Loading component...
Loading component...
View More In This Section
Back to Page
AN OSF HEALTHCARE AND UNIVERSITY OF ILLINOIS COLLEGE OF MEDICINE PEORIA COLLABORATION
The Jump Applied Research for Community Health through Engineering and Simulation (ARCHES) program is a partnership between OSF HealthCare, the University of Illinois Urbana-Champaign (UIUC) and the University of Illinois College of Medicine in Peoria (UICOMP).
The funding supports research involving clinicians, engineers and social scientists to rapidly develop technologies and devices that could revolutionize medical training and health care delivery.
In alignment with the strategic focus on improving patient and family care across key health care areas, the Jump ARCHES program has placed an enhanced emphasis on four critical strategic areas for development:
Research grants have been awarded to the following projects:
Yang Zhao, PhD, University of Illinois Urbana-Champaign
Andrew J Tsung, MD, OSF, University of Illinois College of Medicine Peoria
This project develops a novel, noninvasive electric‑field–based therapy for glioblastoma multiforme (GBM) by introducing chiral fields as an alternative to conventional tumor‑treating fields. The study will first define optimal chiral field parameters that suppress chemotherapy‑resistant GBM growth in vitro using an imaging‑compatible quantitative platform. Building on these findings, a wireless, wearable chiral field device will be developed and evaluated in an orthotopic GBM mouse model, with treatment response assessed using high‑resolution ultrasound localization imaging. By integrating device engineering, quantitative imaging, and clinically relevant tumor models, this work establishes a systematic and translatable framework for electric‑field cancer therapy and generates proof‑of‑concept data to support future clinical translation and commercialization.
Avinash Gupta, PhD, University of Illinois Urbana-Champaign
Megan Kupferschmid, MSN, RN, CCRN, NE-BC, OSF HealthCare
This Phase I project develops and evaluates an AI‑enhanced extended‑reality training platform for pediatric peripheral intravenous catheter (PIVC) placement to address persistent challenges in first‑attempt success and limited access to pediatric expertise. Building on prior pilot work, the study will refine an existing VR module and integrate an AI‑guided trainer that delivers real‑time procedural feedback, structured debriefing and objective performance metrics. Face and content validity will be established through expert evaluation, followed by longitudinal assessment with new graduate nurses focusing on knowledge retention, self‑efficacy and system‑captured performance indicators. In parallel, a mixed‑reality module will be designed to support future comparative effectiveness studies. The platform aims to provide scalable, high‑fidelity pediatric vascular access training across diverse care settings.
Mashfiqui Rabbi, PhD, University of Illinois Urbana-Champaign
Brandyn A. Castro, MD, OSF HealthCare
This Phase I project develops and pilot-tests RecoverML‑Spine, an adaptive AI-driven digital recovery coach for patients undergoing lumbar spine surgery. The system delivers personalized, safety‑constrained behavioral coaching during the critical postoperative period to improve mobility, reduce opioid reliance and enhance functional recovery. A validated behavioral simulator will enable safe offline training of adaptive coaching policies using retrospective clinical data and large-scale wearable datasets. The integrated smartphone application combines clinician-designed rules, reinforcement learning, on-device language generation and multi-layer safety monitoring. A multi-site feasibility pilot will evaluate safety, usability, engagement and preliminary behavioral and clinical outcomes, establishing readiness for a randomized controlled trial and broader surgical applications.
Hyojung Kang, PhD, University of Illinois Urbana-Champaign
Roopa Foulger, OSF HealthCare
This project aims to transform type 2 diabetes mellitus (T2DM) management from reactive care to proactive, individualized intervention by integrating dynamic risk prediction with agentic AI–driven decision support. The study will develop a continuously learning, multi-label risk prediction system to detect concurrent behavioral, clinical and acute care deterioration using longitudinal clinical, behavioral and social determinants of health data. Building on these predictions, a multi-agent AI system will determine whether, when and how to intervene across tiered care pathways, with clinician oversight for high-risk decisions. Integration within a secure Epic sandbox will evaluate feasibility, clinical validity and workflow alignment, supporting scalable, equity-centered chronic disease management.
