Smartphone-measured tremor and tacrolimus trough levels in kidney transplant recipients: an exploratory prospective study.

Abstract

Tacrolimus commonly causes tremor as a side effect, although the relationship between tremor and tacrolimus drug levels is not well established. This work presents the development of a smartphone-based tremor assessment to explore whether tremor features are associated with tacrolimus trough concentrations, including whether individualized models show promise in tremor-sensitive kidney transplant patients. Using smartphone accelerometers, an app was created to record resting and postural tremors in transplant recipients during routine follow-up visits. Tacrolimus trough concentration and doses were recorded at each visit. We evaluated associations between tremor features and tacrolimus trough levels and, in exploratory analyses, assessed regression models both at the cohort level and within individuals with sufficient longitudinal data. Sixty-nine kidney transplant recipients were included. Correlations were found between self-reported tremor severity and tremor features, but no significant association was observed between tacrolimus trough concentration and tremor features at the cohort level (ρ = 0.022, p = 0.447). Population-level regression models showed poor predictive performance. post-hoc exploratory individualized models among patients with adequate longitudinal data (n = 8) provided an averaged root mean squared error of 2.33 ± 1.42 µg/L; three patients achieved root mean squared errors below 1.3 µg/L. Smartphone tremor features did not support population-level prediction of tacrolimus trough concentrations in kidney transplant recipients. Exploratory individualized regression models yielded potentially useful performance in a small subset; consistent with heterogeneity in tacrolimus tremor severity. These findings are hypothesis-generating; larger prospective studies should focus on identifying and validating tremor-sensitive patients for individualized approaches.