Evaluation of the algorithm of Anticlot Assistant: an anticoagulant management system based on mobile health...

The purpose is to evaluate the algorithm of Anticlot Assistant, a novel anticoagulant management system based on mobile health technology which was developed to facilitate patient self-management. The eligible patients managed warfarin therapy with usual care, following the prescriptions of the doctors. The actual prescriptions of doctors and the virtual recommendations by Anticlot Assistant were compared and analyzed. There were no significant differences between the next test dates recommended by Anticlot Assistant and those prescribed by doctors. The mean warfarin dosage prescribed by doctors was lower than that recommended by Anticlot Assistant (2.74 ± 1.17 vs. 2.79 ± 1.21 mg, 95% confidence interval for the difference: -0.01--0.09, P = 0.019, n = 139), resulting in the international normalized ratio a high time below the therapeutic range (TTR) (29.9 ± 17.9%), and a low time above TTR [0.0% (0.0-18.7%)]. A mixed linear model revealed that 'the variations of the dosages prescribed by doctors from those recommended by Anticlot Assistant' were positively correlated with 'variations of next international normalized ratios from TTR' after controlling for other factors (estimate of the effect = 0.231, 95% confidence interval: 0.034-0.428, P = 0.022). Anticlot Assistant can mimic the doctors' prescriptions for the next test date and the warfarin dosages recommended by Anticlot Assistant might be more reasonable than those prescribed by doctors, which indicated that the algorithm was reliable and it was possible for the patients to manage warfarin therapy themselves with the aid of Anticlot Assistant.