Results Files
A description of the results files provided by the toolbox.
Overview
The toolbox provides a table of results for the performance of algorithms in CSV format. This page provides definitions for the results variables.
Variables
The results tables contain a header line (providing the results variables), and then a row providing the results for each algorithm. The variables reported are as follows:
| Variable | Definition |
|---|---|
| alg_no | An arbitrary number allocated to the algorithm |
| m_xa | The abbreviation of the filter-based technique used to extract a respiratory signal (if any). |
| m_xb | The abbreviation of the feature-based technique used to extract a respiratory signal (if any). |
| m_ef | The abbreviation of the frequency-domain technique used to estimate respiratory rate (if any). |
| m_et | The abbreviation of the time-domain technique used to estimate respiratory rate (if any). |
| m_fm | The abbreviation of the modulation-fusion technique used to estimate respiratory rate (if any). |
| m_ft | The abbreviation of the temporal-fusion technique used to estimate respiratory rate (if any). |
| alg_name | The algorithm name (using abbreviations) |
| signal | The input signal |
| two_sd | The precision of the algorithm (i.e. 2 times the standard deviation of the errors), accounting for repeated measures. |
| bias | The bias of the algorithm (i.e. the mean error), accounting for repeated measures. |
| cp1 | The proportion of RR estimates which have an error of <1bpm. |
| cp2 | The proportion of RR estimates which have an error of <2bpm. |
| icp5 | The proportion of RR estimates which have an error of >5bpm. |
| cp1_entire | The proportion of windows (which contain high quality input and reference respiratory signals, and a reference RR) for which the algorithm estimated RR and the error was <1bpm. |
| cp2_entire | The proportion of windows (which contain high quality input and reference respiratory signals, and a reference RR) for which the algorithm estimated RR and the error was <2bpm. |
| icp5_entire | The proportion of windows (which contain high quality input and reference respiratory signals, and a reference RR) for which the algorithm estimated RR and the error was >5bpm. |
| cost_func | TBC |
| tdi95 | The 95th percentile of absolute errors. |
| mape | The mean absolute percentage error |
| mae | The mean absolute error |
| sdae | The standard deviation of absolute errors |
| rmse | The root-mean-square error |
| prop_wins_hq_ref_and_input_signal_and_ref_rr | The proportion of windows (in the entire dataset) which had high quality input and reference respiratory signals, and a reference RR. |
| prop_wins_est | The proportion of windows (which contain high quality input and reference respiratory signals, and a reference RR) for which the algorithm provided an RR estimate and for which there was a reference RR available. |
| total_wins_inc_in_analysis | The total number of windows included in the analysis (i.e. which contain high quality input and reference respiratory signals, and estimated and reference RRs). |
| total_wins_in_dataset | The total number of windows in the dataset. |
| prop_wins_inc_in_analysis | The proportion of windows in the dataset which were included in the analysis. |
Notes:
- Unless otherwise stated, only the following windows are included in the analysis: windows with a high quality reference respiratory signal, a reference RR, a high quality input signal (i.e. ECG or PPG), and an estimated RR.