Among the most frequent questions asked by visitors to PhysioNet are requests for data with specific combinations of characteristics. For example, “Which records include three or more ECG signals and a respiration signal, are at least two hours long, and are from male patients between the ages of 60 and 70?” This and many similar questions are readily answered using the PhysioBank Index. (See how to do this below.)
The PhysioBank Index
All records in PhysioBank that can be viewed by the PhysioBank ATM are indexed in the PhysioBank Index, a 76M text file (last updated Wednesday, 8 July 2015).
You can search the PhysioBank Index using a variety of methods:
- PhysioBank Record Search
This is the most powerful method, and the easiest (use it from within your web browser). Click the link above to begin, or try the hands-on tutorial introduction below.
- The pbsearch software package
This package includes sources for the PhysioBank Record Search (the web client pbsearch, and the server pbsqsd) as well as a command-line client (pbsqc), a stand-alone search engine (pbsqs), and a set of plugins. The software runs on all popular platforms; it can work with a local copy of the Index, or it can read the Index directly from the PhysioNet web server or a mirror via HTTP. This package may be useful if you wish to customize or automate your searches; familiarize yourself with the web-based PhysioBank Record Search first.
- Standard command-line tools such as grep, cut, and uniq
You can also search the Index using Unix (POSIX) utilities for manipulating text files, as described below. These are standard components of GNU/Linux, Mac OS X, and all other Unix and Unix-like platforms; Windows users can get them by installing Cygwin.
If you wish to index your own PhysioBank-compatible records, the portable sources for the PhysioBank Index generator are available in the pbindex software package.
Downloading the PhysioBank Index
You do not need to download the Index in order to use the recommended PhysioBank Record Search. If you wish to another method for searching the Index, you may need to do so.
The Index can be downloaded using any of these commands:
wfdbcat physiobank-index >temp; mv temp physiobank-index
curl http://physionet.org/physiobank/database/physiobank-index >physiobank-index
rsync -avz physionet.org::physiobank-core/database/physiobank-index .
Don't attempt to load the Index directly in your web browser, since its large size will cause problems for most browsers! (wfdbcat is part of the WFDB Software Package, curl is available here, and rsync is here. All three are open-source and run on all popular platforms.)
Contents and format of the Index
Each line of the PhysioBank Index describes one signal, annotation file, or other feature of a single PhysioBank record; there are about 860,000 lines in the Index. All lines pertaining to any given record are consecutive, and the records appear in dictionary order. Here is a sample from the Index:
edb/e0103 Info1 Mixed angina edb/e0103 Info2 1-vessel disease (RCA) edb/e0103 Meds1 nitrates, diltiazem edb/e0103 Info3 Recorder type: ICR 7200 edb/e0103 AgeSex 62 M edb/e0103 ECG1 V4 250 200 adu/mV 7200 edb/e0103 ECG2 MLIII 250 200 adu/mV 7200 edb/e0103 AnnR1 atr 250 7336 7200 0-7200 edb/e0103 AnnR1 atr/(N 250 9 5243 edb/e0103 AnnR1 atr/N 250 7212 7199 0-7200 edb/e0103 AnnR1 atr/s 250 15 6098 859-6956 edb/e0103 AnnR1 atr/(VT 250 2 6 edb/e0103 AnnR1 atr/V 250 82 5069 1892-6961 edb/e0103 AnnR1 atr/(B 250 6 58 edb/e0103 AnnR1 atr/F 250 2 574 2132-2706 edb/e0103 AnnR1 atr/~ 250 8 3613 2682-6295
Each entry (line) of the Index contains up to seven tab-separated fields (columns) that describe a signal, annotation set, or feature associated with the record. For entries describing features (such as the first five lines in the example above), these columns are (from left to right):
- Record name
In each entry in the example above, the record name is edb/e0103, i.e., record e0103 of the collection named edb. (The file DBS contains short descriptions of each collection; edb is the European ST-T Database.)
- Class (feature type)
- AgeSex: The next two columns contain the subject's age and gender. Age of subjects over 89 is PHI, and is stated as 90; unknown age is stated as -1. Ages of infants less than 1 year old may be shown as 0, or as a decimal fraction of a year (e.g., 0.3). Gender is stated as M or F, or ? if unknown. If neither age nor gender is known, this entry is omitted from the Index.
- Diagn: The next column contains one or more free-text diagnoses. This entry appears only if diagnoses have been tagged in the .hea file for the record; in many cases, as in the example above, diagnoses appear as Info since they have not been tagged.
