**6. AI in diabetes and Glycemic control**

Artificial intelligence is a fast-growing field with its applications for persons living with many chronic diseases such as diabetes. There has been global concern about the ever-increasing incidence rate of diabetes with one in two persons undiagnosed

*Unlocking the Potential of Artificial Intelligence (AI) for Healthcare DOI: http://dx.doi.org/10.5772/intechopen.111489*

and untreated [53]. The total number of people living with diabetes is likely to rise to 643 million by 2030 and 783 million by 2045 [53]. A recent study of 300,000 patients with type 2 diabetes on medical therapy found that after 3 months, 31% of patients had discontinued their diabetes medications that number was widened to 44% by 6 months, and to 58% by 1 year [54]. Besides, about 75% of diabetic adults live in lowand middle-income countries with only 5% of those receiving thorough treatment according to guidelines [55]. The best care for diabetes is mostly hindered by lack of real-time crucial health information required to make necessary choices with diabetes control and therapy.

Today, advances in AI have introduced a shift in diabetes care from conventional management approaches to targeted data-driven precision care. There is a spectrum of interventions spread across different care processes in diabetes. AI is not only being applied to predict diabetes risk utilizing genetic data and to diagnose diabetes *via* electronic health record data in clinical decision support but it is also transforming diabetic care and predicting the potential sequelae of diabetes such as nephropathy and retinopathy. Such solutions have enhanced the workflow of both medical staff and patients.

### **6.1 How is AI utilized in diabetic care?**

To help fight diabetes disease and improve its management, AI can play a vital role in diabetic care at many different levels discussed below that can benefit both providers and patients in a team-oriented approach.

#### *6.1.1 Diabetes prediction*

AI can help diagnose diabetes noninvasively and proactively by identifying a subset of populations with the highest risks at a pre-illness stage. Though diabetes prediction models have been generated by conventional statistics, machine learning (ML) can maximize the predictive performance of conventional models to the next level [56]. Algorithms built by ML can do risk stratification by analyzing genomics, lifestyles, mental and physical health, and social media activity. Earlier detection and intervention for at-risk individuals could decrease the incidence of diabetes, and the financial costs associated with uncontrolled diabetes.

#### *6.1.2 Lifestyle guidance for diabetes patients*

Monitoring glucose levels in real-time is being done using wearable devices and continuous glucose monitoring systems of patient symptoms and biomarkers. Continuous glucose monitors (CGM), which are now frequently used by diabetics, acquire a large amount of data that has previously been underutilized. The amount of glucose in the fluid inside the body is measured by CGM. In certain circumstances, the sensor is glued to the back of the arm or is implanted under the skin of the belly rapidly and painlessly. The information is then sent to a wireless-pager-like monitor through a transmitter on the sensor [57].

CGM sensors can be divided into two main categories: Professional CGM sensors and real-time CGM sensors (rtCGM) [58]. Professional CGM sensors are prescribed by healthcare professionals usually for limited periods of time, they record glucose concentration data in blinded modalities (i.e., the patient cannot visualize the data in real-time), and they allow the healthcare professional to retrospectively review the

patient's glycemic trends and make therapy adjustments. Conversely, with real-time CGM sensors (rtCGM) the recorded data are accessible in real-time to the patient, who can use these data for improved decision-making in the daily management of Type 1 diabetes. AI can enable patients to decide what to eat or drink and what level of physical exercise is suitable.
