Database Open Access

Wearable-based signals during physical exercises from patients with frailty after open-heart surgery

Daivaras Sokas Monika Butkuvienė Egle Tamulevičiūtė-Prascienė Aurelija Beigienė Raimondas Kubilius Andrius Petrėnas Birutė Paliakaitė

Published: March 31, 2022. Version: 1.0.0

When using this resource, please cite: (show more options)
Sokas, D., Butkuvienė, M., Tamulevičiūtė-Prascienė, E., Beigienė, A., Kubilius, R., Petrėnas, A., & Paliakaitė, B. (2022). Wearable-based signals during physical exercises from patients with frailty after open-heart surgery (version 1.0.0). PhysioNet.

Please include the standard citation for PhysioNet: (show more options)
Goldberger, A., Amaral, L., Glass, L., Hausdorff, J., Ivanov, P. C., Mark, R., ... & Stanley, H. E. (2000). PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals. Circulation [Online]. 101 (23), pp. e215–e220.


This data collection contains a single-lead electrocardiogram and triaxial acceleration signals of 80 elderly frailty patients, who entered a cardiac rehabilitation program after open-heart surgery. The signals were collected using a chest-based heart rate monitor worn by the patients performing a series of exercise tests for health status assessment. The tests include veloergometry, 6-minute walk test, stair climbing test, timed up and go test, and gait analysis on a treadmill. The signals are supplemented with the demographic and clinical characteristics of the patients including the Edmonton frail scale score and quantitative results of the exercise tests. This data collection could be particularly useful in developing wearable technology-based methods for (1) assessment of gait and balance in frailty patients and (2) monitoring of heart response to daily physical stressors.


Frailty is becoming a major challenge of the aging population [1]. The prevalence of frailty is 17%, whereas pre-frailty may affect up to 60% of individuals ≥ 65 years [2, 3]. Frailty is a geriatric syndrome of reduced physiological reserve leading to increased vulnerability to physical stressors and susceptibility to cardiovascular diseases [4, 5]. As a consequence, frailty is often associated with an increased risk of various adverse healthcare outcomes such as the loss of mobility, disability, emergency visits, institutionalization, hospitalization, and death [6]. Fortunately, frailty can be improved or reversed for some individuals by a timely and proper intervention such as exercise training [7].

Despite the abundance of various indexes and questionnaires for frailty assessment [8], none has been proven to be sufficiently sensitive for detecting pre-frailty, or suitable for monitoring and routine assessment of frailty dynamics in the course of exercise training. Thus, there is a growing interest in new methods, often employing wearable devices, that would enable earlier identification of frailty or routine assessment of frailty-reflecting measures at more frequent time intervals in an unobtrusive way [9]. Such measures should preferably detect and assess physical and autonomic dysfunction, both associated with worsening of frailty [9, 10]. Changes in physical function are reflected in physical activity, gait, and balance measures, and thus the analysis of kinematic signals from wearable inertial sensors is a common assessment approach [8]. Proper cardiovascular functioning is greatly dependent on the autonomic nervous system, and thus autonomic dysfunction may manifest in impaired heart response to physical stressors, which in turn can be assessed using wearable devices with embedded sensors capturing heart activity.

In this data collection, acceleration and electrocardiogram (ECG) signals are the chosen modalities of kinematic and heart activity signals, respectively. The data was obtained from frailty patients entering cardiac rehabilitation program after open-heart surgery. Such population was chosen because frailty in patients after open-heart surgery influences the type and intensity of a cardiac rehabilitation program. Since the response of a frail patient to a tailored exercise training can be very different, convenient tools to assess the effectiveness of the training program routinely are needed. The signals were acquired while patients performed standardized exercise tests enabling assessment of physical and cardiac autonomic function. Most of the selected exercise tests can be modified and implemented routinely and sometimes are even performed unintentionally in activities of daily living. The provided signals and clinical data can be reused in numerous ways, including but not limited to: (1) developing algorithms for automatic detection of exercise tests in acceleration signals; (2) developing wearable technology-based methods for frailty assessment; (3) developing and investigating new frailty reflecting measures; (4) testing methods for quantification of heart response to physical stressors; (5) investigating associations between clinical data, frailty status, and measures estimated from the signals.


