SHM Converge Scientific Abstract ePoster Gallery-Patient Interaction Phenotypes With a Post-Discharge Text Messaging Service and Their Association With Hospital Revisits

Pdf Summary

This study analyzed how patients interacted with a 30-day automated post-discharge text messaging program in the MORE-PC trial. Using unsupervised k-means clustering on participants who sent at least one message, the researchers engineered “conformity” and “engagement” features to identify distinct behavioral phenotypes.

Four interaction groups emerged: <strong>Enthusiasts</strong>, <strong>Minimalists</strong>, <strong>Non-Adapters</strong>, and <strong>High-Need Responders</strong>. Enthusiasts formed the largest group and had high response rates, relatively fast responses, and low error rates. Minimalists responded less often and more slowly. Non-Adapters showed the poorest fit with the program, with low engagement, long response times, and very high error rates. High-Need Responders were a small group who sent many inbound messages, requested calls frequently, and appeared to have the greatest care needs.

The study also examined associations between these phenotypes, patient characteristics, and acute care use after discharge. The findings suggest that messaging behavior may provide useful signals about patient needs and risk beyond the content of the messages themselves.

The authors conclude that identifying engagement phenotypes could help tailor digital health programs: minimalists may need strategies to encourage participation, non-adapters may benefit from extra coaching or different communication methods, and enthusiasts may require little additional support. More broadly, these patterns could be used to improve transitional care messaging systems and other digital health interventions.

Asset Subtitle

Eric Bressman

Meta Tag

Author List Agnes Wang, Anna U. Morgan, Danielle Mowery, Emily Schriver, Eric Bressman, Klea Profka

Category Research

Concept Patient Engagement Phenotype

Concept Unsupervised Clustering

Concept Automated Text Messaging Program

Concept Engagement Feature

Concept Behavioral Phenotyping

Distinguished Non-Finalist

Presenter Organization University of Pennsylvania School of Medicine

Presenting Author Eric Bressman

Track Transitions of Care

Keywords

post-discharge messaging

patient engagement

k-means clustering

behavioral phenotypes

digital health intervention

transitional care

response rates

care needs

acute care use

text messaging program

Patient Engagement Phenotype

Unsupervised Clustering

Automated Text Messaging Program

Engagement Feature

Behavioral Phenotyping