- Infon: Untagged information collected from the .hea file for the record.
- Medsn: One or more medications. As for Diag, this entry appears only if medications have been tagged in the .hea file; otherwise, they appear as Info.
A sequence number n is affixed to each entry for any given class if more than one such entry is possible in a single record (e.g., Info1, Info2, etc.); this is done even if only a single entry is actually present. The sequence numbers are consecutive but not necessarily contiguous, as the example above illustrates. Note that entries describing features are present only for a relatively small number of records.
- Feature value(s)
A string providing the value for the feature named in the previous column. Feature values may include spaces but not tabs. (An AgeSex entry has two feature value fields, which are separated by a tab.)
- Record name
The category of data: either
- a category of signals (defined in sigclasses)
- a category of annotations (either AnnM for machine-derived annotations, or AnnR for reference annotations)
A sequence number n is affixed to each entry for any given class if more than one entry is possible in a single record (e.g., ECG1, ECG2, etc.); this is done even if only a single entry is actually present.
- Signal or annotator name
The index contains an entry for each signal and for each annotation file. In addition, each type of annotation present in a given annotation file is counted separately and listed in its own annotation subset entry. The last eight lines in the example above are annotation subset entries, summarizing the (N, N, s, etc. annotations present in within the atr annotation file. See PhysioBank Annotations for a glossary of commonly used annotations.
- Sampling frequency (Hz)
The number of samples per second. Annotations contain timestamps with a time resolution equal to the reciprocal of the sampling frequency.
- Gain (of signals) or number (of annotations)
Gain is expressed as adu per physical unit. An adu is one analog-to-digital converter unit (the quantization step, which is the smallest measurable difference between samples). An amplitude resolution of 200 adu/mV, as for the ECG signals in the example above, means that two unscaled samples that differ by 20 units represent a potential difference of 0.1 millivolt.
Length of the signal or annotation set, in seconds.
- Time intervals
Interval(s) during which samples or annotations have been recorded if present. The times shown for the beginning and end of each interval are the elapsed times in seconds from the beginning of the record.
- The necessary shell (command-line interpreter, such as bash, csh, ksh, or sh) and command-line tools (grep, cut, sort, etc.) are standard components of GNU/Linux, Mac OS X, and all other Unix and Unix-like platforms; Windows users can get them by installing Cygwin.
- The PhysioNet FAQ includes basic information about standard input and output, I/O redirection, and pipes, powerful and easily-understood concepts that are useful whenever working on the command line.
- On-line tutorials, such as Working with Data or Command Line Essentials: Text and Pipeline, provide more examples of the use of the tools shown here.
- After nearly 30 years, Kernighan and Pike's The Unix Programming Environment remains the best introduction to this approach of tackling problems using tools that each do one job well, and work well together. Used copies are far less expensive than new.
- Records in the QT Database and many of those in the T-Wave Alternans Challenge Database are excerpts of longer records from other PhysioBank data collections.
- Records in the Sudden Cardiac Death Holter Database are complete long-term recordings that were previously excerpted in the MIT-BIH Malignant Ventricular Arrhythmia Database. Similarly, the Long-Term AF Database consists of complete recordings of which brief excerpts appeared in the AF Termination Challenge Database, and one of the 90 records of the Long-Term ST Database is a complete recording that was previously excerpted in the European ST-T Database.
- Several important subsets of the MIMIC II Waveform Database are also indexed as independent records, so that records belonging to them appear more than once in the PhysioBank Index. These include the MIMIC II Waveform Database version 2, containing older versions of about 25% of the current (version 3) MIMIC II Waveform Database; the MIMIC II Waveform Database Matched Subset, containing records that have been matched and time-aligned with those in the MIMIC II Clinical Database; and excerpts of MIMIC II Waveform Database records used in the 2009 and 2010 PhysioNet/CinC Challenges. In addition, records containing estimates of heart rate, blood pressure, and signal quality are associated with about half of the version 2 waveform records.
In entries describing signal and annotation sets, the columns are (from left to right):
Some records have both annotated and unannotated segments; in these cases, the length of the annotation set is shorter than the length of the signals. In annotation subset entries, the duration and time interval reflect the times of the first and last annotations of the associated type, and are generally shorter than the entire annotation set.
