See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/9061890

The LifeShirt An Advanced System for Ambulatory Measurement of Respiratory

and Cardiac Function

Article  in  Behavior Modification · November 2003

DOI: 10.1177/0145445503256321 · Source: PubMed

CITATIONS READS
150 840

3 authors:

Frank H. Wilhelm Walton Roth

University of Salzburg Stanford University
279 PUBLICATIONS   11,506 CITATIONS    273 PUBLICATIONS   15,362 CITATIONS   

SEE PROFILE SEE PROFILE

Marvin Sackner
Preident, Sackner Wellness Products
277 PUBLICATIONS   7,087 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Theories of Panic Attack View project

Nature and Treatment of Panic Disorder View project

All content following this page was uploaded by Frank H. Wilhelm on 28 May 2014.

The user has requested enhancement of the downloaded file.

 The LifeShirt: an advanced system for ambulatory measurement of
respiratory and cardiac function

1
Frank H. Wilhelm, Ph.D., 1Walton T. Roth, M.D.,
& 2Marvin A. Sackner, M.D.

From: 1Department of Psychiatry and Behavioral Sciences, Stanford University, Stanford, CA,
and the Department of Veterans Affairs Health Care System, Palo Alto, CA, USA, and
2
University of Miami at Mt. Sinai Medical Center, Miami Beach, FL.

Address correspondence to: Frank H. Wilhelm, Ph.D., Stanford University / VAPAHCS 116F-
PAD, 3801 Miranda Ave., Palo Alto, CA 94304; 650-493-5000 ext. 65242; fax: 650-493-4901;
fwilhelm@stanford.edu.

Supported by the Department of Veterans Affairs, and NIH Grant MH56094.

VMLA-038-01 1
Abstract

An accurate ambulatory monitor of breathing is needed to observe acute respiratory changes in

patients with medical or psychological disorders outside the clinic (for example, hyperventilation
during panic or apneas during sleep). Significant limitations of existing monitors were their size,
troublesome operation, and difficulty in holding chest and abdomen bands in place during 24-hr
recordings. Recently, a garment with embedded inductive plethysmography sensors for continu-
ous ambulatory monitoring of respiration, ECG, inductive cardiography, motility, postural
changes and other functions has been developed. The signals are displayed and stored on a hand-
held computer (Visor), and then analyzed offline, extracting over 40 clinical parameters relating
to cardiorespiratory function (for example, heart rate, RSA, tidal volume, stroke volume, pre-
ejection period, apnea-hypopnea index, thoraco-abdominal coordination, sighing). In addition,
the device serves as an electronic diary of symptoms, moods, and cognitions. This advanced
system may open a new era in ambulatory monitoring for both clinical practice and scientific re-
search.

Words: 152/150
Keywords: Respiration; Tidal volume; Respiratory sinus arrhythmia; Sleep apnea; Anxi-
ety disorders; Ambulatory monitoring; Review

VMLA-038-01 2
I. Background

Breathing is an important physiological function that has been neglected in clinical research and
practice, in large part because of the difficulties obtaining accurate measurements outside of a
respiratory care hospital unit. In contrast, ECG Holter monitoring has been applied widely for
many years and it has profoundly changed our understanding of heart disease and improved pa-
tient care. The scarcity of respiratory monitoring is unsettling since any disturbance of respira-
tory function is potentially life threatening and carries great weight in diagnostic decisions.
Long-term monitoring of respiratory function outside the hospital would be desirable for a vari-
ety of disorders that are known to be associated with respiratory abnormalities, e.g., chronic ob-
structive pulmonary disease, pulmonary emphysema, restrictive lung disease, asthma, cardiac
failure, or epilepsy.

Furthermore, since emotions can affect respiration profoundly, our understanding of a spectrum
of mental and psychophysiological disorders associated with emotional disturbance would bene-
fit from a detailed monitoring of respiratory function. Dysregulated breathing plays a role in
common clinical syndromes such as functional cardiac disorder and chronic pain (Wilhelm,
Gevirtz, & Roth, 2001), and among mental disorders, especially panic disorder has been linked
to faulty respiratory regulation. This disorder is characterized by episodes of sudden intense
anxiety accompanied by a variety of debilitating physical symptoms, most prominently shortness
of breath and palpitations. Our recent laboratory results have indicated profound instability in
respiratory regulation in patients even when they do not experience panic attacks (Roth, Wil-
helm, & Trabert, 1998; Wilhelm, Gerlach, & Roth, 2001; Wilhelm, Trabert, & Roth, 2001a;
Wilhelm, Trabert, & Roth, 2001b). One study that examined breathing patterns in panic patients
during sleep found an elevated frequency of micro-apneas (Stein, Millar, Larsen, & Kryger,
1995). Respiratory abnormalities are also found in patients with specific phobias during acute
anxiety (Alpers, Wilhelm, & Roth, 2000; Alpers, Wilhelm, & Roth, submitted; Wilhelm & Roth,
1998a).

