Dataset Viewer
Auto-converted to Parquet Duplicate
id
stringlengths
10
10
age
int64
18
89
train-0000
64
train-0001
70
train-0002
70
train-0003
39
train-0004
75
train-0005
82
train-0006
66
train-0007
31
train-0008
67
train-0009
72
train-0010
65
train-0011
40
train-0012
51
train-0013
70
train-0014
79
train-0015
71
train-0016
58
train-0017
32
train-0018
58
train-0019
28
train-0020
71
train-0021
65
train-0022
44
train-0023
59
train-0024
65
train-0025
50
train-0026
41
train-0027
44
train-0028
60
train-0029
55
train-0030
67
train-0031
51
train-0032
56
train-0033
65
train-0034
47
train-0035
60
train-0036
56
train-0037
88
train-0038
69
train-0039
84
train-0040
50
train-0041
26
train-0042
50
train-0043
60
train-0044
84
train-0045
52
train-0046
52
train-0047
78
train-0048
67
train-0049
64
train-0050
62
train-0051
72
train-0052
34
train-0053
86
train-0054
54
train-0055
79
train-0056
57
train-0057
55
train-0058
72
train-0059
72
train-0060
85
train-0061
61
train-0062
63
train-0063
42
train-0064
63
train-0065
47
train-0066
52
train-0067
68
train-0068
59
train-0069
76
train-0070
45
train-0071
57
train-0072
64
train-0073
58
train-0074
53
train-0075
42
train-0076
32
train-0077
62
train-0078
53
train-0079
59
train-0080
83
train-0081
50
train-0082
80
train-0083
69
train-0084
80
train-0085
46
train-0086
42
train-0087
51
train-0088
50
train-0089
65
train-0090
76
train-0091
78
train-0092
78
train-0093
63
train-0094
64
train-0095
55
train-0096
57
train-0097
72
train-0098
44
train-0099
48
End of preview. Expand in Data Studio

ecg-age

Twelve lead resting electrocardiograms with the age of the person they were recorded from, drawn from PTB-XL and repackaged one record per file.

Every record is ten seconds at 500 Hz. Each .npz holds exactly three arrays and nothing else:

  • signal, float32 of shape (12, 5000), the lead voltages in millivolts
  • fs, int64, the sampling rate, always 500
  • lead_names, the twelve leads in order: I, II, III, aVR, aVL, aVF, V1, V2, V3, V4, V5, V6

No other field of the source database travels with the signals. There is no sex, no diagnostic statement, no device, no recording date and no PTB-XL identifier in any file.

Layout

data/heldout/rec-0000.npz .. rec-2009.npz   2010 records, ages withheld
data/train/train-0000.npz .. train-5999.npz 6000 records
data/train/ages.csv                         id,age in whole years for data/train only

Ages are withheld for data/heldout. No age file is published for that folder, here or anywhere else. The 2010 held out records are drawn with a flat age distribution, 134 records in each five year band from 18 to 89, so the young are as numerous as the old. The 6000 training records keep the natural age distribution of the source database. One record per person throughout, and no person appears in both folders.

Source and licence

Derived from PTB-XL version 1.0.3, a clinical twelve lead ECG database of 21799 records from 18869 patients, released on PhysioNet under Creative Commons Attribution 4.0. This repackaging is released under the same licence. Records were taken from the 500 Hz records500 signals, one record per patient, restricted to ages 18 to 89.

Please cite all three of the following, as PTB-XL asks.

Wagner P, Strodthoff N, Bousseljot RD, Kreiseler D, Lunze FI, Samek W, Schaeffter T. PTB-XL, a large publicly available electrocardiography dataset. Scientific Data 7:154 (2020). https://doi.org/10.1038/s41597-020-0495-6

Goldberger AL, Amaral LAN, Glass L, Hausdorff JM, Ivanov PC, Mark RG, Mietus JE, Moody GB, Peng CK, Stanley HE. PhysioBank, PhysioToolkit, and PhysioNet: Components of a New Research Resource for Complex Physiologic Signals. Circulation 101(23):e215-e220 (2000).

Wagner P, Strodthoff N, Bousseljot RD, Samek W, Schaeffter T. PTB-XL, a large publicly available electrocardiography dataset (version 1.0.3). PhysioNet (2022). https://doi.org/10.13026/kfzx-aw45

Downloads last month
5