Cardiac risk assessment based on QTc speculation and trending from past references

A method to construct a predictive time series index based on QTc-intervals is proposed in this paper. Monitored electrocardiography (ECG) data is converted into a root mean square of successive difference (RMSSD) trend-line by first finding the QT intervals [1] and then using [2]. The trend-line is then used as a priori in extrapolating the predictive trend. The next unknown RMSSD is extrapolated using a modified implementation of the Pearson´s Correlation. The last (m-1)^th amount of RMSSD in the priori is the 1^st of the 2 time-series being compared in Pearson´s Correlation; the second time-series is a running window of m^th number of RMSSD from the 1^st RMSSD to the (n-m-1)^th RMSSD. The extrapolated RMSSD is substituted into the priori to compute the next RMSSD. The procedures are reapplied recursively until the required time series has been found. Using our method, the average Mahalanobis distance for all the records in the MIT-Arrhythmia Database is 0.1899.