2023 ARCHES Projects
Spring 2023 Project Awards
STREAM-ED: Simulation to Refine, Enhance and Adapt Management of Emergency
- Hyojung Kang, PhD, University of Illinois Urbana-Champaign
- William Bond, MD, OSF HealthCare
This study aims to develop practical models combining machine learning, discrete event simulation, and optimization techniques to improve emergency department (ED) resource utilization and address ED overcrowding, which is exacerbated by the COVID-19 pandemic and staffing shortages.
Prototype: Intelligent Regulatory Change Management System
- ChengXiang Zhai, PhD, University of Illinois Urbana-Champaign
- Scott Lowry, MHA, CHC, CCEP, OSF HealthCare
This study proposes an Intelligent Regulatory Change Management (IRCM) System that uses natural language processing and artificial intelligence to track and evaluate public policy actions governing OSF HealthCare. This will enable compliance professionals to identify critical changes and determine appropriate courses of action, reducing manual review and improving quality, safety, privacy risk management and efficiency.
Machine Learning of Standardized DICOM Metadata from Imaging Datasets
- Matthew Bramlet, MD, OSF HealthCare
- Brad Sutton, PhD, University of Illinois Urbana-Champaign
The project aims to develop a machine learning-based algorithm that can categorize image parameters directly from signal intensity variations of 2D medical images to enable efficient pipelines for medical image segmentation. The proposed algorithm is expected to estimate patient and image-acquisition information by utilizing machine learning methods in situations where the DICOM header fields are incomplete or unreliable, ultimately allowing for automated characterization of unknown 3D DICOM imaging datasets.
Machine-Guided Staging of Neuroblastic Tumors of Patient Specific 3D Models
- Daniel Robertson, MD, OSF HealthCare
- Brad Sutton, PhD, University of Illinois Urbana-Champaign
The OSF HealthCare Children’s Hospital of Illinois is using segmentation services to create 3D models of neuroblastic tumors for pre-surgical planning. The hospital aims to transition from 2D imaging to 3D modeling to increase the reproducibility of staging analysis, establish a new standard for segmented models of neuroblastic tumors and develop machine-guided tools that can improve upon and automate current recommended image-defined risk factors staging.
Toward Machine-Learned Aortic Arch Measured Diameters
- Matthew Bramlet, MD, OSF HealthCare
- Brad Sutton, PhD, University of Illinois Urbana-Champaign
The original project aims to automate the segmentation and clinical measurement of aortic arch diameters from MRI imaging. The researchers leading this project have successfully completed several steps, including de-identification and curation of datasets, manual segmentation and the development of a novel method for automatically analyzing each aortic arch with promising results, indicating correlation between the automated and clinically derived measurements.
A Field Experiment to Evaluate the Efficacy of Convenient Health Kiosks
- Ann Willemsen-Dunlap, CRNA, PhD, OSF HealthCare
- Ujjal Mukherjee, PhD, University of Illinois Urbana-Champaign
This proposal outlines a field experiment to evaluate the efficacy of health kiosks supported by community health workers (CHWs) in delivering first line preventive health screenings to rural and underserved communities. The project is intended to lead to large-scale development and deployment of health kiosks with the goal of positively impacting social determinants of health and long-term health status of those served.
Contextualizing Nursing Needs for Development of Retention-Support App
- Ann-Perry Witmer, PhD, University of Illinois Urbana-Champaign
- Sheryl Emmerling, PhD, Rn, NEA-BC, OSF HealthCare
The goal of this project is to address the high turnover rate of new nurses by providing a digital app that offers personalized nursing support. The Contextual Engineering (CE) paradigm will be used to assess the needs and values of first-year nurses, including those who have left their positions, to inform the development of the app in the first phase of the project, with the goal of stabilizing the nursing staff, improving the quality of service and reducing operating costs.
Community Health Café: Engaging Digital Innovation and Community-Based Resources to Enhance Health Equities in Underserved Communities
- Scott Barrows, MA, OSF HealthCare
- Joe Bradley, PhD, MA, University of Illinois Urbana-Champaign
The purpose of the community health café is to provide digital access to health and health care resources, including links for assistance with the social determinants of health, health education and connections to public health in underserved communities. The eventual goal is a Medicaid telemedicine option with OSF OnCall. This proposal aims to address critical needs of underserved residents in vulnerable communities and is crucial for their health.
AI-Powered Brain Tumor Segmentation
- Zhi-Pei Liang, PhD, University of Illinois Urbana-Champaign
- Matthew Bramlet, MD, OSF HealthCare
This project aims to enhance the detection and monitoring of brain diseases. Phase 1 of the project focuses on accurate delineation and segmentation of brain tumors using a combination of structural and molecular multimodal brain imaging data and deep learning. The proposed work includes developing brain atlases for AI-powered brain image analysis, computational tools for automated tumor detection and segmentation and evaluating potential clinical applications.
Optimizing Pharmacologic Management of Behaviors in Patients with Autism
- Adam Cross, MD, FAAP, OSF HealthCare
- Ravishankar Iyer, PhD, University of Illinois Urbana-Champaign
This proposal aims to provide physicians with a machine learning model that assists in selecting appropriate medication and dosage strategies for patients with Autism Spectrum Disorder (ASD). By incorporating patient history, genetic information and clinician notes, the model will dynamically adapt the treatment protocol as the patient progresses, ensuring optimal choices for improved behavioral symptoms with a high degree of confidence.
Predicting Medication Non-Adherence in Type 2 Diabetes
- Hyojung Kang, PhD, University of Illinois Urbana-Champaign
- Mary Stapel, MD, OSF HealthCare
Medication adherence is crucial for managing diabetes, but disparities exist, particularly among racial/ethnic minorities and those with lower socioeconomic status. This proposal aims to use data-driven models to identify high-risk individuals and areas for non-adherence to diabetes medication, develop and validate prediction models and implement and evaluate them in clinical practice.
Knowledge Graph Construction with Large Language Models to Predict DKA Occurrence and Severity
- Adam Cross, MD, FAAP, OSF HealthCare
- Jimeng Sun, PhD, University of Illinois Urbana-Champaign
Diabetic ketoacidosis (DKA) hospitalizes over 50,000 American children annually, with underprivileged and underserved children at higher risk. This proposal aims to develop a predictive model using patient-specific knowledge graphs generated from clinical data extracted through name entity recognition and language modeling. Clinicians can use the model to identify high-risk diabetic patients and prevent DKA.