JMIR Publications

ABSTRACT

Background:

The integration of digital tools and patient-reported outcomes (ePROMs/ePREMs) enables longitudinal assessment of health-related quality of life (HRQoL) and treatment satisfaction among pre-exposure prophylaxis (PrEP) users. However, determinants of digital engagement remain poorly understood. Objectives: The primary objective was to assess the usefulness of the Naveta-Phemium digital platform in monitoring the effectiveness, safety, HRQoL, and treatment satisfaction of PrEP among HIV-negative adults at high risk of infection. The secondary objective was to develop and validate a participant classification system to evaluate and predict engagement with the digital tool.

Objective:

The primary objective was to assess the usefulness of the Naveta-Phemium digital platform in monitoring the effectiveness, safety, HRQoL, and treatment satisfaction of PrEP among HIV-negative adults at high risk of infection. The secondary objective was to develop and validate a participant classification system to evaluate and predict engagement with the digital tool.

Methods:

A 24-month prospective study was conducted using the Naveta digital follow-up platform. HIV-negative adults at high risk of infection received a fixed-dose combination of tenofovir disoproxil fumarate (TDF, 245 mg) and emtricitabine (FTC, 200 mg). Clinical safety, HRQoL, and treatment satisfaction were evaluated through laboratory biomarkers and ePROMs/ePREMs (HADS, PROMIS-29, TSQM, P3CEQ). Sociodemographic and behavioral determinants of digital engagement were analyzed using ALGOPROMIA-Classification, a machine learning–based framework for predictive modeling.

Results:

Eighty-one participants were included (mean PrEP duration 689 days). PrEP was well tolerated, and no participant reported moderate or severe adverse effects. Only mild, transient symptoms were recorded, predominantly of gastrointestinal (68.9%) and nervous system (57.8%) types. Renal function remained stable (creatinine 0.86 ± 0.13 mg/dL; eGFR p = 0.498), showing strong temporal consistency (r = 0.63–0.92). HRQoL and psychological well-being remained stable (HADS <7; PROMIS-29 within normative range), and treatment satisfaction was consistently high (TSQM ≈85–87). Engagement with the Naveta platform displayed clear sociodemographic and behavioral gradients, being higher among men, older users, and those with regular exercise or balanced diet. Ensemble machine learning models achieved strong discrimination in predicting engagement (AUC ≈0.87), with Histogram-based Gradient Boosting and Random Forest performing best. SHAP analysis identified healthy diet, physical activity, and higher education as the most influential positive predictors of sustained participation.

Conclusions:

The Naveta digital platform demonstrated the feasibility and safety of long-term PrEP monitoring, with preserved HRQoL, high treatment satisfaction, and stable renal function. Machine learning analysis confirmed that lifestyle and social stability are key determinants of sustained digital engagement in preventive care. Clinical Trial: NAP