Studies

Active and completed research across the Shepherd Research Lab’s three focus areas — cancer detection, body composition, and AI for health. Each study below links to a detailed page with aims, team, funding, and key publications.

Breast Cancer Risk Assessment in Hawaiʻi and the USAPI

U54 Grant #5U54CA143727. A UH Cancer Center / University of Guam partnership studying cancer health disparities among Pacific Islanders since 2003.

Makawalu Study

Portable handheld ultrasound paired with AI detection algorithms for breast-cancer screening in the Pacific — addressing advanced-stage cancer rates in areas without mammography access.

TBDXA.I. — Deep Learning on Total-Body DXA

Self-supervised deep learning applied to Health ABC Study DXA scans to predict cardiovascular disease, mortality, cancer, hip fracture, and metabolic outcomes.

Three-Compartment Breast (3CB)

Dual-energy compositional breast imaging using lipid-protein-water signatures to distinguish malignant from benign lesions and reduce unnecessary biopsies.

SMART — Melanoma Assessment

Deep-learning computer-vision methods to triage skin lesions for biopsy, increasing diagnostic vigilance for benign lesions across multiethnic populations.

Astro3DO

Space-feasible body-composition and body-shape analysis for long-duration space missions — quantifying astronaut fragility and fracture risk via 3D optical scans.

Accessible Breast Cancer Screening in the Pacific (iBE)

Pre-pilot study of the iBreastExam, a hand-held electronic palpation device, as an early-detection tool for women of Pacific Islander ancestry. Completed.

Samsung Bioimpedance System Evaluation

Validation of novel Samsung wrist-worn smartwatch bioimpedance devices against DXA and reference BIA systems. Completed.

Da Kine Body Composition

How hydration status and total body water vary across athletes and age groups — developing correction factors for the four-compartment model. Completed.

Stay in the loop

New publications, study announcements, and research updates — occasional, no spam.