Background: Management of the timing and frequency of meal announcements remains a key determinant of postprandial glycemic control in hybrid automated insulin delivery (AID) systems. Tools that translate routine device data into clinically interpretable insights on meal management behavior for individual patients and broader cohorts remain limited. These tools can potentially provide clinicians and users objective data to formulate personalized meal-related recommendations to safely improve outcomes. Methods: We developed a framework to quantify meal management behavior from routine AID data. The framework combines a meal detection model and a behavioral classification module that categorizes events as on-time, late, or missed based on temporal relationship between detected meals and carbohydrate entries. The meal detection model was trained on data from adults with type 1 diabetes using MiniMed 780G system (NCT04479826) and evaluated in a validation cohort of 157 individuals with type 1 diabetes using advanced hybrid closed-loop (AHCL) (NCT03959423) system. Results: Meal events detection accuracy reached 96.9% in the validation cohort, with precision ≥89.6%, recall ≥93.9%, and area under the curve = 0.996. Application of the tool to the validation cohort showed that a higher proportion of on-time announcements correlated with higher time in range (TIR) (ρ = 0.375, q < 0.001) and lower hyperglycemia. Each 10% increase in on-time announcements corresponded to a 2.66% increase in TIR. Increasing delay of late announcements was associated with worse postprandial glycemic outcomes. Conclusions: This framework enables insights into meal announcement patterns that may support patient counseling, clinical decision support, and large-scale analyses in AHCL systems. Clinical Trial Registration Number: NCT04479826, NCT03959423.
BACKGROUND Diabetes-related foot ulcers (DFUs) that fail to heal within the expected trajectory and become chronic are associated with substantial morbidity. The current clinical standard for identifying chronic DFUs relies on a <50% reduction in wound area over 4 weeks. There is a critical need for tools that can predict wound-healing outcomes at first clinical presentation to enable timely, individualised wound management. We have developed and validated TherMUL, an AI device that analyses the texture of thermal images to predict DFU healing. METHODS Adults with neuropathic DFUs were recruited from 2 specialist foot clinics in Melbourne, Australia. TherMUL, a software-based device, was used to take thermal images during the initial visit. Texture features were extracted using grey-level co-occurrence matrix (GLCM) analysis. Wound-healing status at 12 weeks served as the ground-truth label for model training. A support vector machine (SVM) classifier was trained using the images from 44 healed and 24 unhealed DFUs. A separate unseen dataset was used for a preliminary feasibility assessment. RESULTS The classifier achieved a mean area under the curve (AUC) of 0.635, with a sensitivity of 75.8% for predicting unhealed DFUs, and a specificity of 53% for identifying healed DFUs. In the preliminary evaluation, 5 of 7 predictions correctly matched the 12-week podiatry outcomes. CONCLUSION This study demonstrated early feasibility of using TherMUL to assist clinicians in predicting 12-week DFU healing outcomes at first clinical presentation, with moderate discriminative performance. This non-contact, point-of-care approach has the potential to support earlier identification of DFUs unlikely to heal and enable more timely, targeted wound management strategies.
Background: Diabetic ketoacidosis (DKA) management by variable-rate insulin infusion is increasing with the widespread adoption of electronic glucose management systems (eGMS). Herein, we evaluate the implementation of an updated process for DKA management incorporating the eGMS algorithm. Methods: An eGMS was implemented within the Cedars-Sinai Health System at our tertiary academic center (AC) and community hospital (CH) in January 2024. We compared outcomes after eGMS implementation (post-GMS) to the previous standard of care (pre-GMS). Adult patients admitted to the intensive care unit with a primary diagnosis of DKA and treated with an insulin infusion between January 2023 and December 2024 were included. We separately evaluated outcomes of hypoglycemia (blood glucose, BG <70 mg/dL) and severe hypoglycemia (BG <40 mg/dL) and treatment efficacy. Results: In total, 324 patients were included in the pre-GMS and 211 patients in the post-GMS cohorts. There was a decrease in the prevalence of hypoglycemia (pre-GMS: 11.0%, post-GMS: 4.3%, P = .005) and severe hypoglycemia (pre-GMS: 1.90%, post-GMS: 0.00%, P = .047). At AC, the hospital length of stay (LOS) decreased from 8.6 to 6.0 days ( P = .012). At CH, insulin drip duration increased significantly (pre-GMS: 26.4 h, post-GMS: 36.7 h, P = .004), but there was no impact on LOS. Conclusion: Updating the DKA process and implementing an eGMS improved hypoglycemia rates throughout the institution. Hospital LOS was reduced at the AC, reflecting standardization and streamlining of the DKA processes.
