Abstract Introduction The syndemic of diabesity is prevalent; and currently overconsumption is killing more people than starvation. Lifestyle change is a class I indication for the management of chronic diseases including cardiovascular disease and diabetes; but its implementation in primary care is fragmented. Purpose To determine if a digital therapeutic can positively affect type II diabetes and prediabetes in a primary care setting, in the NHS. Methods We used a large language model Artificial Intelligence in a closed evidenced-based format with cardiovascular and diabetes guidelines, to implement lifestyle change for patients in primary care, that melds seamlessly into the pathways of care of the primary care physician. Fig 1 The GDPR compliant digital therapeutic interacts with the primary care electronic health record to identify suitable patients for enrolment. Patients after informed consent on-board and are exposed to educational webinars at weeks 1, 3, 5, 9, 12, 26 and 52; via an application. Additional asynchronous advice from a lifestyle coach and nudge-psychology educational videos are continuously provided. We enrolled patients with non-insulin dependent type II diabetes and prediabetes and defined engaged patients as those who attended 4 or more webinars/consults after 28 days post-enrolment; versus those who did not (non-engaged). The effects on weight, BMI, HbA1c, total cholesterol and triglycerides were monitored after one year. Statistical analysis used student t-test of enrolled blood tests versus those at one year. Results Engagement was 84% for patients with type II diabetes and 75% for those with prediabetes. The effects of the digital therapeutic on engaged versus non-engaged patients for both type II diabetes and prediabetes is shown in Table 1. The average HbA1c for the engaged Type II Diabetes cohort (n=168) was 61.8 mmol/mol which declined through lifestyle to 51.7 mmol/mol (p0.03 x 10-16). The non-engaged cohort (n= 31) showed an increase in HbA1c from 60.2 to 64.8 mmol/mol. There was also statistically significant reductions in weight (p0.03 x 10-21), BMI (p0.02 x 10-9) and total cholesterol (P0.03) for the engaged versus the non-engaged cohort The average HbA1c for the engaged prediabetes cohort (n=86) was 43.5 and reduced to 41.5 mmol/mol (p0.05 x 10-5); by contrast the non-engaged prediabetes cohort (n=28) had a non-significant reduction in HbA1c of 0.7 mmol/mol. There was also a significant reduction in weight (P0.01 x 10-6), total cholesterol (p0.04) and Triglycerides (p0.003) in the engaged versus the non-engaged cohort. Conclusion A digital therapeutic delivered lifestyle change using artificial intelligence , has a clinically significant impact on metabolic parameters in patients with type II diabetes and prediabetes. This scalable and integrated model works in parallel with primary care practitioners and allied health professionals, conferring added benefits to patients.Figure 1
Colwell et al. (Thu,) studied this question.