In most cases, signals are present throughout, and the last column is omitted in entries that describe such signals. The MIMIC II Waveform Database is an exception to this rule, since many of its signals have been recorded in only a subset of segments; in these cases, the lengths of the signals is less than that of the entire record, and there may be more than one time interval shown.
As for the signal and annotation set lines, the first two columns are the record name and class (data type). The first four feature lines shown above illustrate diagnoses, medications, and two lines of free-text information; the data appear in the third column. The final feature line contains the age (in years) in the third column, and the sex (M, F, or ? in the fourth column). If the subject’s age is over 89, it is shown as 90 (since ages over 89 are protected health information); if the age was not recorded, it is shown as –1.
Using PhysioBank Record Search to search the PhysioBank Index
In this section, we'll answer the question at the top of this page using PhysioBank Record Search. Please follow along in your web browser. Note that your results may vary from those shown below if additional records have been added and indexed since this tutorial was written.
PhysioBank Record Search is controlled from your browser. You can open it from the PhysioNet menu button at the top left corner of most pages on PhysioNet (choose PhysioBank → PhysioBank Search). To follow this exercise, click here to open it in another browser tab or window. You should then be able to go back and forth between this page and the PhysioBank Record Search page.
The upper section of the page, below the PhysioBank Record Search heading, contains a form for composing simple queries, searches that are defined by a Subject, a Relationship, and a Value. Once you have performed any simple queries, the Results section opens beneath the upper section; it contains the results of your queries and additional controls for combining and manipulating them. Near the bottom of the page, on-line help for PhysioBank Record Search appears below the heading How to search for records in PhysioBank. Read the on-line help to familiarize yourself with the controls.
If you have used PhysioBank Record Search within the past week or so on this computer, you will see two buttons in the upper section of the page, labeled “Restore previous session” and “Discard previous results”. Unless you wish to keep the results of your earlier searches, discard them before continuing.
To answer the example question (”Which records include three or more ECG signals and a respiration signal, are at least two hours long, and are from male patients between the ages of 60 and 70?”), our strategy will be to decompose it into simple queries, collect the results of those queries, and combine them. At each step, we'll be able to see how many PhysioBank records fit the simple queries.
Let’s begin. The first simple query will find all records that contain three or more ECG signals. From the Subject menu, select '(#) ECG'. The notation '(#)' indicates that we can specify a minimum number of signals of this type (ECG) in the Name/# box below the Subject menu; type 3 into that box. From the Relationship menu, choose '?' (”defined”), since for this simple query, we only want to know if the 3 ECGs exist. It is unnecessary to enter anything in the Value box (if you do, it will be ignored since the Relationship is '?'). Once the query has been set up, click on Get List.
Immediately below the Results heading, you should now see a line that looks like this:
☐ A  ECG-3 ?
“A” is the tag for the list of results of this query; the bracketed number (which, as noted above, may vary) is the number of PhysioBank records that match the criterion that you have just defined, and the criterion itself appears as a link (”ECG-3 ?”). Click the link (on the search page; the facsimile shown above is not a working link) to view the results if you wish (a list of 25372 record names, probably beginning with challenge/2009/test-set-a/101a/101a).
For the second simple query, let's find all records that include a respiration signal. We might construct a query such as 'Resp ?', but since we also want records that are at least two hours long, we can include that constraint as an element of this query. Select '(#) Respiration' as the Subject, '>=' as the Relationship, and type '2:0:0' (i.e., two hours) as the Value. Don’t forget to erase the '3' that may be left in the Name/# box from the previous query; it’s unnecessary (but harmless) to type a '1' in its place.
After clicking Get List, a second checkbox and result appears. New results appear above old ones, so you should now see this:
☐ B  Resp >= 2:0:0
☐ A  ECG-3 ?
There are many records that satisfy this criterion, too.
If you forgot to erase the '3' before clicking 'Get list', list B will be much shorter, since it will contain only records that include 3 or more respiration signals. Such records do exist (for example, some have thoracic and abdominal impedance plethysmograms and a simultaneously recorded nasal thermistor signal), but they are relatively uncommon.
Next, let’s find records from male subjects ('sex = M'). By now, you may be becoming familiar with the steps of creating a simple query: select the subject, then the relationship, then type a value, and click Get List.
☐ C  sex = M
☐ B  Resp >= 2:0:0
☐ A  ECG-3 ?
You may be surprised that list C has only 9662 records from male subjects, given that list B has about 4 times as many records. For many records, however, information about the subject’s gender (and, as we shall see, age) is not available.