Population and Frailty Assessment

Patients after open-heart surgery, who arrived at Kulautuva Rehabilitation Hospital of Kaunas Clinics (Kulautuva, Lithuania) from the 19th of November 2020 till the 3rd of January 2022, were invited to participate in the study. As a part of the routine clinical evaluation at the beginning of inpatient rehabilitation, healthcare specialists evaluated the degree of frailty based on the Edmonton frail scale (EFS), which assesses nine domains of frailty, namely cognition, general health status, functional independence, social support, medication usage, nutrition, mood, continence, functional performance [11], and asked all patients to perform a 6-minute walk test (6MWT). Out of 337 patients assessed for eligibility, 99 fulfilled the inclusion criteria: (1) age ≥ 65 years; (2) EFS score ≥ 4; (3) 6MWT distance ≥ 150 meters; (4) agreement to participate in the study. The signals from 19 patients were lost due to technical issues, thus data from 80 patients are provided in this data collection.

Study Protocol

ECG, sampled at 130 Hz, and triaxial acceleration signals, sampled at 200 Hz, were obtained using Polar H10 (Polar Electro OY, Kempele, Finland) wearable device. The device was placed below the chest with a textile strap and the signals were sent to a smartphone via a Bluetooth connection in real-time. Patients carried the smartphone in a holder wrapped around the upper arm throughout the monitoring period, except when a healthcare specialist took it out to log the onset of the exercise test with a mobile app. In this study, patients performed five submaximal tests, namely veloergometry, 6MWT, stair climbing, timed up and go test (TUG), and gait analysis. Before and after each physical test, patients were asked to rest in a sitting position for at least 3 minutes. Patients were free to pause or end the test at any time they felt uncomfortable.

Veloergometry is an exercise test used to evaluate the cardiovascular system under conditions of increasing physical load. The patient was asked to cycle on Viasprint 150P ergometer (Ergoline GmbH, Bitz, Germany) starting with a 25 W load and increasing it by 12.5 W per minute until subjective exhaustion or occurrence of any abort criteria, namely shortness of breath, chest pain, leg fatigue, or systolic arterial blood pressure drop 20 mmHg below the initial level or rise above 220 mmHg. Clinical parameters assessed during veloergometry include test duration, maximal load, and maximal heart rate reached.

6MWT is a well-established, convenient, safe, and inexpensive test suitable to assess functional performance [12]. During the 6MWT, the distance an individual can walk on a flat surface under the encouragement of a supervising staff member is measured and compared to the individual-specific reference distance.

Stair climbing was selected as an exercise test assuming that terrain-dependent peculiarities alter gait and balance differently [13]. To perform this test, the patient was asked to climb a set of 12 stairs at a convenient pace without help from a supervising staff member. No clinical parameter was recorded during the test.

TUG test is often used to evaluate balance and assess the risk of falls. During the test, after the command of a supervising staff member, the patient stands up and walks 3 meters forward, then turns around, walks back to the chair, and sits down. The time needed to perform the TUG test was recorded as a clinical parameter.

Gait analysis was performed using a gait and stance analysis system Zebris FDM-T (Zebris Medical GmbH, Isny im Allgäu, Germany) with a treadmill. After the rest period and before the actual gait analysis, the patient was asked to walk on the treadmill to get used to it. After the preparation, the patient walked on the treadmill for 30 s and the gait of this period alone was analyzed by the system. The generated report contained a number of gait parameters, including step length, stride length, step width, stance phase, swing phase, double stance phase, step time, stride time, cadence, and velocity. The system also provided a number of balance parameters obtained from the center of pressure analysis, such as length of gait line, single limb support line, anterior/posterior position, lateral symmetry, and maximal gait line velocity.

Data Description

There are wearable-based signals and clinical data from 80 individual patients in this data collection. Synchronously acquired ECG and triaxial acceleration signals are provided in separate WFDB-compatible records because of different sampling frequencies. If signal acquisition lasted for a single day without interruptions, there is only a single pair of ECG and acceleration records per patient (a single recording session). However, in some cases, the acquisition was interrupted and/or performed on two consecutive days, and thus there are 2 or more pairs of ECG and acceleration records per patient (multiple recording sessions). In total, there are 196 records in this data collection organized in 2 folders according to the signal type.

The records are named as follows:


where xxx is the patient ID, y is the recording session number, and zzzis ecg for the ECG signal and acc for the acceleration signals (ACC_X, ACC_Y, and ACC_Z).

For each record, signals are stored in a signal file (.dat). The record’s header file (.hea) specifies its name, duration, storage format, sampling frequency, and the time of day the signal acquisition started. The record’s annotation file (.atr) contains reference annotations of the onsets of the exercise tests, that have been manually checked for accuracy. Gait analysis is an exception because its annotation mark is a system-provided time of the generated gait analysis report, which has not been manually checked for accuracy. The following codes are used for annotations:

VELO– the onset of veloergometry

6MWT– the onset of the 6MWT

STAIR – the onset of stair climbing

TUG – the onset of the TUG test

GAIT_ANALYSIS – the time of the gait analysis (based on the Zebris system-generated report)

subject-info.csv contains the demographic and clinical characteristics of the patients. Each row is allocated to an individual patient and states patient ID, age, gender, height, weight, EFS score, days after surgery and surgery type, whether the patient is diagnosed with comorbidities related to frailty, and whether the patient is using heart rate-altering medications. The file also contains clinical parameters assessed during the performed exercise tests: 6MWT distance, TUG time, veloergometry measures, and gait and balance parameters from the Zebris report.

test-availability.csv lists which exercise tests are available for each patient and provides the name (without ecg or _acc suffix) of the record containing a particular exercise test. Some tests, marked with a dash, are unavailable because their onsets were not logged with the mobile app or the signals of the entire recording session were not transferred to the smartphone.

Usage Notes

The provided data (as a part of a larger data collection) has already been used in studies aiming to investigate whether kinematic measures extracted from the acceleration signals [14] and ECG-based measures of heart rate response to physical stressors [15] can provide information about frailty transitions during cardiac rehabilitation.

The provided data collection has a great potential for reuse in numerous ways, including but not limited to: (1) developing algorithms for automatic detection of exercise tests in acceleration signals; (2) developing wearable technology-based methods for frailty assessment; (3) developing and investigating new frailty reflecting measures; (4) testing methods for quantification of heart response to physical stressors; (5) investigating associations between clinical data, frailty status, and measures estimated from the signals.  

However, when using the signals, researchers should be aware of these limitations: (1) since the sampling frequency of the Polar H10 device is not perfectly constant (varies up to 0.2%), the ECG and acceleration signals were resampled at uniform sampling frequencies of 130 Hz and 200 Hz, respectively; (2) some ECG signals might be unsuitable for analysis due to poor quality; (3) in some cases, patients were walking longer than 6 minutes while performing the 6MWT; (4) sometimes motion is present before and/or after the exercise test despite the fact that the patient was asked to rest in a sitting position for at least 3 minutes; (5) the Zebris system provides the time of the gait analysis report to the nearest minute, thus it might be difficult to estimate the start and end of the 30-s-long gait analysis precisely; (6) the GAIT_ANALYSIS annotation marker is preceded not only by the resting period but also by the preparation period; (7) it is possible to place the chest strap upside down, so vertical (ACC_X) and mediolateral (ACC_Y) acceleration signals might be inverted in some cases.

When using the subject-info.csv file, researchers should pay attention to the following: (1) heart failure class according to New York Heart Association (NYHA) classification criteria is not provided for the patient 073; (2) the results of the gait analysis for the patient 254 are somehow erroneous; (3) patients 203 and 250 did not perform gait analysis due to dizziness; (4) patient 269 did not perform veloergometry; (5) stride length and stride time are not provided for the patient 318.

Release Notes

Version 1.0: Initial public release


The study protocol is in accordance with the ethical principles of the Declaration of Helsinki and was approved by Kaunas Region Biomedical Research Ethics Committee (No. BE-2-99). All patients gave written informed consent to participate in the study. The data was collected during a clinical trial registered at (No. NCT04636970).


The acquisition of the data was financially supported by the Research Council of Lithuania (Agreement No. S-MIP-20-54). The authors thank Julius Marozas and Vaidotas Marozas for developing data collecting software for a smartphone.

Conflicts of Interest

The authors have no conflicts of interest to declare.


  1. Dent E, Martin FC, Bergman H, Woo J, Romero-Ortuno R, Walston JD. Management of frailty: opportunities, challenges, and future directions. Lancet. 2019; 394(10206): 1376–1386. DOI: 10.1016/S0140-6736(19)31785-4.
  2. Afilalo J, Alexander KP, Mack MJ, Maurer MS, Green P, Allen LA, et al. Frailty assessment in the cardiovascular care of older adults. J Am Coll Cardiol. 2014; 63(8): 747–762. DOI: 10.1016/j.jacc.2013.09.070.
  3. Beard JR, Officer A, De Carvalho IA, Sadana R, Pot AM, Michel J-P, et al. The World report on ageing and health: a policy framework for healthy ageing. Lancet. 2016; 387(10033): 2145–2154. DOI: 10.1016/S0140-6736(15)00516-4.
  4. Morley JE, Vellas B, Van Kan GA, Anker SD, Bauer JM, Bernabei R, et al. Frailty consensus: a call to action. J Am Med Dir Assoc. 2013; 14(6): 392–397. DOI: 10.1016/j.jamda.2013.03.022.
  5. Veronese N, Cereda E, Stubbs B, Solmi M, Luchini C, Manzato E, et al. Risk of cardiovascular disease morbidity and mortality in frail and pre-frail older adults: Results from a meta-analysis and exploratory meta-regression analysis. Ageing Res Rev. 2017; 35: 63–73. DOI: 10.1016/j.arr.2017.01.003.
  6. Hoogendijk EO, Afilalo J, Ensrud KE, Kowal P, Onder G, Fried LP. Frailty: implications for clinical practice and public health. Lancet. 2019; 394(10206): 1365–1375. DOI: 10.1016/S0140-6736(19)31786-6.
  7. Kojima G, Taniguchi Y, Iliffe S, Jivraj S, Walters K. Transitions between frailty states among community-dwelling older people: a systematic review and meta-analysis. Ageing Res Rev. 2019; 50: 81–88. DOI: 10.1016/j.arr.2019.01.010.
  8. Panhwar YN, Naghdy F, Naghdy G, Stirling D, Potter J. Assessment of frailty: a survey of quantitative and clinical methods. BMC Biomed Eng. 2019; 1(7): 1–20. DOI: 10.1186/s42490-019-0007-y.
  9. Vavasour G, Giggins OM, Doyle J, Kelly D. How wearable sensors have been utilised to evaluate frailty in older adults: a systematic review. J Neuroeng Rehabil. 2021; 18(1): 1–20. DOI: 10.1186/s12984-021-00909-0.
  10. Parvaneh S, Howe CL, Toosizadeh N, Honarvar B, Slepian MJ, Fain M, et al. Regulation of cardiac autonomic nervous system control across frailty statuses: a systematic review. Gerontology. 2016; 62(1): 3–15. DOI: 10.1159/000431285.
  11. Rolfson DB, Majumdar SR, Tsuyuki RT, Tahir A, Rockwood K. Validity and reliability of the Edmonton Frail Scale. Age Ageing. 2006; 35(5): 526–529. DOI: 10.1093/ageing/afl041.
  12. ATS Committee on Proficiency Standards for Clinical Pulmonary Function Laboratories. ATS statement: guidelines for the six-minute walk test. Am J Respir Crit Care Med. 2002; 166(1): 111–117. DOI: 10.1164/ajrccm.166.1.at1102.
  13. Wang K, Delbaere K, Brodie MA, Lovell NH, Kark L, Lord SR, et al. Differences between gait on stairs and flat surfaces in relation to fall risk and future falls. IEEE J Biomed Health Inform. 2017; 21(6): 1479–1486. DOI: 10.1109/JBHI.2017.2677901.
  14. Butkuvienė M, Tamulevičiūtė-Prascienė E, Beigienė A, Barasaitė V, Sokas D, Kubilius R, Petrėnas A. Wearable-based assessment of frailty transitions during cardiac rehabilitation after open-heart surgery. Submitted for publication.
  15. Sokas D, Tamulevičiūtė-Prascienė E, Beigienė A, Barasaitė V, Marozas J, Kubilius R, Bailón R, Petrėnas A. Wearable-based assessment of heart rate response to physical stressors in patients after open-heart surgery with frailty. Submitted for publication.


Total uncompressed size: 2.5 GB.

Access the files

Visualize waveforms

Folder Navigation: <base>
Name Size Modified
ANNOTATORS (download) 213 B 2022-02-07
LICENSE.txt (download) 0 B 2022-03-25
RECORDS (download) 2.9 KB 2022-02-07
SHA256SUMS.txt (download) 48.0 KB 2022-03-31
subject-info.csv (download) 20.9 KB 2022-02-08
test-availability.csv (download) 2.6 KB 2022-02-08