Erik Hoyer, vice chair for quality, safety and satisfaction for the Department of Physical Medicine and Rehabilitation at Johns Hopkins, is the site principal investigator overseeing a multisite study, funded by the National Institutes of Health, that aims to develop electronic health record (EHR)-based prognostic indices for hospitalized older adults. These indices will predict six-month and two-year mortality rates to aid in hospice and palliative care referrals and medication deprescribing for patients with and without Alzheimer’s disease and related dementias (ADRD). The project, ending in September 2029, will create models for both EHR integration and ePrognosis, a web-based compendium This work addresses the gap in current prognostic tools for hospitalized older adults, particularly those with ADRD, by incorporating factors such as social determinants of health and functional status data.

Erik Hoyer is also the principal investigator of a project to develop a modern, data-driven system to assess fall risks among older hospitalized patients. The project, enabled by a grant from the Doctors Company Foundation, focuses on identifying novel fall risk factors and dynamic clinical influences on fall risk, such as mobility changes. The goal is to create an automated fall risk assessment tool integrated with electronic medical records to enhance clinical decision-making and reduce hospital falls. The project will end in late fall 2025.

Johns Hopkins physiatrist Erik Hoyer is a co-investigator for research aimed at determining the impact of physical therapy on discharge disposition and care costs for acute hospital inpatients. From January 2023 to July 2024, researchers are investigating how therapy treatment timing and frequency influence functional change and discharge outcomes. Using target trial emulation, the study seeks to provide insights into the optimal allocation of physical therapy resources in hospitals, informing value-based care decisions and enhancing patient care efficiency. This project is funded by the Magistro Family Foundation Research Grant.

Stroke often impairs many aspects of movement, from fine motor control of the fingers to complex whole-body activities such as walking. A major hurdle in understanding and treating motor impairments after stroke has been the inability to collect objective, quantitative movement data in the home or clinic. Clinicians and researchers rely on subjective clinical scales, expensive technology with limited accessibility (e.g., motion capture systems) and other technologies with limited capability (e.g., wearables, instrumented gait mats) for motor assessment. These approaches impose significant limitations on the frequency and granularity of the data that can be collected, precluding a comprehensive understanding of patient-specific motor deficits and deeper insights into timelines of post-stroke recovery during rehabilitation.

Recent developments in artificial intelligence and machine learning (i.e., the advent of “pose estimation” technologies) have made it possible to track human movement with minimal financial cost, time investment and technological requirements. Pose estimation algorithms can rapidly track movement of key features of the body (e.g., shoulders, elbows, fingers) from simple digital videos that are easily recorded using a smartphone or tablet. These approaches have the potential to greatly expand the ability to track and measure motor impairments after stroke in any setting (including the home or clinic) with very little costs of time, money or effort.

The goal of this proposal, led by Ryan Roemmich at Johns Hopkins, is to enable a novel approach for quantitative, video-based movement analysis of the upper extremity after stroke. A state of the art, open-source pose estimation algorithm will be used to perform video-based assessments of movement post-stroke. Two significant advances are anticipated: 1) development and validation of new video-based assessments of upper extremity and fine motor control in people post-stroke against both ground-truth measurements and clinical standards, and 2) new, freely available workflows that make it possible to implement this approach using only simple videos recorded using common household devices. The ultimate goal of this proposal is to make it faster and easier to obtain accurate, quantitative assessments of upper extremity motor function after stroke.