To restrict the search to a range of ages (60-70), we'll use two simple queries ('age >= 60' and 'age <= 70'). When you have run these, your results should look like this:
☐ E  age <= 70
☐ D  age >= 60
☐ C  sex = M
☐ B  Resp >= 2:0:0
☐ A  ECG-3 ?
The final step is to combine all of these results. The 'And' button combines two or more selected lists, producing a new list containing those records that belong to all of its input lists. Select each of the five lists now by clicking on its checkbox, then click 'And' to generate list F:
☐ F  E ∩ D ∩ C ∩ B ∩ A
☐ E  age <= 70
☐ D  age >= 60
☐ C  sex = M
☐ B  Resp >= 2:0:0
☐ A  ECG-3 ?
The final list contains 131 records, all from the 'mimic2wdb/matched/' data collection (the MIMIC II Waveform Database Matched Subset). You can view or download the list of records if you wish by clicking on the link next to F, or you can examine the records using the PhysioBank ATM. If you wish to do this, select list F by clicking its checkbox, then click on 'Choose'. The PhysioBank ATM will appear in place of the PhysioBank Record Search page, and the first record belonging to list F will be preselected as input.
If you are unfamiliar with the ATM, read its on-line help (visible below the How to use the PhysioBank ATM heading), then click '*' (in the ATM control panel under Navigation) to dismiss the help and display the first 10 seconds of the first record. You can use any of the ATM's controls to examine the record as you wish; when you are ready, click on 'Next record' to view the next record in list F.
The '+' and '-' buttons in the ATM control panel are active only while you are examining search results. Use them to mark individual records of interest. When you click on either of them, a '+' or '-' appears after the record's name in the ATM's Record menu (control panel, upper left). Mark at least one record now.
After reviewing as many records as you wish, return to the search page (for example, by clicking the PhysioBank Record Search button in the ATM's page header). When you do, you will see that one or two new lists have been created below list F. These lists, tagged as F+ and F-, contain only those records that you have marked using the ATM's '+' and '-' buttons. You can select list F again and click 'Choose' to return to the ATM if you wish to mark additional records, and these lists will be updated as you do so. Although lists F+ and F- are described as “accepted from F” and “rejected from F”, you can use these lists in any way you wish. Note, however, that a list made by combining them with other lists will not update itself automatically if you make changes in them later on; to get updated results, repeat the actions you used to combine the lists initially.
If you don’t need to refer back to a list, select it and click on Erase to discard it. This helps to avoid confusion if you use PhysioBank Record Search frequently.
Your results are retained for about a week; after that, they will be removed. Download them if you wish to keep them (especially if you have invested effort in choosing individual records using the ATM). When you return, your previous results are identified using a browser cookie (pbs_id); if you use more than one browser, or more than one computer, you will have different cookies (hence different sets of results) for each one unless you synchronize your cookies.
Using standard command-line utilities to search the PhysioBank Index
Begin by downloading the Index using any of the methods above. Open a terminal emulator window and navigate to the directory in which you saved physiobank-index.
There are five records in PhysioBank that include a left ventricular stroke volume signal, which is labelled SV. Finding them is simple: type
grep SV physiobank-index
and the results appear in your terminal window quickly:
