Identifying Language Features Associated With Needs of Ovarian Cancer Patients and Caregivers Using Social Media

Area of Science:

• Oncology
• Natural Language Processing
• Health Informatics

Background:

• Online health communities (OHCs) offer insights into cancer patient and caregiver needs.
• Ovarian cancer (OvCa) patients express diverse unmet needs within OHCs.
• Automated classification models can help clinicians process OHC information.

Purpose of the Study:

• To develop an automated model for classifying ovarian cancer patient and caregiver needs using initial online postings.
• To assess the accuracy of a machine learning model in identifying specific patient needs.

Main Methods:

• Collected and analyzed 853 initial postings from an OvCa OHC.
• Used two annotators to code postings for 12 types of needs.
• Applied a machine learning approach with bag-of-words features to build and evaluate the classification model using F1 score.

Main Results:

• The most frequently reported needs were information, social, psychological/emotional, and physical.
• The model achieved high accuracy in classifying these top needs.
• Psychological terms correlated with psychological/emotional and social needs; medical terms correlated with physical and information needs.

Conclusions:

• Automated classification of OHCs shows potential for supplementing traditional cancer patient needs assessment.
• The developed model can be adapted for other cancer types and improved with domain-specific data.