Dissertation Defense | Brinnae Bent

TOPIC: Discovering Digital Biomarkers of Glycemic Health from Wearable Sensors

KEYWORDS: wearables, digital biomarkers, prediabetes screening, noninvasive glucose monitoring, feature engineering, machine learning, data compression, sampling rate minimization, open source software


Thank you for joining me today for the defense of my dissertation in the Big Ideas Lab at Duke University. Below you will find links to all of the publications presented in the dissertation, as well as resources for further exploration into the topics discussed. If you have questions, be sure to check out the FAQ section. Of course, questions can also be addressed to me via email. If you want to stay up to date on future publications, consider following me on Twitter. Curious what else I have been up to? Visit my website.



Click here for the SparkNotes version of the dissertation.


Objectives


  Objective 1

Investigating Sources of Inaccuracy in Wearable Optical Heart Rate Sensors

Bent, et.al. “Investigating Sources of Inaccuracy in Wearable Optical Heart Rate Sensors.” Nature Digital Medicine. (2020) (Link).

Bent, et.al. “Reply to Matters Arising: Investigating Sources of Inaccuracy in Wearable Optical Heart Rate Sensors.” Nature Digital Medicine. (2021) (Link).

Blog: Can Wearables Keep Up With Us? Bent and Dunn (2020).

Goldsack, Coravos, Bakker, Bent, et.al. “Verification, analytical validation, and clinical validation (V3): the foundation of determining fit-for-purpose for Biometric Monitoring Technologies (BioMeTs).” Nature Digital Medicine. (2020) (Link).

  Objective 2

Optimizing Sampling Rate of wrist-worn optical sensors

Bent and Dunn. “Optimizing sampling rate of wrist-worn optical sensors for physiologic monitoring.” Journal of Clinical and Translational Science. (2020) (Link).

  Objective 3

Data Compression Strategies for Digital Biomarkers (The Biosignal Data Compression Toolbox)

Bent, et.al. “Biosignal Compression Toolbox for Digital Biomarker Discovery.” Sensors. (2021) (Link).

  Objective 4

The Digital Biomarker Discovery Pipeline

Bent, et.al. “The digital biomarker discovery pipeline: An open-source software platform for the development of digital biomarkers using mHealth and wearables data.” Journal of Clinical and Translational Science. (2020) (Link).

Bent, et.al. “Digital Medicine Community Perspectives and Challenges: Survey Study.” JMIR mHealth and uHealth. (2021) (Link).

Blog on The Digital Biomarker Discovery Pipeline. Bent (2020).

The DBDP Website | GitHub | Blog | dbdpED

watch   Objective 5

Proof-of-Concept: Noninvasive Wearables for Remote Monitoring of HbA1c and Glucose Variability

Main manuscript currently under review.

Bent, et.al. “cgmquantify: Python and R packages for comprehensive analysis of interstitial glucose and glycemic variability from continuous glucose monitor data.” Preprint. (Link to Preprint).

cgmquantify Python package on GitHub

wearablevar Python package on GitHub


add_alarm   Objective 6

Engineering Digital Biomarkers of Interstitial Glucose

Main manuscript currently under review.

Bent and Dunn.“Personalized Machine Learning Models for Noninvasive Glucose Prediction Using Wearables”. NeurIPS Machine Learning for Mobile Health Workshop. (2020) (Link).

wearablecompute Python package on GitHub


Frequently Asked Questions

My smartwatch is not accurate in situation X. What do I do?

First, it is important to recognize that most of these devices are not currently approved for or marketed for medical purposes. While it may be in the fine print, these devices are meant for general fitness and wellness and may not be accurate in all situations. We are currently working on this! (Check out our recent paper on this topic, available here). We showed in our study investigating sources of inaccuracy in wearable optical heart rate sensors that activity and device had an effect on accuracy.

As a consumer, it is important to be aware that these smartwatches are not medical devices. A couple things that have been shown to increase accuracy is cleaning the back of the watch regularly and wearing the watch correctly (see the guidelines of your specific smartwatch. We usually don't wear them correctly!)

If you are a researcher or other party using these devices, we recommend evaluating them for fit-for-purpose using the V3 Evaluation Framework.


What are tree-based machine learning models (decision trees, random forests, and gradient boosted models)?

Tree-based machine learning models, including decision trees, random forests, and gradient boosting, are commonly used machine learning methods. I wrote a blog on understanding decision trees, available here.


