A new automated method of classifying raw glucose monitoring data (CGM) has been demonstrated to be feasible. It could help patients with diabetes manage their glucose levels, according to a recent study. Health Data Science.
Recent advancements in CGM devices enable diabetic patients to monitor glucose levels in real time and determine the impact of medication and lifestyle changes on their glucose levels. This is revolutionizing the way doctors, patients and other healthcare professionals can work together to improve patient health outcomes and personalize treatment. The glucose monitoring device readings can be easily interpreted by both the patient or their doctor. However, it can be slow, difficult, and inefficient. The data structure can also be problematic, such as irregular timing and complexity of readings. This makes it difficult to analyze cohort-level data.
By extracting key measurements, physicians can interpret data more quickly and accurately to improve their care. This can be achieved through more efficient and detailed analyses of successful strategies, and targeted interventions at patients most likely for such strategies to improve care, says Kyle Xin Quan Tan, a doctor at NOVI Health Singapore.
“To this end we developed a method of reducing continuous glucose measurements to an easier, distilled set of measurements that capture the most important aspects of patient records,” Sue-Anne Toh (author, doctor at NOVI Health, Singapore, and the National University of Singapore) says. “In a use-case, we demonstrated that there are four ‘glucotypes,’ groups of patients with different glucose dynamics throughout the day.
Many clinical implications have been linked to glucotypes, which are often linked to glycemic response patterns. A person may be assigned to one of these categories quickly, which could allow for personalized lifestyle and medical advice.
Alex R Cook, associate professor at the National University of Singapore, says that “the novel advance that this paper documents, is the means of collapsing CGM measurements into a smaller set of key measurements and further to simplify key measurements into small numbers of glucotypes.” Combining glycemic characteristics with an automated unsupervised classification algorithm could lead to systematic risk stratification, intervention, as well as diabetes management.
He also explained that this process can be used to facilitate other statistical analyses such as assessing if certain drugs or lifestyle changes have an accentuated effect on certain glucotypes.
Researchers are a mix of physicians and academics who frequently use CGM technology in their clinical practice. This allows research results to be more easily applied in practice, such as incorporating CGM technology into patient management tools.
The next steps are to continue the work on account for other variables like the timing of meals to improve the accuracy of eating and other modifications to reduce glycemic excursions. Future research will also focus on automation of personalized feedback triggered by glucose excursions. The team hopes to develop a model which can match glucotypes with the most effective medications and interventions. They can also be studied longitudinally to validate and develop glucotype-specific intervention.
Dialysis patients can rely on the accuracy of glucose monitors that are available off-the-shelf.
Yinan Mao et. al., Stratification for Patients with Diabetes Using Continuous Glucose Monitor Profiles and Machine Learning. Health Data Science (2022). DOI: 10.34133/2022/9892340
Health Data Science
Machine learning-based continuous glucose analysis is promising for personalized diabetes management (2022, Aug 15).
Retrieved 16 Aug 2022
This document is subject of copyright. Except for any fair dealings in private study or research, this document is not subject to copyright.
Part may not be reproduced without written permission. This content is only for informational purposes.
Source: medical xpress.