Andrew M. Smith, University of Illinois Urbana-Champaign
Jun Zhang, MD, PhD, OSF HealthCare, University of Illinois College of Medicine Peoria
This project aims to improve early detection and risk stratification of lung cancer by developing a noninvasive blood-based diagnostic using exosomal microRNA (miRNA) profiling from at-home capillary blood collection. Addressing high false-positive rates in low-dose CT screening, the study will identify miRNA signatures that distinguish malignant lung cancer from benign pulmonary nodules and fibrosis in high-risk patients. Capillary blood samples from approximately 200 screened individuals will be analyzed and integrated with imaging and clinical data to develop predictive malignancy models. The approach enables scalable, remote screening and longitudinal monitoring, with the potential to reduce unnecessary invasive procedures, improve diagnostic accuracy, and expand access to early lung cancer detection in underserved populations.
Sharifa Sultana, University of Illinois Urbana-Champaign
Jun Zhang, MD, PhD, OSF HealthCare, University of Illinois College of Medicine Peoria
This project addresses the care gap faced by cancer survivors aged 50 and older by developing an impairment‑aware digital health intervention tailored to overlapping cognitive, visual, motor and psychosocial challenges. Using a human‑centered, participatory design approach, the study will first identify impairment‑driven requirements through interviews and focus groups with survivors, caregivers and clinicians. These insights will inform co‑design of a high‑fidelity prototype that operationalizes accessibility and interaction design principles. Usability testing with the target population will validate and generalize a design framework for impairment‑aware digital health. The expected outcome is a scalable, patient‑centered framework and prototype that improve engagement, psychosocial well‑being and equitable access to survivorship support.
Manuel Hernandez, PhD, University of Illinois Urbana-Champaign
Christopher Zallek, MD, OSF HealthCare, University of Illinois College of Medicine Peoria
This project develops a multimodal foundation model (FM) of human gait to enable scalable, individualized assessment of neurological gait dysfunction across clinical and real‑world settings. A population‑scale FM will be trained on overground motion‑capture data from typically developing and pathological gaits to learn generalizable gait representations. Validation data will be collected from middle‑aged and older adults with neurologic and non‑neurologic gait using motion capture, wearables, computer vision and fNIRS. The FM will be adapted across sensing modalities and environments to classify gait patterns and monitor longitudinal progression using real‑world data. The approach aims to reduce reliance on specialized laboratories, expand access in rural settings, and improve fall‑risk assessment and monitoring of neurological disease progression.
Mashfiqui Rabbi, University of Illinois Urbana-Champaign
David G. Thompson, MD, FACC, OSF HealthCare
This project develops AI-augmented tools to improve lifestyle medicine goal-setting and adherence by leveraging digital twins built from wearable and mobile data. Using offline physical activity and dietary intake data, individualized digital twins will model patients’ routines to generate achievable, data-driven goals and optimized plans requiring minimal behavior change. These models will be translated into a provider-facing dashboard for collaborative goal and plan customization and a patient-facing mobile app delivering daily reminders, progress feedback and two-way communication. Feasibility will be demonstrated in a pilot with lifestyle medicine patients. The project addresses limitations of self-reported data and infrequent follow-up, enabling proactive, personalized lifestyle interventions and scalable integration of wearables into clinical care.
Jimeng Sun, PhD, University of Illinois Urbana-Champaign
Adam Cross, MD, OSF HealthCare, University of Illinois College of Medicine Peoria
This project aims to deploy and validate a high‑performance, equitable prediction pipeline for catheter‑associated urinary tract infections (CAUTIs) within OSF HealthCare. Leveraging PyHealth models trained on the MIMIC‑IV benchmark, the study will first establish a locally mapped CAUTI cohort from OSF EHR data and evaluate zero‑shot generalization across multiple model architectures. Where performance degradation due to distribution shift is observed, models will be retrained and rigorously audited for calibration and fairness across demographic subgroups. The project will deliver a reproducible, clinically auditable CAUTI prediction pipeline, peer‑reviewed dissemination and a foundation for Phase 2 clinical workflow integration and federal funding.