What is cross validation?

Cross validation is a method to estimate how well a machine learning model performs and there are a number of ways to perform cross validation. To learn more about the different cross validation methods used in this dissertation, please check out this blog I wrote on cross validation.


What is feature engineering?

Feature engineering is the process of using domain knowledge to create features (or variables) that become the input of a machine learning model. I explain feature engineering using my favorite food, pizza, in this blog.


I have a CGM and want to start looking at my data. Where do I start?

As part of dbdpED, I have created a tutorial on how to work with data from a continuous glucose monitor (CGM). For the complete tutorial, check out the blog, available here.


How can I get involved with the DBDP?

If you are new to digital biomarker discovery, we recommend checking out dbdpED, which has tutorials and resources to get started. For the more experienced researcher, we have an extensive user guide for users, contributors, and digital medicine enthusiasts alike! We welcome contributions to the DBDP! Whether it is a complete end-to-end module for a digital biomarker, a method for pre-processing of wearables or mHealth data, or a tool for digital biomarker discovery, your code has a home in the DBDP. Check out this blog to learn more, or visit our website.



Publications

See most up to date publication list at Google Scholar .


Manuscripts currently under peer-review/revisions

*B. Bent, W.E. Hammond. “Data-centric Recommendations for EHR Re-design in a Post-pandemic World”.

*E. Grzesiak, B. Bent, M. McClain, C. Woods, E. Tsalik, B. Nicholson, T. Veldman, T. Burke, C. Nix, S. Evans, Z. Gardener, E. Bergstrom, R. Turner, C. Chiu, P.M. Doraiswamy, A. Hero, R. Henao, G.S. Ginsburg, J.P. Dunn. “Detecting Influenza and the Common Cold Before Symptom Onset with Noninvasive Wearable Sensors”.

*B. Bent, P. Cho, A. Wittman, M. Snyder, M. Crowley, M. Feinglos, J.P. Dunn. “Noninvasive Wearables for Remote Monitoring of HbA1c and Glucose Variability: Proof of Concept”.

*B. Bent, M. Henriquez, J.P. Dunn. “cgmquantify: Python and R packages for comprehensive analysis of interstitial glucose and glycemic variability from continuous glucose monitor data”.


Peer-Reviewed Publications (*indicates journal pub, †indicates conference pub)

*B. Bent, O.M. Enache, B.A. Goldstein, W.A. Kibbe, J.P. Dunn. “Reply: Matters Arising Investigating Sources of Inaccuracy in Wearable Optical Heart Rate Sensors.” Nature Digital Medicine. (2021)

*B. Bent, B. Lu, J. Kim, J.P. Dunn. “Biosignal Compression Toolbox for Digital Biomarker Discovery”. Sensors. (2021).

*B. Bent, I. Sim, J.P. Dunn. “Digital Medicine Community Perspectives and Challenges: Survey Study”. JMIR MHealth and Uhealth. (2021).

*B. Bent, J.P. Dunn. “Wearables in a Pandemic: What are They Good For?” JMIR Mhealth Uhealth. (2020).

B. Bent, J.P. Dunn. “Personalized Machine Learning Models for Noninvasive Glucose Prediction Using Wearables”. NeurIPS Machine Learning for Mobile Health Workshop. (2020).

*M. Trumpis, C.H. Chiang, A. Osborn, B. Bent, J. Li, J.A. Rogers, B. Pesaran, G. Cogan, J. Viventi. “Sufficient Sampling for Kriging Prediction of Cortical Potential in Rat, Monkey, and Human, uECoG”. Journal of Neural Engineering. (2020).

*M. Henriquez, J. Sumner, M. Faherty, T. Sell, **B. Bent. “Machine Learning to Predict Lower Extremity Musculoskeletal Injury Risk in Student Athletes”. frontiers: Sports Science, Technology, and Engineering. (2020). (**corresponding, senior author with funding responsibilities)

*B. Bent, J.P. Dunn. “Optimizing Sampling Rate of Wrist-worn Optical Sensors for Physiologic Monitoring”. Journal of Clinical and Translational Science. (2020).

*B. Bent, K. Wang, E. Gerzesiak, C. Jiang, Y. Qi, Y. Jiang, P. Cho, K. Zingler, F. Ogbeide, A. Zhao, I. Sim, J. Dunn. “Digital Biomarker Discovery Pipeline: An open source software platform for the development of digital biomarkers using mHealth and wearables.” Journal of Clinical and Translational Science. (2020).

