Predictors of Type 2 Diabetes (T2D) Remission using Digital Twin Technology Based on Artificial Intelligence (AI) and Internet of Things (IoT) Technologies

To identify pre-intervention anthropometric and biochemical parameters that predict diabetes remission after applying AI-based IoT technology and post-intervention parameters that are improved in patients achieving remission.

Methods:

Twin Precision Treatment (TPT) is a mobile application and intervention based on the patented Whole-Body Digital Twin (WBDT) enabled precision treatment. An independent ethics committee approved this RCT. 167 patients with T2D completed 180 days of TPT. Remission was defined as an A1C level < 6.5% for ≥180 days without diabetes medication use for at least 90 days. 141 of the 167 patients (84.4%) had remission of diabetes using this definition (group R).

Results:

At baseline, patients in R group, compared with patients in the Non-remission Group (group NR) had shorter duration of diabetes (yrs) (R: 3.3±2.7, 95% CI 2.9 to 3.7) vs. (NR: 4.7±2.4, 95% CI 2.9 to 3.7), p=0.015; lower pre-TPT A1C (R: 8.8±1.8, 95% CI 8.5 to 9.1) vs. (NR: 9.7±1.8, 95% CI 9 to 10.5), p=0.021; lower estimated A1C (eA1C) (R: 8±2.2, 95% CI 7.7 to 8.4) vs. (NR: 9.5±2.5, 95% CI 8.5 to 10.5), p=0.002; lower glucose management indicator (GMI) (R: 7.5±1.9, 95% CI 7.2 to 7.8) vs. (NR: 8.6±1.7, 95% CI 7.9 to 9.3), p=0.006; lower Time Above Range level 2 (TAR2; target >250 mg/dl < 5% time) (R: 17.4±23.8, 95% CI 13.5 to 21.4) vs. (NR: 33.6±31.5, 95% CI 20.9 to 46.3), p=0.003; higher Time In Range (TIR; target 70–180 mg/dl >70% time) (R: 52.5±32, 95% CI 47.2 to 57.9) vs. (NR: 36.6±34.8, 95% CI 22.5 to 50.6), p=0.023; and lower fasting plasma glucose (R: 168.8±54, 95% CI 157.8 to 175.8) vs. (NR: 197.8±82.8, 95% CI 164.4 to 231.3), p=0.015. At 180 days, the difference in the change, in R was better than NR for Time Below Range1 - TBR1 ( < 70–54 mg/dl) < 4% (12.7±16.9, 95% CI 9.9 to 15.5) Vs (5.7±6.6, 95% CI 3.2 to 8.3), p=0.04; HOMA2IR (%) (1±0.43, 95% CI 0.9 to 1.1) Vs (1.22±0.47, 95% CI 1 to 1.4), p=0.019; HOMA2B (%) (92.5±32.3, 95% CI 87.1 to 97.9) Vs (72.8±23.8, 95% CI 63.2 to 82.4), p=0.004; WBC (thousand/mm3) (6453.9±1486.4, 95% CI 6206.4 to 6701.4) Vs (7410.4±1480.1, 95% CI 6812.6 to 8008.2), p=0.003. The baseline and change in body weight, LDL-C, HDL-C, TGs, sdLDL were comparable. The final mean A1C in NR was 6.3% Vs 5.5% in R. The mean baseline anti-diabetic drug count was lower in R (1.73) Vs NR (2.23).

Discussion/Conclusion:

A shorter duration of diabetes, lower A1C, eA1C, GMI, and TAR2 at baseline are reliable predictors for diabetes remission with the TPT intervention. The TBR1, HOMA2B, and HOMA2IR by 180 days improved in patients who achieve remission. TPT is a promising lifestyle intervention for diabetes remission.