Overview

PDM research develops and evaluates the tools of digital medicine: sensors, algorithms, and care models. Our work spans three areas.

RESEARCH AREAS

Image source: Adobe Stock

AREA 01

Care outside brick-and-mortar facilities.

Remote monitoring with active and passive sensors (eyes on the patient), point-of-care evaluation and diagnostics (hands on the patient), and data platforms and analytics for clinical decisions (brains on the patient). We have built, implemented, and evaluated care-at-home programs across hospital, rehabilitation, and chronic disease settings.

AREA 02

Artificial intelligence in medicine.

PDM-developed AI models outperform expert detection for major forms of gallbladder and pancreatic cancer during endoscopy, work led by Dr. Neil Marya. Our ECG-Age model estimates biological age from a standard 12-lead ECG; accelerated aging on this measure is associated with higher risk of hospitalization and death among patients with chronic disease. Additional work includes AI biomarkers for earlier detection of cognitive impairment (Dr. Honghuang Lin), risk stratification algorithms (DECOMP, SUGAR), and smartwatch-based detection of abnormal heart rhythms.

Image source: Adobe Stock

Image source: Adobe Stock

AREA 03

Learning health system.

We evaluate whether innovations actually matter, for whom, and at what cost, and we feed those findings back into clinical decision-making. PDM has evaluated the clinical and cost impact of programs including specialty pharmacy, diabetes care coaching, and tele-pulmonary rehabilitation.

Data Assets

Five years of digital studies and clinical partnerships have produced a research data infrastructure that is unusual for an academic program of this size. The ability to combine EHR data, patient-generated health data, and multi-omics data for multi-modal analysis is a unique asset of PDM.

Clinical research cohorts built on the OMOP common data model, including NExUS (~510,000 participants, internal medicine and emergency care), an acute care cohort (~709,000), an Alzheimers and related dementias cohort (~1.5 million), and disease-specific cohorts in COPD, psychiatry, rheumatology, and more. ECG repository supporting the ECG-Age biomarker program: more than one million 12-lead ECG waveforms linked to longitudinal clinical outcomes. Endoscopic video archive: approximately 100,000 GI procedures recorded through the VIRGO platform, plus nearly 200 bronchoscopy and EBUS procedures, supporting AI development. Recorded cholangioscopy data has been licensed by Boston Scientific alongside PDM-developed AI. Digital study platform data: 16+ decentralized studies with survey, wearable, and sensor data streams from Fitbit, Garmin, HealthKit, Google Fit, and study-specific devices, spanning COVID-19, COPD, infant sleep, maternal health, rural health, and healthy aging.

COLLABORATIONS

UMass Memorial Health

Primary clinical partner for development, validation, and deployment of digital health technologies

VIRGO & Boston Scientific

Endoscopic video AI development and licensing

Morningside

Partnership with a portfolio of digital health companies for rapid demonstration projects and impact evaluation.