Aseem Pradhan1, John Scaringi2, Patrick Gerard3, Ross Arena4, Jonathan Myers5, Leonard A Kaminsky6, Ethan Kung7,8. 1. Department of Mechanical Engineering, Clemson University, Clemson, SC, USA. 2. Department of Bioengineering, Clemson University, Clemson, SC, USA. 3. School of Mathematical and Statistical Sciences, Clemson University, Clemson, SC, USA. 4. Department of Physical Therapy, College of Applied Science, University of Illinois at Chicago, Chicago, IL, USA. 5. Division of Cardiology, VA Palo Alto Healthcare System, Palo Alto, CA, USA. 6. Fisher Institute of Health and Well-Being and Clinical Exercise Physiology Laboratory, Ball State University, Muncie, IN, USA. 7. Department of Mechanical Engineering, Clemson University, Clemson, SC, USA. ekung@clemson.edu. 8. Department of Bioengineering, Clemson University, Clemson, SC, USA. ekung@clemson.edu.
Abstract
PURPOSE: Blood pressure, cardiac output, and ventricular volumes correlate to various subject features such as age, body size, and exercise intensity. The purpose of this study is to quantify this correlation through regression modeling. METHODS: We conducted a systematic review to compile reference data of healthy subjects for several cardiovascular parameters and subject features. Regression algorithms used these aggregate data to formulate predictive models for the outputs-systolic and diastolic blood pressure, ventricular volumes, cardiac output, and heart rate-against the features-age, height, weight, and exercise intensity. A simulation-based procedure generated data of virtual subjects to test whether these regression models built using aggregate data can perform well for subject-level predictions and to provide an estimate for the expected error. The blood pressure and heart rate models were also validated using real-world subject-level data. RESULTS: The direction of trends between model outputs and the input subject features in our study agree with those in current literature. CONCLUSION: Although other studies observe exponential predictor-output relations, the linear regression algorithms performed the best for the data in this study. The use of subject-level data and more predictors may provide regression models with higher fidelity. SIGNIFICANCE: Models developed in this study can be useful to clinicians for personalized patient assessment and to researchers for tuning computational models.
PURPOSE: Blood pressure, cardiac output, and ventricular volumes correlate to various subject features such as age, body size, and exercise intensity. The purpose of this study is to quantify this correlation through regression modeling. METHODS: We conducted a systematic review to compile reference data of healthy subjects for several cardiovascular parameters and subject features. Regression algorithms used these aggregate data to formulate predictive models for the outputs-systolic and diastolic blood pressure, ventricular volumes, cardiac output, and heart rate-against the features-age, height, weight, and exercise intensity. A simulation-based procedure generated data of virtual subjects to test whether these regression models built using aggregate data can perform well for subject-level predictions and to provide an estimate for the expected error. The blood pressure and heart rate models were also validated using real-world subject-level data. RESULTS: The direction of trends between model outputs and the input subject features in our study agree with those in current literature. CONCLUSION: Although other studies observe exponential predictor-output relations, the linear regression algorithms performed the best for the data in this study. The use of subject-level data and more predictors may provide regression models with higher fidelity. SIGNIFICANCE: Models developed in this study can be useful to clinicians for personalized patient assessment and to researchers for tuning computational models.
Authors: Nathalie Mm Benda; Maria Te Hopman; Arie Pj van Dijk; David Oxborough; Keith P George; Dick Hj Thijssen; Thijs Mh Eijsvogels Journal: Eur J Prev Cardiol Date: 2016-02-08 Impact factor: 7.804
Authors: Aron S Bode; Wouter Huberts; E Marielle H Bosboom; Wilco Kroon; Wim P M van der Linden; R Nils Planken; Frans N van de Vosse; Jan H M Tordoir Journal: PLoS One Date: 2012-04-04 Impact factor: 3.240