Agenda for the 2025 Future Trends of mRehab Sessions
Timed presentations
Session 1. Introduction and Overview of Issues, Challenges and Opportunities for Mobile Rehabilitation (10 minutes)
Mike Jones, PhD, Director Emeritus, Crawford Research Institute, Shepherd Center
John Morris, PhD, Research Scientist, Shepherd Center
Session 2. Empowering Therapist’s Decision Making with Their Patients’ Data: Using Home Exercise Program Data to Inform Clinical Care (50 minutes with Q&A)
Raeda Anderson, PhD, Research Scientist, Shepherd Center
Dalise Robinson, MS, CCC-SLP, Outpatient Therapy Manager, Acquired Brain Injury, Shepherd Center
Home exercise programs (HEP) are a key component of rehabilitation recovery assigned by physical, Occupational, and Speech Therapists for patients to build skills and stamina over their rehabilitation journey. Often, outpatient acquired brain injury patients struggle to remember their HEP assignments and inaccurately report adherence. This study generated a clinician-driven interactive dashboard updates daily to display HEP assignment and adherence to inform clinicians on their patient’s assignment and adherence to HEPs, allowing for a clearer understanding of patient activity and progress. This presentation will cover how this project developed over time from a scientific study to implemented for all outpatient brain injury patients at Shepherd Center.
Break: 10 minutes
Session 3. RehaBot: Artificial Intelligence-Enabled Chat to Support Behavior Activation by People Living with Brain Injury (50 minutes with Q&A)
Amanda Rabinowitz, PhD, Associate Director & Director of the Brain Injury Neuropsychology Lab, Jefferson Moss Rehabilitation Research Institute
This presentation will review the concept and characteristics of conversational agents (or, chatbots, service bots, AI chat). We will then discuss our experiences co-designing our conversational agent (Rehabot) to support behavior activation (BA) in persons living with the effects of moderate to severe brain injury.10,11 A conversational agent is a piece of software that can carry on a conversation with a user. Chatbots are able to handle complex multistep interactions, such as making hotel reservations or ordering coffee. Advances in Natural Language Processing (NLP) have made conversational interactions ever more flexible and natural,12,13 and thereby potentially more useful for supporting community dwelling individuals with brain injury and other debilitating conditions. This presentation will discuss our use case, initial designs, engaging end-users in the co-design process, validation, testing and deployment of Rehabot. We will conclude with an interactive discussion of rapid advances in artificial intelligence and large language models (LLMs) that offer expanded opportunities for designing interventions using AI chat.
Break: 10 minutes
Session 4. Smart Coach: Predicting Patient Performance and Engagement Using Large Activity Data Sets (50 minutes with Q&A)
David Reinkensmeyer, Professor of Engineering, University of California, Irvine
George Collier, Senior Data Scientist, Shepherd Center
Sangjoon Kim, Principal Investigator, Korea Institute of Science and Technology
This presentation introduces the concept of exercise “perseverance” and habit formation to understand the dynamics of behaviors that can be supportive or not supportive of adherence to home exercise prescriptions.14 Research on exercise perseverance in the general public has suggested the importance of early exercise frequency and schedule consistency (in terms of which days of the week exercises are performed) because they encourage habit formation. After reviewing the literature on exercise perseverance and habit formation, we will discuss our own research on exercise perseverance using a large dataset of community dwelling individuals living with the effects of stroke. We will review models for grouping users based on the timing and frequency of their exercise patterns across a six-month period. Our models of patient types can be used to predict the “decay” of exercise perseverance (decline in exercise frequency and consistency over time) that all patients experience, even those with the best formed exercise habits. Our research may help in designing effective home rehabilitation programs after stroke.
