Segmenting User Behavior: A Clustering Approach to Understanding Digital Health Engagement

Data and code for: "Segmenting User Behavior: A Clustering Approach to Understanding Digital Health Engagement"This repository contains a de-identified dataset and Python analysis code to fully reproduce all manuscript figures and the baseline characteristics table. The dataset comprises 535 survey responses from 210 participants across three timepoints (baseline, 6 months, 12 months), linking REDCap survey data with mobile app paradata from an mHealth HIV self-testing intervention. Variables include HIV testing behavior, PrEP use, health literacy, app usability (ITUES), and in-app usage metrics (session counts, time, test counts).To reproduce the analysis:```pip install -r requirements.txtpython run_analysis.py```This generates 10 figures (PNG + PDF) in results/figures/ and a baseline characteristics table in results/tables/. K-means clustering (k=2 and k=3) on in-app usage identifies engagement subgroups; analyses cover Pearson correlations, demographic distributions, and longitudinal clinical outcomes. See README.md for full documentation and DATA_DESCRIPTION.md for variable definitions.