*Y. Jiang, Y. Qi*, K. Wang, B. Bent, R. Avram, J. Olgin, J. Dunn. “EventDTW: An Improved Dynamic Time Warping Algorithm for Aligning Biomedical Signals of Uneven Sampling Frequencies.” Sensors. (2020)

*J.C. Goldsack, A. Coravos, J. Bakker, B. Bent, A. Dowling, C. Fitzer-Attas, A. Godfrey, J.G. Godino, N. Gujar, E. Ismailova, C. Manta, B, Peterson, B. Vandendressche, W. Wood, W. Wang, J. Dunn. “Verification, Analytical Validation, and Clinical Validation (V3): The Foundation of Determining Fit-for-Purpose for Biometric Monitoring Technologies (BioMeTs)”. Nature Digital Medicine. (2020).

*B. Bent, B.A. Goldstein, W.A. Kibbe, J.P. Dunn. “Investigating Sources of Inaccuracy in Wearable Optical Heart Rate Sensors.” Nature Digital Medicine. (2020)

*CH. Chiang, S. Won, A. Orsborn, KJ. Yu, M. Trumpis, B. Bent, C. Wang, Y. Xue, S. Min, V. Woods, C. Yu, BH. Kim, SB Kim, R. Huq, J. Li, KJ. Seo, F. Vitale, H. Fang, Y. Huang, K. Shepard, B. Pesaran, JA Rogers, J. Viventi. “Development of a neural interface for high-defintion, long-term recording in rodents and nonhuman primates”. Science Translational Medicine. (2020).

B. Bent, CH. Chiang, C. Wang, N. Lad, A. Kent, J. Viventi. “Simultaneous Recording and Stimulation Instrumentation for Closed Loop Spinal Cord Stimulation.” IEEE Neural Engineering Conference Proceedings. (2019).

B. Bent, A.J. Williams, R. Bolick, K. Chiang, M. Trumpis, J. Viventi. “3D Printed Cranial Window System for Chronic μECoGRecording.” IEEE Engineering Medicine and Biology Conference Proceedings. (2018).

†A.J. Williams, M. Trumpis, B. Bent, K. Chiang, J. Viventi. “A Novel μECoG Electrode Interface for Comparison of Local and Common Averaged Referenced Signals.” IEEE Engineering Medicine and Biology Conference Proceedings. (2018).

*V. Woods, M. Trumpis, B. Bent, C.H. Chiang, K. Palopoli-Trojani, C. Wang, M. Insanally, R.C. Froemke, and J. Viventi, “Chronic​reliability of μECoG arrays implanted for greater than one year in rodents,” Journal of Neural Engineering (2018).

†M. Sahraee Ardakan, M. Emami, AK Fletcher, M. Trumpis, B. Bent, J. Viventi. “Learning Nonlinear Dynamical Networks in Neural Systems”. Conference on Cognitive Computational Neuroscience Proceedings. (2017).

*J. Dieffenderfer, H. Goodell, S. Mills, M. McKnight, S. Yao, F. Lin, E. Beppler, B. Bent, V. Misra, Y. Zhu, O. Oralkan, J. Strohmaier, J. Muth, D. Peden, and A. Bozkurt. “Low Power Wearable Systems for Continuous Monitoring of Environment and Health for Chronic​ Respiratory Disease.” Journal of Biomedical and Health Informatics (2016).

†Laura Gonzales, Katherine Walker, Sindhuja Challa, B. Bent. “Monitoring a Skipped Heartbeat: A Real-time Premature Ventricular Contraction (PVC) Monitor”. IEEE Virtual Conference on Applications of Commercial Sensors (2016).

*B. Bent, Dr. Alper Bozkurt. “Miniaturizing Plethysmography for use in a Multifunctional Health Monitoring Device withApplications for Asthma Analysis.” State of North Carolina Undergraduate Research Journal: Explorations Vol. X (2015).

†Dieffenderfer, James P., Henry Goodell, B. Bent, Eric Beppler, Rochana Jayakumar, Murat Yokus, Dr. Jesse S. Jur, and Dr.Alper Bozkurt. "Wearable Wireless Sensors for Chronic Respiratory Disease Monitoring." IEEE Body Sensor Networks (2015).