edb/e0112 AnnR1 atr/(SVTA 250 16 49 edb/e0114 AnnR1 atr/(SVTA 250 3 9 edb/e0607 AnnR1 atr/(SVTA 250 1 2 edb/e1304 AnnR1 atr/(SVTA 250 2 3 incartdb/I33 Info2 PVCs, SVEBs, supraventricular couplets incartdb/I34 Info2 SVEBs, couplets, paroxysmal supraventricular tachycardia with aberrated QRS incartdb/I70 Info2 SVEBs with wide QRS, WPW incartdb/I73 Info2 PVCs on bradycardia, SVEBs, couplets mghdb/mgh050 Info50 34 min VT vs. SVT with ABERRANCY mghdb/mgh208 Info10 Runs of VT/SVT mitdb/114 AnnR1 atr/(SVTA 360 1 5 mitdb/201 AnnR1 atr/(SVTA 360 1 2 mitdb/207 Info7 episode of SVTA. mitdb/207 AnnR1 atr/(SVTA 360 1 51 mitdb/209 AnnR1 atr/(SVTA 360 10 102 mitdb/220 AnnR1 atr/(SVTA 360 8 14 mitdb/222 AnnR1 atr/(SVTA 360 4 8 mitdb/234 AnnR1 atr/(SVTA 360 1 26 qtdb/sel114 AnnR6 atr/(SVTA 250 1 5 slpdb/slp59 SV1 SV 250 7.93846 adu/ml 14400 slpdb/slp60 SV1 SV 250 7.90293 adu/ml 21300 slpdb/slp61 SV1 SV 250 958.995 adu/cc 22200 slpdb/slp66 SV1 SV 250 9.957 adu/ml 13200 slpdb/slp67x SV1 SV 250 5.25615 adu/ml 4620 vfdb/426 AnnR1 atr/(SVTA 250 3 81 vfdb/607 AnnR1 atr/(SVTA 250 2 26 vfdb/611 AnnM1 qrs/(SVTA 250 16 49 vfdb/611 AnnR2 atr/(SVTA 250 1 0 vfdb/611 AnnR3 qrsc/(SVTA 250 16 49
Most of these results are records containing supraventricular tacharrhythmias (annotated as '(SVTA'), and others contain SV in comments. These are easily ignored, but it's also possible to improve the search using either
grep $'\tSV\t' physiobank-index
(if you are using the bash shell), or
grep -P '\tSV\t' physiobank-index
(if you are using GNU grep). Either of these commands interprets the sequence \t as a tab character in the search pattern, so that the results contain only lines from the index in which SV appears surrounded by tabs (i.e., in a column by itself):
slpdb/slp59 SV1 SV 250 7.93846 adu/ml 14400 slpdb/slp60 SV1 SV 250 7.90293 adu/ml 21300 slpdb/slp61 SV1 SV 250 958.995 adu/cc 22200 slpdb/slp66 SV1 SV 250 9.957 adu/ml 13200 slpdb/slp67x SV1 SV 250 5.25615 adu/ml 4620
Getting (re)acquainted with the command line
If you've ever used any version of Unix, or even MS-DOS, the examples on this page may look familiar. If not, consult any introductory book or on-line tutorial about Unix or GNU/Linux. Here are a few places to start:
If we want to find records that have at least 3 ECG signals, we can look for ECG3:
grep ECG3 physiobank-index
This results in a very long list of records that quickly scrolls off the screen. If we want to know how long the list is, we can use wc to count the lines:
grep ECG3 physiobank-index | wc -l
(The pipe symbol, '|', connects a pair of commands; it means ”take the standard output of the command on the left and feed it to the standard input of the command on the right”.) When this page was written, there were 25,372 recordings with at least 3 ECG signals in PhysioBank. We can save the entire list by redirecting the standard output into a file, like this:
grep ECG3 physiobank-index >ECG3-records
The '>' collects the standard output of the command, which would otherwise be shown in the terminal window, and saves it in a file (ECG3-records).
Suppose what we really want are the longest such recordings. Here's how to find the 3 longest cases:
grep ECG3 physiobank-index | cut -f 1,6 | sort -nr -k2 | head -3
(This command uses pipes to chain four commands together, each one reading the output of the previous one; cut selects the first and sixth fields — the record name and the duration — from each line output by grep; sort rearranges the lines in reverse numerical order of the second field output by cut; and head discards all but the first three lines output by sort.) The output lists 3 recordings, each containing over 400 hours of ECG3:
mimic2wdb/31/3101148/ 2299827 mimic2wdb/37/3752730/ 1845755 mimic2wdb/32/3212213/ 1842540
There is a caveat, however: these recordings are all from the MIMIC II database, and the signals are not necessarily continuous; in fact, they may not even be simultaneously available. To find a set of long records with at least 3 continuous, simultaneous ECG signals, we can exclude the MIMIC databases and the similar Challenge 2009 database from the search:
grep ECG3 physiobank-index | \ egrep -i -v "mimic|challenge/2009" | \ cut -f 1,6 | sort -nr -k2 | head -3
(Here the \ characters indicate the command continues on the following line.) The results are:
ltstdb/s30691 85860 ltstdb/s30731 85845 ltstdb/s30801 85821
These examples illustrate the flexibility of using standard command-line tools to search within the PhysioBank Index. If these tools are already familiar, it’s easy to perform much more complex searches, including many that would be very difficult to perform using a relational database and SQL.
How many records are in the index?
As of January 2012, over 36,000 record sets from over 50 collections are included in the PhysioBank Index. Many record sets include two or more records, and some records belong to more than one collection, so the number of record names in the index is nearly 73,000.
