Sentiment analysis in medication adherence: using ruled-based and artificial intelligence-driven algorithms to...

Area of Science:

• Artificial Intelligence in Pharmacy
• Natural Language Processing Applications
• Computational Linguistics in Healthcare

Background:

• Sentiment analysis (SA) is an underutilized AI innovation in pharmacy for assessing patient feelings towards medications.
• SA utilizes artificial intelligence and natural language processing to analyze text for emotions and sentiments.

Purpose of the Study:

• To evaluate Valence Aware Dictionary for Sentiment Reasoning (VADER) and Emotion English DistilRoBERTa-base (DistilRoBERTa) models.
• To identify patient sentiments and emotions concerning pharmacotherapy using SA.

Main Methods:

• Utilized a dataset of 320,095 anonymized patient medication experiences.
• VADER analyzed sentiment polarity (-1 to +1); DistilRoBERTa classified emotions into seven categories.
• Employed scikit-learn's Python module for performance metrics.

Main Results:

• VADER achieved 0.70 overall accuracy, with strong recall for negative sentiments (0.80) and good precision for positive sentiments (0.73).
• DistilRoBERTa revealed positive emotions correlated with medication effectiveness, ease of use, and satisfaction.
• Both models demonstrated consistent and reliable results in analyzing patient pharmacotherapy sentiments.

Conclusions:

• VADER and DistilRoBERTa successfully analyzed patient sentiments and emotions towards pharmacotherapy.
• Findings support the integration of SA in clinical pharmacy for enhanced patient care.
• This research paves the way for more personalized and effective patient management strategies.