Artificial Intelligence-Based Co-Facilitator (AICF) for Detecting and Monitoring Group Cohesion Outcomes in Web-Based...

Area of Science:

• Computational linguistics
• Artificial intelligence
• Online health support

Background:

• Online support groups (OSGs) offer cost-effective cancer care.
• Monitoring group cohesion in text-based OSGs is challenging due to limited nonverbal cues.
• The Artificial Intelligence-based Co-Facilitator (AICF) was developed to analyze conversations and identify therapeutic outcomes.

Purpose of the Study:

• To develop and evaluate a method for training the AICF to monitor group cohesion in OSGs.
• To assess the AICF's capability in detecting group cohesion within online cancer support group conversations.

Main Methods:

• Utilized a text classification approach for extracting group cohesion mentions.
• Trained the AICF model using human-annotated messages from Cancer Chat Canada.
• Employed word embedding models to identify similar expressions of group cohesion.
• Compared AICF performance against the Linguistic Inquiry Word Count (LIWC) software.

Main Results:

• AICF was trained on 80,000 messages and tested on 34,048 messages.
• Human experts validated AICF's ability to classify group cohesion with an F1-score of 0.82 after retraining.
• AICF demonstrated slightly superior performance in identifying group cohesion compared to LIWC.

Conclusions:

• Machine learning combined with human input can effectively detect group cohesion in OSGs.
• AICF can assist therapists by identifying group dynamics for real-time intervention.
• AICF enhances patient-centered care in web-based support settings.