Development and Validation of a Pattern-Recognition Engine for Visualization of Glycemic Patterns in Individuals Performing Low-Frequency SMBG

Abstract

Background and Aims: Innovative technologies supporting healthcare professionals (HCPs) and patients in the review of self-monitoring blood glucose (SMBG) data may aid clinical and educational approach to diabetes management. MyStar Connect® (MSC) is a diabetes data management software that helps HCPs review patient data and adjust diabetes therapy. Filtering mechanisms are needed to visualize data efficiently. We developed a pattern recognition engine (PRE) for visualization of glycemic patterns of SMBG data in MSC. Method: 3-month of SMBG data from 184 insulin- and non-insulin treated patients, at low testing frequency (0.5-2 tests per day), were independently evaluated by three expert assessors. Daily and weekly hypo- and hyperglycemic patterns were identified, and compared to those identified by the PRE. Discrepancies were reconciled through discussion between the assessors. The rate of agreement between assessors and PRE was evaluated. The analyses were conducted separately for hypoglycemic and hyperglycemic patterns. Results: Daily patterns: 56 hypoglycemic and 247 hyperglycemic patterns were identified by the PRE. No disagreement was noted in the assessment of hypoglycemic patterns between the PRE and the assessors. In 23 cases of hyperglycemic patterns, a disagreement was noted, due to the PRE missing a pattern considered as present by the assessors. After further clarification, 18 cases were reconciled. Therefore, the rate of agreement between the PRE and the assessors was 100% for hypoglycemic patterns and 98% for hyperglycemic patterns. Weekly patterns: The assessors recognized that the PRE provided information not easily inferable by simple review of SMBG data. PRE was therefore considered reliable for all the patterns identified. Conclusion: The implementation of PRE can represent a useful tool to guide HCPs’ treatment decisions and educate patients performing SMBG at a low frequency. Disclosure G. Vespasiani: Consultant; Self; METEDA Srl. Other Relationship; Spouse/Partner; METEDA Srl. A. Nicolucci: None. M. Saleh: Employee; Self; Sanofi. J. Sieber: Employee; Self; Sanofi. Stock/Shareholder; Self; Sanofi. G. Prosperini: None. C. Suraci: None. M. Galetta: None. G. Grassi: None.