Quantitative Research

Ongoing research

SCOS Cuffless Blood-Pressure Research

Timothy leads an IRB-supervised positional signal-to-noise-ratio experiment, applies XGBoost to 80+ engineered cardiac-waveform features for blood-pressure prediction and hypertension classification, and develops wearable test hardware.

Multi-panel scientific figure showing SCOS optical measurement principles, cardiac waveforms, and simultaneous measurement configurations.
Outcome
Implemented a positional signal-to-noise-ratio measurement protocol and analysis workflow, with attachable hardware for wireless wearable benchmark testing.
Timothy’s contribution
Timothy leads protocol development, participant recruitment, measurements, and data analysis; works with 80+ engineered waveform features; and uses Autodesk Inventor and additive manufacturing for wearable test hardware.
Role
Undergraduate Researcher
Institution
Boston University Biomedical Optical Technologies Lab
Date
February 2026–Present
Technologies
Python, XGBoost, Autodesk Inventor
Methods
IRB-supervised positional signal-to-noise-ratio experiment, Measurement protocol, recruitment, and measurements, Dataset construction from raw trial data, Cardiac-waveform feature engineering, Blood-pressure prediction, Hypertension classification, Clinical-cohort comparison, CAD test-base development
Visit institutional contextRelated experience: Undergraduate Researcher

Advisor-authored publication informing the work

Ariane Garrett, Byungchan Kim, Nil Z. Gurel, Edbert J. Sie, Benjamin K. Wilson, Francesco Marsili, John P. Forman, Naomi M. Hamburg, David A. Boas, and Darren Roblyer, “Speckle contrast optical spectroscopy for cuffless blood pressure estimation based on microvascular blood flow and volume oscillations,” Biomedical Optics Express 16, 3004–3016 (2025).

View publication (opens in a new tab)

Problem and constraints

Cuffless blood-pressure estimation depends on useful optical measurements and a repeatable analysis workflow. Timothy's current experiment investigates positional signal-to-noise ratio, while his analytical work connects cardiac-waveform features to blood-pressure prediction and hypertension classification.

The experiment is IRB-supervised. Participant-level data is private, and this case study reports implemented methods without claiming clinically validated performance.

Timothy's role

Timothy leads measurement-protocol development, participant recruitment, measurements, and data analysis for the positional experiment. He also applies XGBoost to 80+ engineered cardiac-waveform features, analyzes hypertensive and non-hypertensive subjects, presents feature patterns, and develops wearable benchmark-test hardware.

Collaboration

Postdoctoral advisor Ariane Garrett provides research advisement and code support for an existing XGBoost workflow. Earlier dataset construction used raw scientific-trial data from studies Garrett ran. That earlier data provenance remains separate from the IRB-supervised positional experiment Timothy now leads.

Technical architecture and methods

Measurement protocol and dataset

Timothy creates and implements the protocol for positional signal-to-noise-ratio measurements and leads recruitment, measurement, and analysis. His earlier project-specific dataset organized advisor-run trial data for repeated analysis and clinical-cohort comparison.

Feature and modeling workflow

The analytical workflow uses 80+ engineered cardiac-waveform features. Timothy applies XGBoost to predict blood pressure and classify hypertension and analyzes feature patterns from hypertensive and non-hypertensive subjects.

The modeling builds on an existing XGBoost workflow adapted under Garrett's advisement and with advisor-assisted code.

Physical development

Timothy uses Autodesk Inventor and additive manufacturing to develop an attachable base for wireless wearable benchmark testing.

Engineering decisions and research constraints

The research connects experimental protocol development, traceable data preparation, cohort comparison, an established modeling workflow, and physical test hardware. Advisor support and earlier trial provenance remain part of the technical account.

Preliminary findings have been presented internally. No participant count, model-performance number, or clinical-validation result is asserted in this case study.