Lunch: 60 minutes
Session 5. Transformative AI Data Capture and Reporting: Informing Clinical Decision Making with Remote Activity and Health Metrics (50 minutes with Q&A)
Brad Willingham, PhD, Director of Multiple Sclerosis Research, Shepherd Center
George Collier, Senior Data Scientist, Shepherd Center
Jacob Cartwright, Data Scientist, Shepherd Center
This educational session is divided into two parts, highlighting the technical infrastructure and applied intelligence that enable clinicians to utilize complex remote data for rehabilitation decision-making. The first part introduces the foundational elements of Remote Therapeutic Monitoring (RTM) systems and focuses on how enterprise cloud technologies, such as secure data warehouses, enable the integration of multimodal data from discrete remote monitoring sources. We will discuss how physiologic, behavioral, and self-reported data from wearables, patient-reported outcome platforms, tele-rehabilitation apps, and EHRs can be securely harmonized and organized to support real-world rehabilitation research and care delivery. Clinical rationales (e.g., care continuity, intervention timing) for unifying disparate data streams will be highlighted. The second part focuses on the application of artificial intelligence, particularly large language models (LLMs), to synthesize complex datasets into concise, clinically meaningful summaries. We define the role of LLMs in rehabilitation and explore how they differ from traditional analytic tools in their ability to reason across multimodal inputs and generate context-aware, decision-supportive content. In addition to demonstrating clinical utility, we will address key considerations for implementation, including data privacy, security safeguards, and access controls that protect sensitive health information. We also outline transparency and safety guardrails, such as human-in-the-loop validation, that are essential for building clinical trust and ensuring appropriate use of AI-generated outputs.15,16
Break: 10 minutes
Session 6. From Lab to Real Life: Translating Research Innovations into Clinical and Commercial Impact (50 mins with Q&A)
Deborah Backus, PT, PhD, FACRM, VP, Research & Innovation
To fully realize the potential of novel technology-enabled mobile rehabilitation interventions, it is essential to translate research and technology advances into meaningful clinical programs that are practical and impactful for patients and providers.17,18 Successful integration into healthcare systems requires thoughtful implementation that prioritizes clinical relevance, cost-effectiveness, institutional technological support, and streamlined workflows to foster adoption and sustainability across diverse clinical programs.19,20 This presentation reviews challenges and opportunities for validation, adoption, scale-up implementation, spread, and sustainability of complex digital innovations. The presentation will highlight the challenge of driving technology-enabled innovation in complex clinical rehabilitation settings due to the latter’s dynamic and adaptive nature.21
Session 7. Panel Discussion and Wrap-up: The Future of Mobile Rehabilitation for People with Disabilities (50 minutes)
Moderator: Mike Jones, PhD, Director Emeritus, Crawford Research Institute, Shepherd Center
This interactive session will address some of the major technological and clinical rehabilitation trends and future directions for mobile rehabilitation. Topics include: exercise adherence, artificial intelligence, conversational agents, large language models, remote therapeutic monitoring (RTM), remote physiologic monitoring (RPM), wearable and environmental sensors, co-design, clinician buy-in to new tools for patient engagement, and implementation of new technology-enabled interventions into clinics and hospital systems.
Learning Objectives
Identify socio-economic, policy and technology trends supporting the development and implementation of mobile rehabilitation solutions.
Discuss processes for engaging rehabilitation clinicians in the development and adoption of dashboards for patient activity data.
Identify the opportunities for supporting patients with brain injury to engage in at-home physical activity using AI-enabled chat routines.
Discuss the distinction between open-AI and closed-AI systems. Discuss the comparative performance of major commercial AI solutions in summarizing patient electronic medical records.
Discuss the distinction between perseverance and adherence to home exercise programs and identify patient types based on early performance.
Define the role of large language models (LLMs) in clinical reporting and list 3 ways they improve upon conventional data processing methods.
Explain 2 AI safety guardrails (e.g., transparency, human validation) that support the clinical appropriateness of AI-generated summaries.
Identify 3 workflow efficiencies clinicians can gain from using LLM-generated reports in rehabilitation practice.
Identify challenges and opportunities for implementing and sustaining innovative mRehab interventions into clinical practice.