Please Click Here for Conference Presentations & Invited Talks

Conference Presentations & Invited Talks


December 2020 | Invited Speaker, Endocrinology Grand Rounds, Duke University (virtual), "Feasibility of using smartwatches to monitor glycemic health"

May 2020 | Invited Speaker, Digital Medicine Society Webinar Series (virtual), "Investigating Sources of Inaccuracy in Wearable Optical Heart Rate Sensors"

April 2020 | Invited Speaker, Duke Center for Health Informatics Seminar Series (virtual), "Validating Wearable Sensors for Digital Biomarker Discovery" [Link]

February 2020 | Speaker, Banff International Research Station Workshop Use of Wearable & Implantable Devices in Health Research (Banff, Canada), "Open-source digital biomarker development" [Link]

October 2019 | Poster Presentation, Biomedical Engineering Society Conference BMES (Philadelphia, PA), "Determining the Optimal Sampling Frequency of Photoplethysmography from Wearables for Heart Rate Variability"

April 2019 | Selected Talk, GradX (Durham, NC), "What Happens to the Body when you Run 100 Miles?"

January 2019 | Poster Presentation, North American Neuromodulation Society Conference NANS (Las Vegas, NV), "Simultaneous Recording and Stimulation Instrumentation for Closed Loop Spinal Cord Stimulation"

July 2018 | Oral Presentation, Engineering, Medicine, and Biology Conference EMBC (Honolulu, HI), "3D Printed Cranial Window System for Chronic µECoG Recording"

June 2018 | Poster Presentation, Neural Interfaces Conference NIC (Minneapolis, MN), "3D Printed Cranial Window System for Chronic µECoG Recording"

March 2018 | Selected Talk, Broader Impacts through Research Seminar (Durham, NC), "How an interactive classroom and a robotic arm has turned mistrust of neural implants into understanding for senior citizens". 3rd Place Award.

December 2017 | Oral Presentation, Regional BMES Symposium (Durham, NC), "Flexible, High-density µECoG Electrodes for in vivo Applications", co-presented with A. Williams.

October 2017 | Poster Presentation, Biomedical Engineering Society Conference BMES (Phoenix, AZ), "In Vivo Evaluation of Chronic Reliability of High Resolution, Low-cost µECoG Arrays"

May 2016 | Oral Presentation, Abrams Research Fellowship Final Symposium (Raleigh, NC), "Quantifying Sleep Quality with Smartband Technology"

April 2016 | Poster Presentation, National Conference on Undergraduate Research NCUR (Asheville, NC), "Miniaturizing Photoplethysmography for use in a Multifunctional Health Monitoring Device with Applications in Asthma Analysis"

November 2015 | Poster Presentation, State of NC Undergraduate Research Symposium SNCURS (High Point, NC), "Miniaturizing Photoplethysmography for use in a Multifunctional Health Monitoring Device with Applications in Asthma Analysis"

October 2015 | Selected Poster Presentation, Society of Women Engineers National Conference (Nashville, TN), "Miniaturizing Photoplethysmography for use in a Multifunctional Health Monitoring Device with Applications in Asthma Analysis"

October 2015 | Poster Presentation, Biomedical Engineering Society Conference BMES (Tampa FL), "Miniaturizing Photoplethysmography for use in a Multifunctional Health Monitoring Device with Applications in Asthma Analysis"

July 2015 | Oral Presentation, University of Michigan Summer Research Symposium (Ann Arbor, MI), "Characterizing Micelles for Application in Nanoscale Electrochemical Voltage Sensors"

November 2014 | Poster Presentation, State of NC Undergraduate Research Symposium SNCURS (Raleigh, NC), "Quantifying Sleep Quality with Sleepiband"


References - Defense Presentation


Additional Resources

Exploring Machine Learning

Blog on Supervised vs. Unsupervised ML by Bent (2021)

Blog on Cross Validation by Bent (2020)

Blog on Feature Engineering by Bent (2021)

Blog on Linear Regression by Bent (2021)

Blog on Decision Tree Based Models by Bent (2020)


Creating Data Visualizations (like the ones shown in this dissertation)

Blog on Data Viz Word Clouds by Bent (2020)

Blog on Data Viz Geographical Maps by Bent (2020)

Blog on Data Viz Donut Plots by Bent (2020)

Blog on Pie Charts by Bent (2020)