Predicting disease progression in patients with bicuspid aortic stenosis using mathematical modeling

Darae Kim, Dongwoo Chae, Chi Young Shim, In Jeong Cho, Geu Ru Hong, Kyungsoo Park, Jong Won Ha

Research output: Contribution to journalArticlepeer-review

6 Scopus citations


We aimed to develop a mathematical model to predict the progression of aortic stenosis (AS) and aortic dilatation (AD) in bicuspid aortic valve patients. Bicuspid AS patients who underwent at least two serial echocardiograms from 2005 to 2017 were enrolled. Mathematical modeling was undertaken to assess (1) the non-linearity associated with the disease progression and (2) the importance of first visit echocardiogram in predicting the overall prognosis. Models were trained in 126 patients and validated in an additional cohort of 43 patients. AS was best described by a logistic function of time. Patients who showed an increase in mean pressure gradient (MPG) at their first visit relative to baseline (denoted as rapid progressors) showed a significantly faster disease progression overall. The core model parameter reflecting the rate of disease progression, α, was 0.012/month in the rapid progressors and 0.0032/month in the slow progressors (p < 0.0001). AD progression was best described by a simple linear function, with an increment rate of 0.019 mm/month. Validation of models in a separate prospective cohort yielded comparable R squared statistics for predicted outcomes. Our novel disease progression model for bicuspid AS significantly increased prediction power by including subsequent follow-up visit information rather than baseline information alone.

Original languageEnglish
Article number1302
JournalJournal of Clinical Medicine
Issue number9
StatePublished - Sep 2019

Bibliographical note

Publisher Copyright:
© 2019 by the authors. Licensee MDPI, Basel, Switzerland.


  • Bicuspid aortic valve
  • Mathematical model
  • Progression


Dive into the research topics of 'Predicting disease progression in patients with bicuspid aortic stenosis using mathematical modeling'. Together they form a unique fingerprint.

Cite this