iCAMP role: 31 subjects and 800+ hours of chest-worn wearable data.
Biomedical sensing
Wearable Physiological Data Analysis
iCAMP analysis of 800+ hours of chest-worn ECG/accelerometer data across 31 subjects, supporting fall-risk modeling from 0.73 baseline AUC to 0.969 mixed-model AUC.
Part of Data Analyst · iCAMP Research Group, University of Arizona · Sep 2014 – Jan 2017
completed2014–2017WearablesPhysiological dataECGAUC
Typedata analysis
SkillsECG/HRV, Accelerometers, Feature engineering, Prediction workflows
Selected facts
Quantitative details and source-backed proof points.
Fall-risk/frailty model reference: 0.73 baseline AUC, 0.921 posture/activity AUC, and 0.969 mixed-model AUC.
Workflow included ECG filtering, R-wave detection, R-R interval correction, HRV features, accelerometer calibration/reorientation, posture/activity classification, labeling, and prediction support.
Project summary
Why it exists, what I built, and what I learned.
Why I built it
Wearable physiological studies needed reliable feature extraction, calibration, labeling, and model-support workflows from noisy human sensor data.
What I built
ECG filtering, R-wave detection, R-R interval correction, HRV feature generation, accelerometer calibration/reorientation, posture/activity classification, labeling, and prediction support.
What worked
The pipeline converted raw chest-worn ECG and acceleration into model-ready physiological, posture, and activity features.
What failed
Human wearable data required careful calibration, labeling, windowing, and artifact handling before model metrics were meaningful.
What I learned
Physiological modeling depends on signal preparation and context labeling as much as the downstream classifier.
Stack
Tools, systems, and technical areas involved.
ECGHRVAccelerometryR-wave detectionFeature extractionLabelingAUC analysis
Links and direction
Public links and next steps.
Use this page as early evidence for wearable physiological data and feature-engineering experience.
Related projects
Other projects in the same neighborhood.
Medical devices
FlowSense Clinical / ACE
Clinical wearable and DSP/ML workflow for noninvasive CSF shunt-flow assessment; improved from a 74% baseline to 0.81 AUC / 82% blinded validation accuracy.
Medical devicesFlowSense Home / Lynx
Home-use hydrocephalus wearable advanced across 3 design generations, 100+ units, 200+ participants, 2,800+ wear hours, and 70% → 96% reliability improvement.
Medical devicesWound Monitoring Platform / Tabby
NIH R43 Phase I multimodal wound patch with thermal/humidity/temperature sensing, 40 mAh battery, 20-subject dataset, 172 logs, and 5-fold wound-model validation.