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

Abstract

Introduction: Understanding patients' everyday experience is essential to improve patient centered care in sarcoidosis. So far, patient perspectives are based on survey- and qualitative research.

Aim: We aimed to assess patient-driven perspectives on their care trajectories using a novel machine learning-driven approach (MLD).

Methods: We used the largest Dutch sarcoidosis patient platform as the data source of patient stories. The patients' stories were extracted with permission. We applied topic modelling (to generate topics among the posts), and sentiment analysis (to find tone of voice in the topics). To validate the findings, we read the top 50 most relevant posts of each topic. An in-depth patients' disease trajectory map was made.

Results: Based on 4969 forum posts, 30 final topics and 10 upper themes were generated, which formed the basis for the "patient journey-map" which shows patients' perspective across the care pathway. Important decision moments could be identified, as well as care "tracks" at home and hospital and topics associated with positive or negative emotions. Most patients' perspectives were about symptoms (mainly negative sentiment), disease-modifying medication (mainly neutral sentiment), and quality of life (negative, neutral and positive).

Discussion: A major part of living with sarcoidosis takes place outside the view of the hospital, but this part often remains invisible. MLD is an innovative approach, providing a comprehensive overview of patients' perspectives on health and care. Integrating, these findings in the design of health care delivery has the potential to improve patient-centered care.