# Developing statistical analysis and prediction tools for continuous glucose monitoring (CGM) use in hospitalized patients with diabetes > **NIH NIH R01** · EMORY UNIVERSITY · 2024 · $370,716 ## Abstract Project Summary Diabetes mellitus is one of the leading causes of death and disability in the United States. With rapidly ris- ing incidences in the last two decades, diabetes has been reported in over 20% of hospitalized adult patients. Glycemic control plays an important role to help reduce hospital complications, mortality, and health care cost for these patients. To achieve optimal glycemic control, continuous glucose monitoring (CGM), which evaluates interstitial glucose every 1–5 minutes, offers many advantages over the traditional point of care (POC) capillary glucose testing before meals and at bedtime, including: a panoramic view of glycemic profiles, real-time detec- tion of hypoglycemia and hyperglycemia, remote glucose management, and lower care labor and cost. With the fast growing utilization of CGM for inpatient diabetes management, developing effective and robust statisti- cal methods tailored to translate the abundant data from CGM to sound clinical decisions is of timely importance; however, this area is largely unexplored. From closely working on a series of multi-center clinical studies that investigated the reliability, safety, and efficacy of CGM for inpatient use, we recognize substantial barriers for the existing analytic approaches for CGM data to meet this critical need. These include inadequately accounting for special data issues in inpatient CGM studies, and inefficient use of the rich information from CGM to inform more individualized diabetes care. Specifically, time-in-range (TIR), defined as the percentage of time that glucose readings are within a target glycemic range over a specified amount of time, is a key metric for evaluating glycemic control based on CGM. To evaluate TIR in the hospital setting, a prevalent issue is that some patients are discharged before sufficient CGM data are captured. As inpatient glucose excursion patterns can be highly variable over time, current data analyses that simply impute TIR based on the shorter, incomplete glucose monitoring can lead to biased inferences on TIR. Another notable caveat relates to the assessment of hypoglycemia and hyperglycemia, which plays an important role in treatment decisions. Existing analyses mainly use the counts of these events but waste the valuable timing information that is uniquely available in CGM studies. This leads to a missed opportunity for detailed individual profiling and dynamic prediction of hypoglycemia or hyperglycemia risk that can help guide customized inpatient care of diabetes. In this application, we aim to fill in these gaps through (i) developing rigorous statistical methods that are elegantly suited to thoroughly evaluate TIR and other similar key outcomes with the special data complications in hospital CGM studies properly addressed; (ii) broadening the paradigm of current CGM data analyses with a new framework of analyzing hypoglycemia and hyperglycemia outcomes that effectively utilizes the timing information and confers a much im... ## Key facts - **NIH application ID:** 10812815 - **Project number:** 1R01DK136023-01A1 - **Recipient organization:** EMORY UNIVERSITY - **Principal Investigator:** Limin Peng - **Activity code:** R01 (R01, R21, SBIR, etc.) - **Funding institute:** NIH - **Fiscal year:** 2024 - **Award amount:** $370,716 - **Award type:** 1 - **Project period:** 2024-02-10 → 2028-01-31 ## Primary source NIH RePORTER: https://reporter.nih.gov/project-details/10812815 ## Citation > US National Institutes of Health, RePORTER application 10812815, Developing statistical analysis and prediction tools for continuous glucose monitoring (CGM) use in hospitalized patients with diabetes (1R01DK136023-01A1). Retrieved via AI Analytics 2026-05-25 from https://api.ai-analytics.org/grant/nih/10812815. Licensed CC0. --- *[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*