• NLP tools can enhance large-scale screening, early symptom detection, and clinical decision-making.