A rheumatologist, neuroradiologist, and computer scientist have joined forces to blend medical expertise with AI/machine learning to reimagine the future of lupus diagnosis and care.
Multimodal approaches for diagnosing autoimmune diseases are nascent, and most focus only on omics data or electronic health records, ignoring other sources of informative patient data. In this innovative project, entitled Automated Reasoning & Interpretation for Early Lupus Detection (ARIEL), the multidisciplinary team is developing a cohesive, data-driven approach for early, and efficient diagnosis of lupus through multimodal analyses.
According to Dr. Saira Sheikh, who is the Lead Principal Investigator of the project, “Our approaches will significantly advance multimodal machine learning in general, and for autoimmune diseases specifically, tackling a highly relevant disease like lupus and allowing us to identify and accelerate care for those at greatest risk for poor outcomes.”
Dr. Saira Sheikh, the Linda Coley Sewell Distinguished Professor of Medicine at UNC, is a nationally renowned physician, researcher, and clinical trials investigator. She is trained and triple board certified in Internal Medicine, Rheumatology and Allergy & Immunology and is the Director of Clinical Trials at the UNC Thurston Arthritis Research Center. She also serves as the Chair of the Lupus Clinical Investigators Network, which is the premier and largest network of lupus clinical trial centers across the United States and Canada, managed by Lupus Research Alliance’s affiliate Lupus Therapeutics, LLC. At UNC, Dr. Sheikh leads a robust research team that focuses on answering scientific questions that directly impact the care of patients with lupus.
Dr. Yueh Lee, Co-PI, Professor and Vice Chair of Translational Research in Radiology at UNC has expertise in translational research, combining his clinical training as a diagnostic radiologist and technical training in the physics of medical imaging. He brings his extensive expertise in building bridges between engineering and clinical medicine to the project. His focus is on developing and translating technologies with the potential to make a significant patient and societal impact
“This project demonstrates the critical importance of close collaborations between clinicians and computer scientists in the development of the next generation of diagnostic tools for advancing human health, and the importance of a biomedical research environment that fosters collaboration,” states Dr. Lee
Dr. Marc Niethammer, Co-PI, is the Halıcıoğlu Endowed Chair in Health AI and a Professor of Computer Science & Engineering as well as Neurological Surgery at UC San Diego. He is also a Research Professor in Computer Science at UNC Chapel Hill. Dr. Niethammer leads the Biomedical Image Analysis Group. He is a leader in the field of medical image computing and machine learning approaches for biomedical data. His research is highly interdisciplinary and heavily motivated by solving medical problems with direct relevance for clinical care. His focus is on leading the technical developments of the project and interfacing with the clinical leads.
“Our project brings together an amazing team of computer scientists with experts in lupus and radiology to develop open-source, patient-oriented algorithms for better patient care. Our approaches will be applicable to a wide variety of diseases and leverage any available clinical data type,” explains Dr. Niethammer.
The team believes that their approach could be transformative by improving the diagnostic abilities of primary care physicians as well as subspecialists, to enable better health outcomes for patients far and wide. While the focus is on lupus as the driving biological problem, the computational approaches they aim to develop will be applicable to other autoimmune diseases and beyond. Artificial Intelligence has the potential to revolutionize the early diagnosis of lupus by analyzing vast amounts of clinical and imaging data to detect subtle patterns that may be missed by traditional diagnostic methods.
“This grant is truly a multidisciplinary and collaborative team effort, bringing together our exceptional clinical and technical teams. Given the complexity and variability of lupus symptoms, which often mimic other conditions, AI-powered tools such as machine learning algorithms and predictive models can help identify early indicators of the disease with greater accuracy and speed. This can lead to reduced misdiagnosis rates, faster interventions, and improved long-term outcomes for patients. As AI continues to evolve, its integration into clinical workflows holds promise for transforming lupus diagnosis from a lengthy, uncertain process into a more precise and timelier one,” remarks Dr. Sheikh
A strength of this project is the incorporation of community partner engagement throughout every step of the process, ensuring that the tools & data outputs are grounded in real-world needs & applications, shaped by those who receive & deliver care. According to Dr. Sheikh, “The voices and needs of patients & clinicians inform all our technical developments. This ensures that our tools are relevant and clinically meaningful.”
Other key members of the Sheikh Research team on this project that bring extraordinary experience and expertise are Tessa Englund, PhD, MPH, (Senior Research Scientist) and Claire Timon, MS (Social Clinical Research Assistant). Becki Cleveland, PhD, Director of Biostatistical Operations at NC TraCS is also a key project partner.
The technical team also includes an exceptional group of computer scientists at UNC. Dr. Mohit Bansal, PhD, John R. & Louise S. Parker Distinguished Professor at UNC brings expertise in natural language processing and multimodal machine learning. Hongtu Zhu, PhD, Kenan Distinguished Professor, brings expertise in statistical learning, medical image analysis, precision medicine, biostatistics, artificial intelligence, and big data analytics. Junier Oliva, PhD provides strong machine learning experience with a focus on understanding data at an aggregate, holistic level. Tianlong Chen, PhD brings expertise in building accurate, trustworthy, and efficient machine learning systems. Kelyne Kenmogne (UNC RENCI) is the project manager.
Year 2 of this award also incorporates additional project partners focused on developing federated learning approaches for cancer clinical trial matching. This includes Ghulam Rasool, PhD at Moffitt Cancer Center, Hongfang Liu, PhD and Liwei Wang, PhD at The University of Texas Health Center at Houston, Wei Zhang, PhD at Atrium Health Wake Forest Comprehensive Cancer Center and Jihad Obeid, MD and Paul Heider, PhD at the Medical University of South Carolina.
This NIH OT award (OT2OD038045-01) is supported by the Office of the Director (OD) and the Office of Data Science Strategy (ODSS)