Search for a command to run...
Aim. To develop a predictive model for the success of dual-task training in patients after on-pump coronary artery bypass grafting (CABG). Material and methods. The study included patients (n=51) who underwent dualtask training (postural balance and three cognitive tasks) after elective CABG. All patients underwent extensive neuropsychological testing at three time points: 1-2 days before CABG, 2-3 days after surgery, and before hospital discharge (11-12 days after CABG). At the last two points, postoperative cognitive dysfunction was assessed in patients according to the 20-20 criterion. To build a prognostic model determining the training success, the STATISTICA 10.0 software module "Data mining", "General Classification and Regression Trees", and the Standard C&R method were used. Results. We established that the highest probability of training success in the form of the absence of postoperative cognitive dysfunction on the 11-12th day after CABG were demonstrated by the following combinations of factors: on-pump time <86 min + temperature <35,45 о C + memory integrated index ≥0,439; onpump time <86 min + temperature <35,45 о C + memory integrated index ≥0,439; on-pump time ≥86 min + secondary education + neurodynamics integrated index <0,287 + temperature <35,45 о C; onpump time ≥86 min + secondary education + neurodynamics integrated index ≥0,287 + on-pump time <104; on-pump time ≥86 min + higher education + attention integrated index ≥0,09. The developed model demonstrated high predictive value, achieving 100% sensitivity and 95,83% specificity. Conclusion. The success factors of dual-task training were the initial level of cognitive status and education, as well as surgery-related factors (on-pump time and temperature). The study results offer the challenge for improving the medical decision support system and can be used to create personalized programs aimed at restoring the cognitive status of patients after cardiac surgery.
Published in: Russian Journal of Cardiology
Volume 31, Issue 2, pp. 6368-6368