Research question
This ongoing work examines how speckle contrast optical spectroscopy data and cardiac-waveform analysis can support cuffless blood-pressure estimation. The project joins experimental measurements, clinical-data preparation, quantitative feature engineering, machine-learning workflows, cohort comparison, and wearable test-base development.
Timothy's role begins after the scientific trials: postdoctoral advisor Ariane Garrett ran the trials that produced the raw data, and Timothy created a project-specific dataset for his analysis. That distinction keeps the experimental and analytical contributions clear.
Timothy's contribution
Timothy constructed the project-specific dataset, engineered more than 700 cardiac-waveform features, performed clinical-data analysis, and contributed blood-pressure prediction and adapted hypertension-classification work. He compared relevant clinical cohorts, developed CAD for an attachable wearable test base, and presented preliminary findings to the lab team.
His modeling work is explicitly advisor-supported. Under Garrett's advisement and with advisor-assisted code, Timothy adapted an existing XGBoost pipeline for his dataset. The work does not claim that he independently created the XGBoost system or the underlying hypertension-classification approach.
Advisor-supported workflow
Garrett's experimental work and advisement establish the scientific context for the project. Timothy translates raw trial outputs into a structured analysis workflow, develops project-specific features and comparisons, and returns preliminary findings to the lab team for review. Advisor assistance remains part of the code and modeling process throughout that loop.
Dataset and waveform methods
Timothy structured raw trial data into a project-specific dataset designed for repeatable analysis. He then derived more than 700 features from cardiac waveforms, creating a broad quantitative representation for ongoing blood-pressure prediction, hypertension-classification, and clinical-cohort comparison work.
The public description stops at implemented methods. It does not publish participant-level data, report model scores, or imply clinical validation.
Machine-learning workflow
The project adapts an existing XGBoost pipeline rather than presenting a newly invented model. Timothy configured that established workflow for his dataset under Garrett's advisement and with advisor-assisted code, then used it within the ongoing prediction and classification investigation.
This attribution is part of the technical account: Timothy's contribution lies in dataset construction, feature engineering, clinical analysis, cohort comparison, and project-specific adaptation within an advisor-supported research system.
Wearable test-base development
Timothy also developed CAD in Autodesk Inventor for an attachable wearable test base. This physical-development work connects the analytical pipeline to the practical geometry and repeatability of a measurement setup.
Current status
The research is ongoing. Timothy has presented preliminary findings internally and implemented meaningful methodological capabilities, but no quantitative model-performance results are approved for public display. This page makes no claim of clinically validated performance.