Brad Sutton, PhD, University of Illinois Urbana-Champaign
Matthew Bramlet, MD, OSF HealthCare, University of Illinois College of Medicine Peoria
This project develops a novel framework for constructing anatomically consistent 4D digital twins of cardiovascular anatomy from routine, retrospectively ECG‑gated cardiac CT imaging. The core innovation is a mesh‑matching approach that preserves strict node‑level correspondence across time, enabling accurate tracking of anatomical deformation throughout the cardiac cycle. Using this framework, the study will derive CT‑based left ventricular endocardial strain and evaluate its association with cardiac amyloidosis and will quantify aortic distensibility with spatially resolved 3D deformation maps. By transforming static CT datasets into dynamic, quantitative functional diagnostics, the project enables scalable, clinically actionable assessment of myocardial and vascular biomechanics without additional imaging burden.
Wawrzyniec Dobrucki, PhD, University of Illinois Urbana-Champaign
Matthew Bramlet, MD, OSF HealthCare, University of Illinois College of Medicine Peoria
This project develops a clinically deployable, AI-assisted liver imaging platform that integrates diagnostic-grade segmentation with volume-aware radiomic analysis to support surgical planning and longitudinal disease assessment. The study will create high-fidelity automated segmentations of liver parenchyma and hepatic tumors from CT imaging, enabling anatomically precise 3D modeling for clinical and VR-based workflows. Building on these segmentations, a tumor volume–aware radiomics framework will decouple biologically meaningful imaging features from size-related confounding, allowing robust characterization of tumor heterogeneity, liver parenchymal biology and tumor–parenchyma interactions. The platform will correlate volume-independent imaging biomarkers with clinical endpoints, enhancing reproducibility, interpretability and clinical decision support in hepatic oncology.
Avinash Gupta, PhD, University of Illinois Urbana-Champaign
Chris Zallek, MD, OSF HealthCare, University of Illinois College of Medicine Peoria
This project develops and evaluates a virtual reality (VR)–based robotic trail maker (RTM) platform to support scalable, patient-specific upper extremity rehabilitation for individuals with neurological conditions. The system integrates a 6‑degree‑of‑freedom robotic device with an immersive VR environment, enabling therapists to design customized tasks, trajectories and adaptive assistive forces. Human subject studies with healthy participants and patients will assess safety, usability, cybersickness, workload and performance while collecting standardized kinematic data. Exploratory artificial intelligence models will be trained on these data to characterize motor impairment and investigate automated task selection and assistance modulation. The platform aims to reduce therapist burden while maintaining high-quality, data-driven rehabilitation.
Brad Sutton, PhD, University of Illinois Urbana-Champaign
Matt Bramlet, MD, OSF HealthCare, University of Illinois College of Medicine Peoria
This project develops a fully automated, clinically integrated framework for objective characterization of renal tumor complexity from standard-of-care multiphase CT imaging. The study will generate high-fidelity, anatomically precise 3D renal digital twins through automated segmentation of kidney anatomy, tumors, collecting system and hilar structures. A segmentation-driven co-registration framework will integrate arterial, venous and delayed phases into a unified anatomical model. Building on this representation, automated algorithms will compute objective RENAL nephrometry scores, replacing subjective manual assessment. Validation against expert clinician scoring and clinical outcomes will assess reproducibility and decision impact. The platform aims to improve surgical planning, risk stratification, and treatment selection for localized kidney cancer.
