Brugada Syndrome Diagnosis: Three Approaches to Combining Diagnostic Markers

Predictive Models for Mutation Carriers in Brugada Syndrome Screening

“…As mentioned before, the diagnosis of BS is definitive only when the Brugada pattern is identified in an ECG, but this entails the problem of the constant fluctuations of the Brugada pattern [10], [19]. Earlier studies [11], [18], [20] have opened up new options in the diagnosis of BS, identifying five ECG markers with good potential to distinguish mutation from non-mutation carriers [18], and exploring ways of linearly combining these markers in order to enhance the ability of each one to discriminate between the two groups [11]. Exploring the logistic regression models of Henriques et al [11] further, in this study we assess the predictive ability of those models and also we apply shrinkage techniques to improve calibration and future predictions.…”

“…Earlier studies [11], [18], [20] have opened up new options in the diagnosis of BS, identifying five ECG markers with good potential to distinguish mutation from non-mutation carriers [18], and exploring ways of linearly combining these markers in order to enhance the ability of each one to discriminate between the two groups [11]. Exploring the logistic regression models of Henriques et al [11] further, in this study we assess the predictive ability of those models and also we apply shrinkage techniques to improve calibration and future predictions. First, the stability of coefficient estimates will be evaluated through 95% bootstrap percentile confidence intervals, which will be presented in Section 3.…”

“…In Henriques et al [11], several methodologies were applied to combine the five ECG markers presented above, in order to obtain a prediction model for mutation carriers. Three combination alternatives were highlighted: combination of PR and QRSf, combination of these two and dQT, and combination of the five markers.…”

“…In earlier studies, [11], [18], [20], five ECG markers were identified to have good ability to discriminate between carriers and non-carriers of the genetic mutation. Henriques et al [11] explored different ways of combining these five markers in order to enhance the ability of each one to discriminate between the two groups. The studied models exhibited excellent discriminative ability, opening the possibilities for diagnostic tools in Brugada Syndrome.…”

“…Yet, these models have the weaknesses mentioned above: they are based on a small sample data set and some of the five markers are relatively high correlated. In this study, we use bootstrapping to obtain estimates of prediction accuracy for the logistic regression models studied in [11]. Furthermore, we use shrinkage methods to improve calibration and future predictive accuracy of the models.…”