Precision-Medicine Diagnostic Support in Hospitalized Veterans with Acute Kidney Injury.

Acute kidney injury (AKI), defined as a sudden loss of kidney function, affects up to 20% of hospitalized Veterans and is strongly associated with chronic kidney disease, poor quality of life, and death. Clinical practice guidelines recommend timely identification of the cause of AKI; however, few patients receive a standard diagnostic evaluation. Some causes of AKI require specific treatments beyond supportive care (i.e., treatment-specific AKIs (tsAKI)), can be challenging to diagnose, and can require even more detailed evaluation including kidney biopsies. Clinical decision support systems (CDSS) have shown early promise to address these barriers; however, impact on clinical outcomes have been minimal, with most limited to systems that alert providers to the presence of AKI and provide general supportive recommendations at a single point in time. Significance to VA: AKI occurs in 200,000 hospitalizations/year in the VA, and Veterans are at elevated risk for poor AKI-related outcomes. Facilitating timely diagnosis and treatment for AKI can have a significant impact on the burden of kidney disease in the Veteran population. Innovation and Impact: The proposed work aligns with and complements federal objectives to improve precision-based approaches to kidney disease by targeting unaddressed areas of clinical phenotyping and personalized care. We will be among the first to advance the science of CDSS in AKI by using a combination of approaches that 1) link patient-level data to provide iterative cognitive support to a variety of providers, 2) advance the VA’s capacity for translation of AI- research into clinical practice, and 3) improve understanding of processes of care provided to different types of AKI. Specific Aims: We will develop and test an Artificial Intelligence (AI)-assisted automated and comprehensive precision CDSS tool (PRECISE-AKI) to provide iterative cognitive support of general and nephrology-based providers in the diagnostic evaluation of AKI. To accomplish this, we will: 1) identify facilitators, barriers, and user preferences in information content and appropriate workflow integration for providers managing Veterans with AKI, 2) develop and validate machine learning algorithms to predict the likelihood of tsAKI etiologies that have differential treatment considerations, 3) develop a clinical workflow support tool to facilitate iterative diagnostic and treatment decision-making for inpatient providers caring for Veterans with AKI, and 4) evaluate the PRECISE-AKI CDSS tool in a feasibility and usability pilot study. Methodology: PRECISE-AKI’s user interface will focus on 3 diagnostic decision points in AKI evaluation: 1) initial identification of patients with AKI to improve rates of standard evaluation and screen for possible tsAKIs, 2) guide further diagnostic evaluation and nephrology referral among patients with markers of AKI severity or where initial diagnostic evaluation suggests the potential for tsAKIs, and 3) provide cognitive support to nephrologists that can provide predicted probabilities of the likelihood of tsAKIs. Our approach will leverage a wide array of clinical data, including the national clinical data and the VA experience with AKI-related kidney biopsies, and apply best practices in human-computer interaction and implementation science to assess user preferences in information content and workflow integration. We will execute development within a framework of iterative information visualizations and developed through user-centered design while ensuring assessments for bias and fairness. Path to Translation/Implementation: Should the findings be promising, we will pursue a multi-site implementation trial. We have partnered with local clinical service champions, the National Artificial Intelligence Institute, the VHA National Kidney Health Program, and the American Association of Kidney Patients Veteran Health Initiative to facilitate expanded implementation and future use adaptations. Project Number: 1I01RD000654-01A1 | Fiscal Year: 2026 | NIH Institute/Center: Veterans Affairs (VA) | Principal Investigator: Edward Siew (+1 co-PI) | Institution: VETERANS HEALTH ADMINISTRATION, NASHVILLE, TN | Activity Code: I01 | Study Section: HSR-3 Healthcare Informatics & Access to Care[HSR3] View on NIH RePORTER: https://reporter.nih.gov/project-details/11243334