Neural nets and prediction of the recovery rate from neuromuscular block.

摘要: Background and objective: The aim was to train artificial neural nets predict the recovery of a neuromuscular block during general anaesthesia. It assumed that initial/early data with simultaneously measured physical variables as inputs into well-trained back-propagation net would enable rough estimate remaining time. Methods: Spontaneous from (electrically evoked electromyographic train-of-four responses) were recorded following known affect block: multiple minimum alveolar concentration, end-tidal CO2 peripheral central temperature. Results: mean prediction errors, absolute root-mean-squared errors correlation coefficients all significantly better than those average-based predictions used in study. error - employing concentrations whole period (the time E2/E1 = 0.30 E4/E1 0.75; E1 first response train-of-four, E2 second etc.) smaller other nets, or same only initial (from 0.25). Conclusions: Neural could individual times method here, which supposed be more accurate guesses by any clinician. concentration monitored variable influenced rate, but it did not aid prediction.