Yiwen Dong, University of Illinois Urbana-Champaign
Christopher Zallek, MD, OSF HealthCare, University of Illinois College of Medicine Peoria
This project develops and translates an AI-powered system for personalized, interpretable visualization of patient motion during neurologic examinations. Addressing the limitations of time-intensive, subjective documentation, the system integrates motion analysis with vision–language models to generate patient-specific digital avatars that visually and semantically represent functional status. The platform enables before–after comparison of neurologic function using clinician-selected metrics, scores and narrative descriptions. An automated pipeline for exam recording, processing and reporting will be validated against clinician observations and pilot-deployed at OSF HealthCare. The expected outcome is a scalable solution that improves documentation efficiency, clinical decision-making, patient communication and longitudinal assessment of neurologic function.
Yiwen Dong, University of Illinois Urbana-Champaign
Tiffani Stroup Franada, DO, OSF HealthCare, University of Illinois College of Medicine Peoria
This Phase I project develops and validates video-derived digital biomarkers to sensitively track functional impairment progression in multiple sclerosis (MS). The study will first create a biomechanics-constrained gait reconstruction pipeline capable of accurate skeletal motion recovery from single-camera clinical videos with heavy occlusion. Using reconstructed motion, the project will extract interpretable digital biomarkers reflecting motor fatigue, cognitive burden and balance dysfunction—domains poorly captured by EDSS. Biomarkers will be validated in a pilot cohort of MS participants against clinician observations, standardized functional tests and patient-reported outcomes. The results will establish a patient-centered, noninvasive framework for scalable MS progression monitoring and inform Phase II longitudinal deployment.
Jimeng Sun, PhD, University of Illinois Urbana-Champaign
Shoji Samson, DO, OSF HealthCare, University of Illinois College of Medicine Peoria
This project aims to improve diagnosis and management of retinopathy of prematurity (ROP) by integrating artificial intelligence into neonatal care. Using longitudinal clinical data and retinal imaging, the study will develop robust predictive models to identify preterm infants at high risk for severe ROP and to model individualized disease progression. The project will quantify the contribution of key clinical risk factors and leverage deep learning to enhance diagnostic precision and clinical decision-making. By reducing subjectivity and resource burden in ROP screening, this AI-driven framework seeks to enable earlier intervention, optimize clinical workflows and improve visual outcomes for vulnerable preterm infants, aligning with Jump ARCHES priorities in innovative, scalable healthcare solutions.
Yiwen Dong, University of Illinois Urbana-Champaign
Adam Cross, MD, OSF HealthCare, University of Illinois College of Medicine Peoria
This project develops an interpretable, AI-driven clinical decision-support framework for pediatric neuromuscular care by integrating physiological measurements with clinical domain knowledge. A multimodal knowledge graph will be constructed to link quantitative data (e.g., gait kinematics, muscle activity) with pathophysiology, clinical observations and interventions. Building on this representation, a collaborative multi-agent system will perform evidence-based pathological analysis and treatment planning through iterative reasoning and critique. The framework will be validated in cerebral palsy cohorts using real-world gait and clinical data, comparing recommendations against expert decisions. The study aims to improve accuracy, consistency and transparency of personalized treatment recommendations and support scalable, data-informed clinical decision-making.
Harshal Mahajan, PhD, University of Illinois Urbana-Champaign
Megan Kupferschmid, MSN, RN, CCRN, NE-BC, OSF HealthCare
This Phase I project develops and evaluates an extended-reality (XR) teleoperation training platform to support safe, scalable use of assistive mobile manipulators in healthcare. The study focuses on preparing nurses to operate the Stretch robot for selected patient-support tasks through human-centered XR training. Using participatory design, the project will identify nursing workflow needs, target robot-assisted tasks, and performance requirements, then develop an XR training environment simulating realistic care scenarios. Usability, workload, confidence, safety behaviors, and task performance will be compared between XR-trained and conventionally trained nurses using the physical robot. Results will establish feasibility and inform broader clinical adoption and external funding.