The MIMIC Database and the MIMIC II Waveform Database consist of record sets (pairs of records acquired simultaneously from each subject: a waveform record of signals sampled at 125 or 500 Hz, and a numerics record of vital signs sampled once per second or once per minute).
Records (or excerpts of records) may belong to more than one data collection:
Each file listed below contains an index for a PhysioBank data collection (or part of one). These files are concatenated to form the PhysioBank Index.
Name Last modified Size Description Parent Directory - vfdb 13-Feb-2012 21:00 5.9K MIT-BIH Malignant Ventricular Ectopy Database ucddb 03-Mar-2012 08:19 22K UCD Sleep Apnea Database twadb 13-Feb-2012 21:06 53K T-Wave Alternans Challenge Database tpehgdb 25-Feb-2012 03:38 317K Term-Preterm EHG Database szdb 13-Feb-2012 21:02 2.4K Post-Ictal Heart Rate Oscillations in Partial Epilepsy svdb 14-Feb-2012 18:53 23K MIT-BIH Supraventricular Arrhythmia Database stdb 13-Feb-2012 21:01 5.0K MIT-BIH ST Change Database slpdb 13-Feb-2012 21:01 10K MIT-BIH Polysomnographic Database sleep-edfx 23-Oct-2013 00:08 33K Sleep-EDF Database sleep-edf 25-Feb-2012 03:37 3.6K Sleep-EDF Database simfpcgdb 12-Feb-2014 00:22 2.4K shhpsgdb 13-Feb-2012 21:04 1.0K Sleep Heart Health Study Polysomnography Database shareedb 31-Jan-2015 04:37 44K European ST-T Database sddb 13-Feb-2012 21:04 17K Sudden Cardiac Death Holter Database qtdb 14-Feb-2012 18:53 178K QT Database ptbdb 13-Feb-2012 21:03 2.1M PTB Diagnostic ECG Database ob1db 01-Aug-2012 01:40 288 nstdb 14-Feb-2012 18:53 5.6K MIT-BIH Noise Stress Test Database nsrdb 14-Feb-2012 18:53 6.5K MIT-BIH Normal Sinus Rhythm Database nsr2db 13-Feb-2012 21:01 14K Normal Sinus Rhythm RR Interval Database nifecgdb 13-Feb-2012 21:01 33K Non-Invasive Fetal ECG Database mvtdb_data 25-Feb-2012 03:37 55K Spontaneous Ventricular Tachyarrhythmia Database motion-artifact 04-Mar-2014 00:11 19K mitdb 13-Feb-2012 21:00 28K MIT-BIH Arrhythmia Database mimicdb_numerics 08-Jul-2015 00:03 97K MIMIC Database Numerics mimicdb 13-Feb-2012 16:58 64K MIMIC Database mimic2wdb_matched 08-Jul-2015 10:39 8.5M MIMIC II Waveform Database Matched Subset mimic2wdb_39 08-Jul-2015 08:20 4.4M MIMIC II Waveform Database, version 3 part 9 mimic2wdb_38 08-Jul-2015 07:32 4.3M MIMIC II Waveform Database, version 3 part 8 mimic2wdb_37 08-Jul-2015 06:46 4.2M MIMIC II Waveform Database, version 3 part 7 mimic2wdb_36 08-Jul-2015 06:00 4.1M MIMIC II Waveform Database, version 3 part 6 mimic2wdb_35 08-Jul-2015 05:12 4.4M MIMIC II Waveform Database, version 3 part 5 mimic2wdb_34 08-Jul-2015 04:23 4.2M MIMIC II Waveform Database, version 3 part 4 mimic2wdb_33 08-Jul-2015 03:33 4.3M MIMIC II Waveform Database, version 3 part 3 mimic2wdb_32 08-Jul-2015 02:25 4.0M MIMIC II Waveform Database, version 3 part 2 mimic2wdb_31 08-Jul-2015 01:38 4.5M MIMIC II Waveform Database, version 3 part 1 mimic2wdb_30 08-Jul-2015 00:49 4.5M MIMIC II Waveform Database, version 3 part 0 mimic2db_numerics 08-Jul-2015 11:01 3.5M MIMIC II Waveform DB, v2 Numerics [deprecated, use v3] mimic2db 27-Feb-2012 