Automatically classifying question types for consumer health questions.

Area of Science:

• Computational linguistics
• Health informatics
• Artificial intelligence

Background:

• Consumer health questions are crucial for accessing medical information.
• Existing methods for classifying medical questions lack accuracy for consumer queries.
• Automatic retrieval of answers from health resources requires effective question classification.

Purpose of the Study:

• To develop and evaluate a machine learning-based method for automatically classifying consumer health questions.
• To introduce thirteen distinct question types for improved medical answer retrieval.
• To address the limitations of previous approaches in consumer health question classification.

Main Methods:

• Developed a machine learning model for classifying consumer health questions.
• Defined thirteen specific question types tailored for health queries.
• Identified and annotated three key question elements to enhance classification accuracy.
• Compared the proposed method against previous techniques for medical question classification.

Main Results:

• The proposed method represents a novel machine learning approach for consumer health question classification.
• Previous methods were found insufficient for achieving high accuracy on this specific task.
• Manual and automatic classification of three question elements improved overall classification performance.
• Analysis highlighted the inherent difficulties and areas for future research in this domain.

Conclusions:

• The developed machine learning method offers a significant advancement in classifying consumer health questions.
• The identified question elements are crucial for improving classification accuracy.
• Further research is needed to overcome the challenges and achieve high-performance consumer health question classification systems.