Remote monitoring using an AI-powered voice robotic assistant for control of metabolic syndrome parameters
M A Ajadi, S R Chumalaeva, V Y U Zektser, M F Petrukchnova, Y U N Belenkov, N V Khabarova, M V KozhevnikovaAbstract
Background
Metabolic syndrome (MetS) affects approximately 25% of the global adult population and substantially elevates cardiovascular morbidity and mortality. Despite guideline-directed therapies, sustained metabolic control and treatment adherence remains difficult to achieve. Advances in digital health demonstrate that remote monitoring (RM) can support self-management in chronic cardiometabolic conditions. However, the impact of AI-powered voice assistants on comprehensive MetS parameters in routine clinical care requires rigorous evaluation.
Purpose
To assess the effect of AI-voice robotic assistant RM program on anthropometric measures, laboratory parameters, quality of life (QOL), and treatment adherence in patients with MetS compared to standard care alone.
Methods
This comparative study included 121 patients with MetS (66 women, 55 men): 71 in the monitoring group (MG) (standard care plus daily monitoring via AI-powered voice-assistant questionnaires) and 50 in the control group (CG) (standard care only). After 1 month follow-up, anthropometric and laboratory parameters were assessed; QOL was evaluated using SF-36, and treatment adherence scale was graded as low, moderate, or high. Statistical analysis was performed in StatTech v.4.12.2, with p<0.05 considered significant.
Results
Baseline characteristics were comparable between groups. After one month, the MG demonstrated significant reductions in BMI (33.15 [32.23;34.08] to 31.90 [31.00;32.80] kg/m²), fasting glucose (8.09 [7.48;8.71] to 6.37 [6.11;6.64] mmol/L), total cholesterol (5.71 [5.31;6.11] to 5.21 [4.85;5.58] mmol/L), LDL-C (3.83 [3.74;3.92] to 3.22 [3.14;3.30] mmol/L), and triglycerides (2.04 [1.72;2.36] to 1.49 [1.29;1.69] mmol/L) (all p<0.001). HDL-C remained unchanged. Uric acid decreased in MG; women (362.9 [342.1;383.7] to 332.5 [313.2;351.8] μmol/L) and men (452.8 [421.3;484.3] to 418.6 [389.4;447.8] μmol/L) (p<0.001). ALT, AST, GGT all improved (p<0.001). No significant changes occurred in the CG. QOL scores increased in both groups (MG: 71.94 [70.29;73.60] to 78.48 [77.11;79.85]; CG: 71.10 [69.88;72.32] to 72.38 [71.08;73.68]; both p<0.001). Baseline adherence were similar (MG: 54.9% low, 45.1% moderate, 0% high; CG: 54.0% low, 42.0% moderate, 4.0% high). Post-intervention, low adherence was eliminated in the MG - 80.3[69.1–88.8]% achieving high adherence (p<0.001), the CG showed minimal change (56.0% - low, 6.0 [1.3–16.5]% - high adherence p=0.016).
Conclusion
This study demonstrates that integrating an AI-powered voice robotic assistant into RM program produces clinical improvements in metabolic parameters, QOL, and treatment adherence in patients with MetS. The achievement of high adherence in 80% of MG participants addresses a critical therapeutic gap in cardiovascular prevention. These findings suggest that scalable, low-burden voice-assisted digital interventions can effectively translate guidelines into sustained behavioral change.