00:56 4.8M MIMIC II Waveform DB, v2 [deprecated, use v3] mimic2cdb 13-Feb-2012 16:59 7.9K MIMIC II Clinical Database Public Subset mghdb 13-Feb-2012 16:57 513K MGH/MF Waveform Database meditation_data 25-Feb-2012 03:36 13K Heart Rate Oscillations during Meditation ltstdb 13-Feb-2012 16:57 272K Long Term ST Database ltdb 14-Feb-2012 18:52 2.7K MIT-BIH Long-Term ECG Database ltafdb 25-Feb-2012 03:36 27K Long Term AF Database incartdb 13-Feb-2012 21:04 60K St Petersburg INCART 12-lead Arrhythmia Database iafdb 13-Feb-2012 16:56 23K Intracardiac Atrial Fibrillation Database gaitndd 13-Feb-2012 16:56 6.3K Gait in Neurodegenerative Disease Database gaitdb 13-Feb-2012 16:55 1.3K Gait in Aging and Disease Database gait-maturation-db_data 26-Feb-2012 22:31 5.8K Gait Maturation Database fantasia 13-Feb-2012 16:55 14K Fantasia Database excluded 14-Feb-2012 18:53 1.0K Recordings excluded from nsrdb erpbci 12-Feb-2014 00:22 943K emgdb 13-Feb-2012 16:55 1.1K Examples of Electromyograms ehgdb 23-Dec-2014 05:53 247K eegmmidb 13-Feb-2012 16:55 5.2M EEG Motor Movement/Imagery Dataset edb 13-Feb-2012 16:51 50K European ST-T Database ecgiddb 08-Mar-2014 00:13 106K earndb 25-Feb-2012 03:35 702K Evoked Auditory Responses in Normals drivedb 13-Feb-2012 21:04 5.4K Stress Recognition in Automobile Drivers cudb 13-Feb-2012 16:45 7.4K CU Ventricular Tachyarrhythmia Database ctu-uhb-ctgdb 19-Feb-2014 02:10 1.0M crisdb 13-Feb-2012 16:45 923K CAST RR Interval Sub-Study Database chfdb 14-Feb-2012 18:52 6.6K BIDMC Congestive Heart Failure Database chf2db 13-Feb-2012 16:41 8.8K Congestive Heart Failure RR Interval Database chbmit 26-Feb-2012 20:58 1.1M CHB-MIT Scalp EEG Database challenge_2015_training 07-Mar-2015 03:02 235K challenge_2014_set-p2 04-Mar-2015 00:18 6.6K challenge_2014_set-p 25-Oct-2014 00:13 45K challenge_2013_set-b 03-Apr-2013 00:10 22K challenge_2013_set-a 03-Apr-2013 00:10 17K challenge_2011_sim 25-Feb-2012 03:28 24K Challenge 2011 Pilot Set challenge_2011_set-b 13-Feb-2012 16:39 345K Challenge 2011 Test Set B challenge_2011_set-a 13-Feb-2012 16:37 690K Challenge 2011 Training Set A challenge_2010_set-c 13-Feb-2012 16:35 38K Challenge 2010 Test Set C challenge_2010_set-b 13-Feb-2012 16:35 38K Challenge 2010 Test Set B challenge_2010_set-a 13-Feb-2012 16:34 36K Challenge 2010 Training Set A challenge_2009_test-set-b 26-Feb-2012 20:56 340K Challenge 2009 Test Set B challenge_2009_test-set-a 26-Feb-2012 20:55 79K Challenge 2009 Test Set A cebsdb 12-Dec-2014 00:16 19K cdb 13-Feb-2012 21:00 13K MIT-BIH ECG Compression Test Database capslpdb 26-Jul-2012 01:58 122K MIT-BIH Polysomnographic Database bpssrat 26-Jun-2012 14:55 2.7K Blood Pressure in Salt-Sensitive Dahl Rats apnea-ecg 13-Feb-2012 16:33 36K Apnea-ECG Database ahadb 26-Feb-2012 20:50 438 AHA Database [sample excluded record] aftdb 13-Feb-2012 16:33 24K AF Termination Challenge Database afpdb 25-Feb-2012 03:37 72K PAF Prediction Challenge Database afdb 14-Feb-2012 18:52 9.7K MIT-BIH Atrial Fibrillation Database adfecgdb 13-Dec-2012 00:11 2.6K aami-ec13 13-Feb-2012 16:33 1.0K ANSI/AAMI EC13 Test Waveforms