Innovating care for people with sarcoidosis using a machine learning-driven approach.

Area of Science:

• Digital Health
• Patient-Reported Outcomes
• Computational Social Science

Background:

• Understanding the patient's everyday experience is crucial for enhancing patient-centered care in sarcoidosis.
• Current patient perspectives are primarily derived from traditional survey and qualitative research methods.
• The daily realities of living with sarcoidosis often remain outside the scope of clinical observation.

Purpose of the Study:

• To employ a novel machine learning-driven approach (MLD) to assess patient-driven perspectives on their sarcoidosis care trajectories.
• To gain deeper insights into the patient journey beyond conventional research methodologies.
• To identify key decision points and emotional tones within patient narratives.

Main Methods:

• Utilized the largest Dutch sarcoidosis patient platform, extracting patient stories with consent.
• Applied topic modeling to identify key themes and sentiment analysis to gauge emotional tone within patient narratives.
• Validated findings through manual review of top posts and constructed an in-depth patient disease trajectory map.

Main Results:

• Analysis of 4969 forum posts yielded 30 topics and 10 themes, forming a comprehensive patient journey map.
• Identified critical decision points, distinct care pathways (home and hospital), and associated emotional sentiments.
• Patient perspectives predominantly focused on symptoms (negative sentiment), disease-modifying medication (neutral sentiment), and quality of life (mixed sentiment).

Conclusions:

• A significant portion of the sarcoidosis patient experience occurs outside of hospital settings and is often overlooked.
• Machine learning-driven approaches offer an innovative method for a holistic understanding of patient health and care perspectives.
• Integrating these patient-driven insights into healthcare delivery design can significantly improve patient-centered care for sarcoidosis.