Computer-aided therapeutic diagnosis for anorexia

Area of Science:

• Psychiatry
• Computational Linguistics
• Artificial Intelligence

Background:

• Anorexia nervosa etiology is not fully understood, necessitating effective diagnostic tools.
• Clinical diagnosis involves identifying causes like body image issues, food, family, and peer influences.
• Computer-aided tools are recommended to assist psychologists in diagnosis and treatment planning.

Purpose of the Study:

• To present a novel computer-aided therapeutic diagnosis method for anorexia nervosa.
• To enhance the diagnostic process for anorexia nervosa using automated analysis.
• To facilitate timely intervention and improve patient prognosis.

Main Methods:

• The proposed method involves four stages: patient free statements, Recurrent Neural Network (RNN) sentiment analysis, basic emotion intensity assessment (Nencki Affective Word List), and dictionary-based sentiment analysis for difficulty areas.
• Utilized RNN for general sentiment analysis and a dictionary approach for specific emotion and difficulty assessment.
• Employed natural language processing techniques for analyzing patient-reported data.

Main Results:

• Sentiment analysis achieved 72% effectiveness with RNN and 51% with the dictionary-based method.
• A significant 10% difference in sadness intensity was observed between control and research groups using the dictionary method.
• The study demonstrates the potential of automated analysis in quantifying patient-reported experiences.

Conclusions:

• Rapid sentiment analysis of patient statements can expedite anorexia nervosa diagnosis and therapy initiation.
• The automated method may reduce patient aversion to therapy, improving communication and engagement.
• This approach offers a promising avenue for improving treatment outcomes and prognosis in anorexia nervosa.