Natural Language Processing Applications in the Clinical Neurosciences: A Machine Learning Augmented Systematic...

Area of Science:

• Artificial Intelligence
• Clinical Neurosciences
• Natural Language Processing

Background:

• Natural language processing (NLP) applications in medicine are widespread, yet limited in clinical neurosciences.
• Recent advancements in deep transformer models (e.g., XLNet, BERT, T5, RoBERTa) and transfer learning have matured NLP capabilities.
• The clinical neurosciences field can benefit from NLP's potential for automating tasks and analyzing complex data.

Purpose of the Study:

• To systematically review existing NLP applications within the clinical neurosciences.
• To explore NLP techniques for literature synthesis and data analysis in this field.
• To demonstrate the practical potential of NLP for a clinical audience.

Main Methods:

• Systematic literature review of 48 articles meeting inclusion criteria.
• Application of NLP for keyword identification, text summarization, and document classification.
• Utilized deep transformer models including XLNet, BERT, T5, and RoBERTa.

Main Results:

• NLP has been applied in clinical neurosciences for literature synthesis, data extraction, patient identification, automated reporting, and outcome prediction.
• A significant increase in NLP publications in this field over the last five years was observed.
• Document classifiers showed moderate performance (XLNet AUC=0.66, BERT AUC=0.59, RoBERTa AUC=0.62), with T5 providing acceptable abstract summaries.

Conclusions:

• NLP applications are growing in clinical neurosciences, offering valuable tools for research and practice.
• NLP facilitates literature synthesis, data extraction, and outcome prediction, enhancing efficiency.
• Further exploration and implementation of NLP technologies can significantly advance clinical neuroscience research and patient care.