Background/Purpose: Lupus commonly affects women of reproductive age whose pregnancies are associated with higher rates of complications and fetal loss. There is a knowledge gap about pregnancy among lupus patients. Patients often turn to social media for health-related information, experiences and might feel more comfortable discussing their fears and concerns online than in the doctor’s office. Sentiment analysis of social media posts can help better understand the unmet needs when it comes to pregnancy care in lupus, help model comprehensive discussions in the clinic with patients of reproductive age, spread awareness, disseminate right educational content through social media to decrease the knowledge gap.

Aim of this study was to apply an artificial intelligence–based approach to analyze public sentiments on social media toward pregnancy in lupus to better understand the public attitude and concerns.

Methods: Analysis was conducted on 5000 posts of over 42K members in the public Facebook group ‘lupus & autoimmune disease awareness and sharing’. Posts about pregnancy were identified using the keywords pregnancy, pregnant. Posts without questions, experiences or information about pregnancy were disregarded. Data transformation, cleaning, sentiment determination (SI score), and analysis was carried out using R package syuzhet, NRC lexicon and free online tool danielsoper.com. Independent t-tests were used to compare the average positive and negative sentiment scores. Broad themes of the posts were identified manually.

Results: There were 295 pregnancy-related posts between January 2020-June 2022. 124 posts met the criteria: 70 (56%) with positive (SI score 96+/-13), 50 (40%) negative sentiment (SI score 96+/-9) (p >0.05) and 4 (3%) neutral sentiment. Net sentiment ratio [(number of positive – number of negative)/(number of total)] was 16% indicating overall positive sentiment toward pregnancy.

The predominant positive sentiments were trust, joy and anticipation. Negatives were additionally classified as fear, sadness, anger and disgust. The broad themes that emerged were questions about chances of conception, general information about pregnancy planning, pregnancy outcomes, flares, medication use, treatment of various symptoms arising during pregnancy. 34% of posts said they received no pre-pregnancy guidance from physicians. 61% of posts expressed the need for longer discussion with the physician regarding pregnancy.

Conclusion: There were 40% of posts associated with negative sentiment toward pregnancy in lupus including fear and sadness. Knowledge gaps were identified in the areas of pregnancy planning and medication management. Preconception counseling is identified as an unmet need and possible intervention in outpatient clinics to improve pregnancy outcomes.

« Back to ACR Convergence 2022

ACR Meeting Abstracts - https://acrabstracts.org/abstract/artificial-intelligence-enabled-sentiment-analysis-of-public-attitude-toward-pregnancy-in-lupus-social-media-group/