Comparability of Heart Rate Turbulence Methodology: 15 Intervals Suffice to Calculate Turbulence Slope – A Methodological Analysis Using PhysioNet Data of 1074 Patients

Figures

• 

  Figure 1. Lorenz plot patterns and their corresponding smoothed histograms of SD1. The plots show 30-min chunks of different measurements that were used in the analysis as an example of different Lorenz plot patterns. The Poincaré plots in the top show a (A) comet, (B) side lobe pattern, (C) fan pattern, and (D) island pattern. The corresponding plots below show the histograms after smoothing. The lobe, fan, and island patterns have an SD2/SD1 ratio of less than 1.5 and would, therefore, be excluded. The histogram of the island patterns includes more than one local maximum, which is another criterion for exclusion. Data: (A) nsr008 from NSR2DB, (B) nsr018 from NSR2DB, (C) 01 from LTAFDB, and (D) f018 from CRISDB.

• 

  Figure 2. Standard VPCS (stVPCS) files with distinct heart rate turbulence (HRT). Exemplary tachograms of two files used to calculate the stVPCS that show distinct HRT. Both files are in class HRTA. The upper row shows a zoomed in tachogram, the row below the respective tachogram zoomed out. Data: (A) e145a from CRISDB and (B) nsr010 from NSR2DB.

• 

  Figure 3. stVPCS files without distinct HRT. Exemplary tachograms of two files used to calculate the stVPCS that do not show distinct HRT. Both files are in class HRTB. The upper row shows a zoomed in tachogram, the row below the respective tachogram zoomed out. Data: (A) s20491 from LTSTDB and (B) chf202 from CHF2DB.

• 

  Figure 4. The stVPCS calculated from 47 files that matched all filter criteria and had the best HRV parameters.

• 

  Figure 5. Averaged HRT parameter values of all files grouped by the respective turbulence timing (TT) value. The color of the boxes and the gray background graph illustrate the number of files in the different groups. The common threshold of TO (0%) (A) and TS (2.5 ms/RR) (B) is represented as a blue line. The common cut-off of TS is used for nTS, because no threshold for nTS has been established yet (C). The median of the groups with low TT (TT ≤ 10 for TO and TT ≤ 6 for TS) lie on the low-risk side of the threshold. For nTS, only the medians of the groups with a TT of 2–4 lie above the threshold.

• 

  Figure 6. Dynamic time warping (DTW) analysis of exemplary postRRs grouped by their respective TT. The left side (A) shows the analysis of the stVPCS with the full postRRs sequences, the right side (B) with the cut sequences. From up to down the plots show the comparisons for TT 1 (1), 3 (2), and 7 (3). The full averaged sequence of TT 1 (A.1) lacks the initial bend and shows an immediate IL incline. The sequence of TT = 3 fits the stVPCS the best, both in the full (A.2) and in the cut version (B.2).

• 

  Figure 7. Distances calculated by DTW. The differences between stVPCS and the averaged VPCSs of all files grouped by their respective TT are shown in dark blue (full sequences) and light blue (cut sequences). The distances of the cut sequences are lower than the distances of the full sequences. In both TT 3 has the lowest distance to the stVPCS. The gray background graph and the corresponding axis on the right illustrate the number of files in the different groups.

• 

  Figure 8. Linear regression analysis of TT and its respective TTSD of all files included in the analysis.