An interactive visualization dashboard for predicting the effect of sacubitril/valsartan initiation in patients with...

Sacubitril/valsartan (S/V) reduces mortality and hospitalization rates in patients with heart failure with reduced ejection fraction (HFrEF), but low adherence remains a challenge. Early initiation of S/V is recommended, yet no practical tool currently exists to effectively communicate its benefits to outpatients or assess patient stability before S/V initiation during hospitalization.

We collected data retrospectively from 527 HFrEF patients who started S/V between March 2017 and January 2020 at the National Taiwan University Hospital, with follow-up through September 2022. A modern stepwise variable selection approach was applied to fit the optimal Cox's proportional-hazards model to address nonlinear covariate effects and potential multicollinearity. Penalized smoothing splines were used to visualize nonlinear effects and identify cutoff values for continuous covariates. The model was then integrated into an interactive Streamlit dashboard for real-time simulation and risk prediction based on patient-specific covariates.

We identified 20 key variables - 12 associated with increased risk and 8 conferring protective effects. Some variables presented actionable cutoff values. The predictive model helped estimate individualized hazard ratios and covariate-adjusted survival curves, which serve as both a patient engagement platform in outpatient settings and a practical tool for physicians to assess inpatient stability before initiating S/V.

Our interactive visualization dashboard can potentially improve medication adherence and clinical outcomes by involving patients in their treatment journey and facilitating more informed decision-making. A pragmatic clinical trial is currently underway to further evaluate the dashboard's clinical utility and effectiveness in both outpatient and inpatient environments.