Abstract
BACKGROUND Prediction of New York Heart Association (NYHA) functional class in heart failure (HF) has relied on electronic medical record (EMR) data, omitting psychosocial determinants now collectible through patient-facing applications. OBJECTIVE This study aimed to determine whether patient-reported measures of anxiety, perceived social support, and health literacy, collected through a patient-facing digital health application, improve identification of advanced NYHA functional limitation beyond routinely available EMR data in adults with HF. METHODS Single-center pilot observational study of adults with HF receiving care in inpatient and outpatient settings. Clinical and demographic data were obtained from the EMR; patient-reported measures were collected through a patient-facing digital health application. The primary outcome was self-reported NYHA Class III-IV versus Class I-II. A two-stage framework compared Stage 1, comprising prespecified EMR-derived variables (age, sex, HF duration, comorbidity burden, guideline-directed medical therapy, and diuretic use), with Stage 2, which added the patient-reported measures. Logistic regression, elastic-net, random forest, and gradient-boosted tree models were evaluated using discrimination, calibration, reclassification, and prediction-error measures with repeated cross-validation and sensitivity analyses. RESULTS The Stage 2 elastic net (primary model) achieved cross-validated AUC 0.66 (95% CI 0.55, 0.77) versus Stage 1 (EMR-only) AUC 0.57 (0.45, 0.70); the primary incremental-discrimination test was non-significant (AUC difference +0.09, DeLong p = 0.094). In exploratory analyses, the continuous NRI was 0.45 (95% CI 0.08, 0.83), but its event component was zero for both linear models. Two tree-based comparators reached nominal DeLong significance (p = 0.003, p = 0.007) but were not considered confirmatory: the random-forest EMR-only baseline was at chance (AUC 0.49), and gradient-boosted-tree calibration and Brier score worsened with the added predictors. Under fully nested cross-validation the elastic-net Stage 2 AUC was 0.62 with a repeat-level increment of +0.10 (2.5-97.5 percentile range across repeats 0.02 to 0.16), and the bootstrap optimism-corrected Stage 2 AUC was 0.69. In MICE-pooled multivariable logistic regression, GAD-7 was the only individually significant predictor (OR 1.15 per point, 95% CI 1.02 to 1.30, p = 0.022). CONCLUSIONS In this single-center pilot observational study, adding patient-reported psychosocial measures to routinely available EMR-derived clinical variables produced a numerical but statistically nonsignificant improvement in discrimination of advanced self-reported NYHA functional limitation. Anxiety was the only patient-reported measure with a consistent individual association across analytic approaches, although interpretation is limited by concurrent collection of predictor and outcome through the same application. Larger multisite studies with independently assessed functional outcomes and external validation are needed to determine whether patient-generated psychosocial data add meaningful incremental value for HF risk stratification.

This publication has 29 references indexed in Scilit: