Skip to content
Publicly Available Published by De Gruyter February 18, 2014

Individualized biomonitoring in heart failure – Biomon-HF “Keep an eye on heart failure – especially at night”

  • Thomas Vollmer EMAIL logo , Patrick Schauerte , Matthias Zink , Sigrid Glöggler , Johannes Schiefer , Michael Schiek , Udo Johnen and Steffen Leonhardt

Abstract

In the project “Individualized Biomonitoring in Heart Failure (Biomon-HF),” innovative sensors and algorithms for measuring vital signs, i.e., during the nocturnal sleep period, have been developed and successfully tested in five clinical feasibility studies involving 115 patients. The Biomon-HF sensor concepts are an important step toward future patient-customized telemonitoring and sensor-guided therapy management in chronic heart failure, including early detection of upcoming HF exacerbation and comorbidities at home. The resulting preventable disease complications and emergencies and reduction of consequences of disease are very important advantages for the patients, causing relief for medical staff and, thus, offer an enormous potential for improvements and cost savings in healthcare systems.

Introduction

Chronic heart failure (CHF) is a disease with the highest mortality rate for elderly people, with a prevalence of 5–10% for persons aged 65 and older. As a primary single causal and curative therapy is rarely possible, the treatment often simply aims for the prevention of complications (palliation). Therefore, the diagnosis and treatment of concomitant diseases (comorbidities) are of an outstanding importance. Besides that, an early detection of an exacerbation of the clinical pattern is central.

The care and the backing of affected patients, especially in rural areas, become increasingly complicated as the density of family doctors decreases. Against this backdrop, researchers from Aachen, Germany, aim for the development of individualized nightly telemonitoring utilizing external, noninvasive sensors for measurement-guided ambulant therapy management in heart failure, timely diagnosis of HF exacerbation, and early detection of concomitant diseases, i.e., arrhythmia, sleep-disordered breathing, and hypertension, at home.

For this purpose, the sleeping period of the patients which amount to 30% of a day, is used to monitor the vital parameters more intensively and more comfortably at home. This sleeping period enables an optimized signal quality of the sensors, due to the more stable and relatively motionless measurement period.

The remote supervision of the patients has a huge potential for cost savings, resulting from reductions of emergencies and relief for medical personnel.

This paper will give an overview about the Biomon-HF project.

Biomon-HF work packages

The Biomon-HF project has been organized in four parallel sensor and algorithm development-related work packages (WP 1–WP 4) and one cross-cutting work package for the clinical studies and development of an electronic study data management platform (WP 5).

WP 1: Bioimpedance spectroscopy

In work package 1, the feasibility of using bioelectrical impedance measurements to detect pulmonary edema is investigated. Bioelectrical impedance spectroscopy (BIS) is a measurement technique based upon the different electrical properties of body tissues. Owing to the conductive behavior of the body fluid, the bioelectrical impedance is related to the fluid volume and its distribution. The BIS is a low-cost, rapid, accurate, and noninvasive method. BIS allows measurement of electrical body impedance at various frequencies, generally ranging from 5 kHz to 1 MHz [6, 12–15].

WP 2: Ballistocardiography bed sensor

Recording vibrations caused by the mechanical activity of the heart by means of highly sensitive sensors integrated into the bed frame or mattress is one popular approach for unobtrusively measuring cardiac activity. These methods have been given a variety of names such as ballistocardiography (BCG), mechanocardiography, kinetocardiography, or seismocardiography [2]. The sensor principle is shown in Figure 1. In work package 2, the feasibility of ballistocardiography for detecting cardiac arrhythmia and sleep-disordered breathing is investigated.

Figure 1 Ballistocardiography bed sensor principle.
Figure 1

Ballistocardiography bed sensor principle.

WP 3: Individualized blood pressure monitoring

Nocturnal monitoring of blood pressure during a 24-h ambulatory cycle is still characterized by discomfort, either caused by the noise in inflation or by excessive pressure exerted to the upper arm of the patient. This often is the reason for noncompliance to the method, when indicated over a longer time period for weeks or months.

Within the Biomon-HF work package 3 “individualized nocturnal blood pressure monitoring for ambulatory therapy guidance”, the need for an increase in compliance to the measurement method was identified and expressed as a target. The key to achievement was seen in finding the most “gentle” measurement procedure in order to decrease the discomfort of patient sleep.

Commercially available devices for oscillometric blood pressure measurement come along with the major disadvantage of cuff inflation to suprasystolic blood pressure. This procedure today is required for the most accurate determination of blood pressure. Yet, biomathematical models more and more allow for the calculation of central hemodynamic indices, e.g., central (aortic) blood pressure, based upon secondary data derived by oscillometry at lower pressure levels; secondary data such as pulse waveform or velocity. This relatively new concept has provided the pathway for the idea to record secondary data at low pressure level and to use this information to derive blood pressure values.

Following this path, an individual calibration of calculations on a per-patient base is mandatory – only an individual calibration allows for an accurate determination of recordings within the physiological bandwidth of the measured patient.

Related technical questions to the idea, such as the required sampling rate and data volume, could only be observed in a real patient setup.

WP 4: In-ear photoplethysmograpy (PPG) sensor

In work package 4, the feasibility of detecting sleep apnea and atrial fibrillation with a novel reflective pulseoximetric sensor (Figure 2), which can be worn in the ear channel is investigated. In the previous work, it could be demonstrated that measurement of heart rate (HR), arterial oxygen saturation (SpO2), and respiratory-related information can be performed with reliable accuracy [7, 8]. As these results were obtained under laboratory conditions, in Biomon-HF, the in-ear sensor is studied in a clinical sleep laboratory and atrial fibrillation study setting.

Figure 2 In-ear PPG sensor [9].
Figure 2

In-ear PPG sensor [9].

WP 5: Clinical studies and electronic study data management platform

Clinical studies

For feasibility evaluation of the Biomon-HF sensor concepts, five clinical study settings were defined, which allow for comparison of the deployed sensors with clinical reference measurements.

The defined study settings are:

  1. Cardiac recompensation – noninvasive thoracic bioimpedance and respiratory measurements of decompensated patients during recompensation compared to standard measurements.

  2. Pleural effusion – noninvasive thoracic bioimpedance and respiratory measurements compared to standard measurements before and after pleural drainage.

  3. Atrial fibrillation (AF) – ballistocardiography and in-ear photoplethysmography (PPG) sensor measurements to investigate the heartbeat before and after cardioversion of atrial fibrillation.

  4. Sleep apnea – ballistocardiography and in-ear photoplethysmography (PPG) sensor compared to polysomnography (PSG) at the sleep laboratory to investigate the heartbeat and respiration during apnea episodes.

  5. Tilt table – course and behavior of blood pressure and heartbeat in a patient with orthostatic syncopal diagnostic on a tilt table measured by a novel noninvasive tool.

Study approval

All five studies received a positive approval by the Ethics Committee of the University Hospital of Aachen. This procedure includes generating study protocols, informed consent, and a patient information brochure for each study and ethical and regulatory classification according to the German Medical Device Act (MPG) and “Berufsordnung für Ärztinnen und Ärzte” (BOÄ).

Screening, recruiting, measuring of patients, and documentation were performed by physicians assisted by four medical PhD students. In addition, support was given by the Clinical trial center of the University Hospital of Aachen as well as the associated engineers according to good clinical practice (GCP), regulatory requirements, and laws.

The clinical trial period of each study was defined as 1 year, with up to 50 patients per study. The recruiting was finished in time for all trials.

Electronic study data management platform

For conducting the measurements in the clinical studies, two measurement carts have been designed carrying the Biomon-HF research sensors required in a particular study as well as clinical reference measurements provided by a patient monitor with an integrated PC (iPC).

Figure 3 shows the measurement cart for the sleep apnea study during an electrical safety test in the sleep laboratory of the University Hospital Aachen.

Figure 3 EN IEC 60601-1 electrical safety testing of Biomon-HF measurement cart at UKA sleep laboratory.
Figure 3

EN IEC 60601-1 electrical safety testing of Biomon-HF measurement cart at UKA sleep laboratory.

For the management of the collected study data, a full electronic study data management workflow and system has been developed and implemented as shown in Figure 4. All sensor data are captured semiautomatically by an electronic data capture (EDC) application (session control application) shown in Figure 5. The EDC application automatically sends clinical study data in CDISC ODM format over a web service interface to an electronic Case Report Form (eCRF) system (OpenClinica, Figure 6).

Figure 4 Overview of full electronic clinical study data management system implemented in Biomon-HF.
Figure 4

Overview of full electronic clinical study data management system implemented in Biomon-HF.

Figure 5 Biomon-HF session control application.
Figure 5

Biomon-HF session control application.

Figure 6 Example of an OpenClinica eCRF form.
Figure 6

Example of an OpenClinica eCRF form.

Results

At the time of writing, the data analysis of the clinical study data is still ongoing, and therefore, in this section, only some first results can be briefly reported. For a detailed description of these first results, the interested reader is referred to the publications written by the Biomon-HF project members thus far [1–20].

Bioimpedance spectroscopy (WP 1)

In preparation for the cardiac recompensation study, an animal study has been conducted in the course of the project to investigate the influence of extracellular water on the bioimpedance. The multifrequency impedance measurements were performed in five female pigs. Animals were connected to an extracorporeal membrane oxygenation device during a lung disease experiment, and fluid balance was recorded. Every 15 min, the amount of fluid infusion and the weight of the urine drainage bag was recorded. From the fluid intake and output, the fluid balance was calculated. These data were compared with values calculated from a mathematical model, based on the extracellular tissue resistance and the Hanai Mixture theory. The extracellular tissue resistance was also measured with BIS [13]. Figure 7 shows the electrode position on the animal.

Figure 7 BIS electrode position on the animal [13].
Figure 7

BIS electrode position on the animal [13].

The results of measurements in one pig are shown in Figure 8. As expected, body impedance decreased with an increase in infused volume.

Figure 8 Plot of BIS measurement on the complex impedance plane, the fading color indicates the increase of liquid [13].
Figure 8

Plot of BIS measurement on the complex impedance plane, the fading color indicates the increase of liquid [13].

In Figure 9, some example daily BIS measurements from one patient in the cardiac recompensation study are shown from day 1 to day 8 of hospitalization. The BIS measurements show a clear trend while the patient is recompensating.

Figure 9 Example BIS measurement from the cardiac recompensation study.
Figure 9

Example BIS measurement from the cardiac recompensation study.

Ballistocardiography bed sensor (WP 2)

The signals measured with the ballistocardiography bed sensor contain a mixture of cardiac and respiratory components, which can potentially be used for detection of comorbidities of heart failure like apnea or arrhythmia. However, the separation of the cardiac and respiratory components has proven to be difficult, as there is significant overlap in the spectra of both components. In WP 2, an algorithm for the separation task has been developed, which can overcome the problem of overlapping spectra. Additionally, a model has been developed for the generation of artificial ballistocardiograms, which are used to analyze the separation performance. Furthermore, the algorithm has been tested on preliminary data from the AF clinical study. The developed nonlinear signal separation scheme utilizes a locally projective noise reduction (LPNR) algorithm and is described in detail in [3, 16, 17]. Figures 10 and 11 show the respiratory and cardiac signal separation scheme applied on a clinical BCG signal.

Figure 10 Respiratory components extracted with LPNR and FIR filters from the BCG signal, together with reference signals. From top to bottom: reference signal abdominal effort, BCG component extracted with LPNR and FIR filter and original BCG [16].
Figure 10

Respiratory components extracted with LPNR and FIR filters from the BCG signal, together with reference signals. From top to bottom: reference signal abdominal effort, BCG component extracted with LPNR and FIR filter and original BCG [16].

Figure 11 Cardiac components extracted with LPNR and FIR filters from the BCG signal, together with reference signals. From top to bottom: reference signal ECG, BCG component extracted with LPNR and with FIR filter and original BCG [16].
Figure 11

Cardiac components extracted with LPNR and FIR filters from the BCG signal, together with reference signals. From top to bottom: reference signal ECG, BCG component extracted with LPNR and with FIR filter and original BCG [16].

The first Biomon-HF bed sensor measurements from the atrial fibrillation (AF) study and algorithms for automatic AF detection have been reported in [2], and the feasibility of the automatic detection of AF from cardiac vibration signals (ballistocardiograms/BCGs) recorded by unobtrusive bed-mounted sensors has been demonstrated. The proposed system is intended as a screening and monitoring tool in home-healthcare applications and not as a replacement for ECG-based methods used in clinical environments. Based on the BCG data recorded in the study, seven popular machine learning algorithms have been evaluated in [2]. Figure 12 shows examples of cardiac components extracted with LPNR from the recorded bed sensor signals in the AF study for three signal epoch classes (normal, AF, and artifact). The vertical-dashed lines indicate the occurrence of R peaks in the simultaneously recorded ECG.

Figure 12 Examples of cardiac components extracted with LPNR from the recorded bed sensor signal for three signal epoch classes. Vertical-dashed lines indicate the occurrence of R peaks in the simultaneously recorded ECG [2].
Figure 12

Examples of cardiac components extracted with LPNR from the recorded bed sensor signal for three signal epoch classes. Vertical-dashed lines indicate the occurrence of R peaks in the simultaneously recorded ECG [2].

The first BCG data analysis results from the sleep apnea study look very promising as well as can be seen in Figure 13.

Figure 13 Example BCG bed sensor measurements from the sleep apnea study. From top to bottom: PSG references (flow, abdominal effort, thoracic effort, SpO2), BCG bed sensor. Gray shaded epochs show apnea.
Figure 13

Example BCG bed sensor measurements from the sleep apnea study. From top to bottom: PSG references (flow, abdominal effort, thoracic effort, SpO2), BCG bed sensor. Gray shaded epochs show apnea.

Individualized blood pressure monitoring (WP 3)

For evaluation of the individualized blood pressure monitoring approach, a special designed hardware and software was developed, allowing for recordings in the tilt table feasibility study.

The hardware was equipped with an oscillometric blood pressure unit, allowing for both, the determination of blood pressure and the recording of pulse waves. Additionally a one-channel ECG was included in the unit. Peak detection allowed for the calculation of transit time, by determination from signal conduction at the heart to the arrival of signal at the upper arm cuff. All calculations were performed in real time with an in-build digital signal processor. The determination of blood pressure was processed by validated algorithms as used by the I.E.M. Mobil-O-Graph blood pressure device.

With respect to data management, the hardware was controlled via PC software. The software initiated measurements of transit time, pulse waveforms, and blood pressure via predefined intervals or manual trigger. All data were stored in raw data format, allowing for retrospective analysis of ADC values, transit times, blood pressure values, and information on date and time.

The feasibility study has successfully been conducted at the University Hospital Aachen, and the preliminary data review allows for an optimistic view and for the further pursuance of the approach over and beyond the Biomon-HF project.

In-ear photoplethysmography sensor (WP 4)

The evaluation of the in-ear photoplethysmography (PPG) sensor in the sleep apnea study revealed good correlation of the in-ear sensor signal with the reference ECG and PSG. An example measurement is shown in Figure 14, and Figure 15 presents the scatter plot for the heart rate derived from in-ear PPG sensor and reference ECG from polysomnography (PSG), respectively.

Figure 14 Comparison of in-ear PPG sensor with reference ECG from PSG.
Figure 14

Comparison of in-ear PPG sensor with reference ECG from PSG.

Figure 15 Scatter plot for heart rate (HR) derived from in-ear PPG sensor (LAVIMO) and reference ECG.
Figure 15

Scatter plot for heart rate (HR) derived from in-ear PPG sensor (LAVIMO) and reference ECG.

Finally, the in-ear photoplethysmography (PPG) sensor-derived arterial oxygen saturation (SpO2) in comparison with the reference SpO2from the PSG finger PPG is shown in Figure 16 (bottom). Both show good correlation. The reference respiratory flow measured by the PSG is plotted in Figure 16 (top). The amplitude modulation in the respiratory flow indicates sleep apnea episodes, which can be seen in the SpO2signals as well.

Figure 16 Example in-ear SpO2 measurement (bottom) from sleep laboratory study with PSG reference flow (top) and SpO2 (bottom).
Figure 16

Example in-ear SpO2 measurement (bottom) from sleep laboratory study with PSG reference flow (top) and SpO2 (bottom).

Conclusion

In the Biomon-HF project, innovative sensors and algorithms for measuring vital signs, i.e., during the nocturnal sleep period have been developed and successfully tested in five clinical feasibility studies involving 115 patients.

The investigated sensor concepts and algorithms are an important step toward future patient-customized telemonitoring and sensor-guided therapy management in chronic heart failure, including early detection of upcoming HF exacerbation and comorbidities, i.e., cardiac arrhythmia and sleep apnea, at home.

The resulting preventable disease complications and emergencies and reduction of consequences of disease are very important advantages for the patients, causing relief for medical staff and, thus, offer an enormous potential for improvements and cost savings in healthcare systems.


Corresponding author: Thomas Vollmer, Philips Research, Personal Health Solutions Group, Aachen, Germany, E-mail:

Acknowledgments

The Biomon-HF project is part of the in.nrw innovating medical technology cluster Aachen, Germany, and is performed in collaboration between the University Hospital Aachen Department of Internal Medicine I (Cardiology, Pneumology, and Angiology) and the Department of Neurology, the Central Institute ZEA-2 Electronic Systems of the Research Center Jülich, the Philips Chair for Medical Information Technology at the Helmholtz-Institute for Biomedical Engineering of the RWTH Aachen, and the industrial partners Philips Research, I.E.M. and Takeda Pharma. The project is co-funded by the European Union (ERDF – European Regional Development Fund – Investing in your future) and the German federal state North Rhine-Westphalia (NRW), under the operational program “Regional Competitiveness and Employment” 2007–2013 (EFRE).

References

[1] Blanik N, Santos S, Venema B, Leonhardt S. Long-term vital parameter monitoring. BEMBS 2011, Boston.Search in Google Scholar

[2] Brüser C, Diesel J, Zink MDH, Winter S, Schauerte P, Leonhardt S. Automatic detection of atrial fibrillation in cardiac vibration signals. IEEE J Biomed Health Inform 2013; 17: 2013, 162–171.10.1109/TITB.2012.2225067Search in Google Scholar PubMed

[3] Schiek M (2011): Phase based methods for data fusion; Jahrestagung DGBMT 2011.Search in Google Scholar

[4] Schiek M. Biosignals meet Apps – future application scenarios and current obstacles; Biomedical Engineering/Biomedizinische Technik; Jahrestagung der Deutschen Gesellschaft für Biomedizinische Technik 2012, DGBMT, Jena, Germany, 09/16/2012–09/19/2012 online/print (2012).Search in Google Scholar

[5] Schiek M, Schanze: Signale und Artefakte; Workshop “Innovative Verarbeitung bioelektrischer und biomagnetischer Signale” – bbs2012, bbs2012, Berlin, Germany, 04/19/2012–04/20/2012 (2012).Search in Google Scholar

[6] Ulbrich M, Weyer S, Mühlsteff J, Walter M, Leonhardt S. Evaluation of a swine model for impedance cardiography – a feasibility study. In: International Workshop on Impedance Spectroscopy (IWIS), 26.-28.09.12, Chemnitz; 2012.Search in Google Scholar

[7] Venema B. Long-term monitoring of cardiac and respiratory activity using optoelectronic sensor. In: 15th International Student Conference, Prague, Czech Republic, 2011.Search in Google Scholar

[8] Venema B, Blanik N, Blazek V, Gehring H, Opp A, Leonhardt S. Advances in reflective oxygen saturation monitoring with a novel in-ear sensor system: results of a human hypoxia study. IEEE Trans Biomed Eng 2012; 59: 2003–2010. doi: 10.1109/TBME.2012.2196276.10.1109/TBME.2012.2196276Search in Google Scholar PubMed

[9] Venema B, Blazek V, Blanik N, Schiefer J, Leonhardt S. A feasibility study evaluating innovative in-ear pulse oximetry for unobtrusive cardiovascular homecare monitoring during sleep. In: IEEE EMBS Special Topic Conference on Point-of-Care Healthcare Technologies, Bangalore 2013, India.10.1109/PHT.2013.6461300Search in Google Scholar

[10] Venema B, Santos S, Leonhardt S. In-ear arterial oxygen saturation monitoring with a single optoelectronic sensor for long-term conditions. In: IFMBE World Congress 2012, Peking, China.10.1007/978-3-642-29305-4_356Search in Google Scholar

[11] Vollmer T, Schauerte P, Zink M, et al. Individualized biomonitoring in heart failure – Biomon-HF “Keep an eye on heart failure – especially at night”. Biomed Tech 2014; 59: 103–111.10.1515/bmt-2013-0024Search in Google Scholar PubMed

[12] Weyer S, Röthlingshöfer L, Walter M, Leonhardt S, Bensberg R. Evaluation of bioimpedance spectroscopy for the monitoring of the fluid status in an animal model. In: 2012 Ninth International Conference on Wearable and Implantable Body Sensor Networks (BSN 2012).10.1109/BSN.2012.25Search in Google Scholar

[13] Weyer S, Röthlingshöfer L, Walter M, Leonhardt S. Evaluation of bioelectrical impedance spectroscopy for the assessment of extracellular body water. Acta Polytechnica 2012; 52: 120–124.10.14311/1660Search in Google Scholar

[14] Weyer S, Röthlingshöfer L, Walter M, Leonhardt S. Evaluation of bioelectrical impedance spectroscopy for the assessment of extracellular body water. In: POSTER 2012, 16th International Student Conference on Electrical Engineering.10.14311/1660Search in Google Scholar

[15] Weyer S, Ulbrich M, Leonhardt S. A model-based approach for analysis of intracellular resistance variations due to body posture on bioimpedance measurements. J Phys Conf Ser 2013; 434: 012003.10.1088/1742-6596/434/1/012003Search in Google Scholar

[16] Yao Y, Brüser C, Piertzyk U, Leonhardt S, v. Waasen S, Schiek M. Model-based verification of a non-linear separation scheme for ballistocardiography. IEEE Trans Biomed Health Inform 2013; 18: 174–182.10.1109/JBHI.2013.2261820Search in Google Scholar PubMed

[17] Yao Y, Brüser C, Vollmer T, Schiek M. Signal separation for ballistocardiography via locally projective noise reduction. In: World Congress on Medical Physics and Biomedical Engineering 2012.10.1007/978-3-642-29305-4_132Search in Google Scholar

[18] Yao Y, Dammers J, Schiek M. (2012): Identifikation der Stimulus bezogenen intrinsischen Moden von MEG-Signalen; DGBMT-Workshop “Innovative Verarbeitung bioelektrischer und biomagnetischer Signale” April 2012.Search in Google Scholar

[19] Yao Y, Zink M, Schauerte P, Schiek M. Analyse von Langzeitregistrierungen invasiv erhobener Bioimpedanzdaten. In: Workshop “Innovative Verarbeitung bioelektrischer und biomagnetischer Signale” – bbs2012, bbs2012, Berlin, Germany, 04/19/2012–04/20/2012 (2012).Search in Google Scholar

[20] Zink M, Brüser C, Venema B, et al. Heart beat analysis with reflective in-ear pulse photoplethysmography sensor and ballistocardiographic foil. In: 79th Annual Meeting of the German Cardiac Society (DKG), Mannheim, 2013.Search in Google Scholar

Received: 2013-6-14
Accepted: 2013-11-21
Published Online: 2014-2-18
Published in Print: 2014-4-1

©2014 by Walter de Gruyter Berlin/Boston

Downloaded on 28.3.2024 from https://www.degruyter.com/document/doi/10.1515/bmt-2013-0024/html
Scroll to top button