Listening and Learning from the Voices of Patients with Myelodysplastic Syndromes and Their Caregivers: A Comprehensive Social Media Analysis Using Machine Learning and Natural Language Processing-Based Algorithms
Author(s)
Marwah R1, Mishra S2, Jana A1, Gross B1, Couturiaux S1, Calara R3, Sabate E3, Goldmann E4, Hogea C3
1Definitive Healthcare, Framingham, MA, USA, 2Definitive Healthcare, Boston, MA, USA, 3Gilead Sciences Inc, Foster City, CA, USA, 4New York University, New York, NY, USA
OBJECTIVES: This study aims to develop an agile reusable framework to collect and analyze evolving social media data to systematically capture insights into patients and caregivers' lived experiences related to myelodysplastic syndromes (MDS). Discerning unmet needs of individuals affected by MDS and their caregivers can help inform development of appropriate patient support tools. METHODS: An extensive Google search was performed for English-language websites relevant to MDS using validated URLs and keywords, followed by scraping algorithms to gather, clean, and standardize relevant information posted from 2008 to 2022. We deployed advanced algorithms such as Latent Dirichlet Allocation, Latent Semantic Analysis, BERT, Hierarchical Clustering, and Negative Matrix Factorization to analyze the underlying semantic and temporal structure of the data. RESULTS: The data collected comprised approximately 5.5 million words from 42,000 posts across 5500 threads, involving about 4000 users predominantly from the US, UK, and Canada. Forty-nine percent were classified as patients, 42% as caregivers, and 9% as others. User engagement, measured as the number of days users were active (posting), ranged from 2 to 1000+ days (median, 10 days) with 1% considered superusers (top 2 percentile users by activity). Common emotions included worry (17%), frustration (16%), and hope (13%) stemming from the potential severity of diagnoses and treatment, but also hope for better prognosis. We uncovered diverse topics encompassing voices of patients and caregivers around disease management, access to treatment, communication with healthcare providers, impact on quality of life, and lack of knowledge resources, compounding to latent emotions such as anxiety and stress. CONCLUSIONS: Uncovering and mapping underlying themes with sentiments along the MDS patient journey can inform areas of need for patient-centered care and development of patient-focused solutions with the potential to improve the patient experience. Ongoing exploratory research includes focus on higher-risk MDS patients and specific related issues.
Conference/Value in Health Info
2023-11, ISPOR Europe 2023, Copenhagen, Denmark
Value in Health, Volume 26, Issue 11, S2 (December 2023)
Code
PCR134
Topic
Economic Evaluation, Methodological & Statistical Research, Patient-Centered Research
Topic Subcategory
Artificial Intelligence, Machine Learning, Predictive Analytics, Novel & Social Elements of Value, Patient Engagement, Patient-reported Outcomes & Quality of Life Outcomes
Disease
No Additional Disease & Conditions/Specialized Treatment Areas, Oncology
Explore Related HEOR by Topic