For the examination of respiratory dysregulation in anxious patients we have used a measure-
ment technique called respiratory inductive plethysmography (IP), first implemented in the Res-
pitrace™ system (Sackner, Nixon, Davis, Atkins, & Sackner, 1980). IP is the gold standard for
unobtrusive respiratory monitoring and has been used widely in clinical and research settings.
Approximately 1,600 published scientific studies have used this technology and established it as

VMLA-038-01 3
the standard for non-invasive assessment of the pattern of breathing. This technique approxi-
mates the amount of air moved by the respiratory system by measuring the expansion and con-
traction of both the rib cage (RC) and abdominal (AB) compartments (Konno & Mead, 1967).
The RC motion reflects activity mostly of the intercostal muscles and to a lesser extent the acces-
sory muscles while AB motion mostly reflects activity of the diaphragm. An IP sensor consists
of a sinusoidal arrangement of electrical wires embedded in elastic cotton bands (Respiband™).
A high frequency, low voltage oscillating current is passed through the wires to generate a mag-
netic field needed to measure the self-inductance of the coils, which is proportional to the cross-
sectional area surrounded by the band (Watson, Poole, & Sackner, 1988). No electrical currents
are passed through the body.

Calibration of the RC and AB bands by spirometry or a fixed volume bag and summing of the
two signals allows measurement of tidal volume in ml. A number of studies have shown that the
accuracy of tidal volume estimates from IP recording is sufficient for clinical decision making in
babies, children, and adults under resting, sleeping, and exercising conditions, and even in pa-
tients with chronic obstructive pulmonary disease (e.g., Adams, Zabaleta, Stroh, & Sackner,
1993; Tobin et al., 1983). In resting healthy adults, 89% of the breath-by-breath tidal volumes
measured by this method have been shown to be within ±10% of simultaneous spirometric
measurements, and 100% within ±20% (Tobin, Guenther, Perez, & Mador, 1987). No systematic
bias was found, meaning that tidal volume can be estimated with much greater accuracy when
averaged over periods of minutes. One important advantage of this method is its minimal intru-
siveness since measurement methods requiring a face mask or mouthpiece are known to alter
natural breathing (Askanazi et al., 1980). In addition, it allows long-term respiratory monitoring
inside and outside the laboratory.

In a series of studies with anxious patients we have used a small, portable (4.5 x 2.5 x 1.5 in) IP
device (Respitrace™, Ambulatory Monitoring, Inc., Ardsley, NY) connected to a digital recorder
(Vitaport, Becker Meditec, Karlsruhe, Germany) outside the laboratory (Alpers et al., 2000; Wil-
helm, Alpers, Meuret, & Roth, 2001; Wilhelm & Roth, 1996; Wilhelm & Roth, 1997; Wilhelm
& Roth, 1998a; Wilhelm & Roth, 1998c). The application of this IP technology in ambulatory
patients is more problematic than recording in supervised, resting patients in the laboratory. The
bands react not only to respiratory activity, but also to postural changes, bending of the trunk, or
vibrations of abdomen, chest, or breasts due to accelerative forces on the body. Walking pro-
duces rhythmical fluctuations on top of the breathing signal (see Wilhelm & Roth, 1996).

VMLA-038-01 4
In addition to these artifact problems, one general dilemma with the analysis of respiratory data,
either visually or by computer algorithm, is that there is no definite range of acceptable values of
respiratory activity. There are a number of reasons for this: First, respiratory activity is multi-
dimensional in nature in that derived parameters like depth, rate, and pauses can vary independ-
ently from each other. The respiratory waveform can therefore assume a wide variety of pat-
terns. Second, naturally occurring extreme values associated with sighing or breathing suspen-
sion cannot be excluded from consideration. Third, respiratory activity can vary profoundly
from breath to breath. Fourth, speech is accompanied by complicated, variable breathing pat-
terns. And fifth, voluntary control of breathing can produce any variety of waveforms. All these
factors make detection of artifacts in respiratory recordings more difficult than in other physio-
logical channels.

Previous attempts to measure respiration ambulatorily (Anderson, Coyle, & Haythornwaite,

1992; Martinez et al., 1996) used a monitoring device that reduces respiratory data on-line to av-
erages of respiratory rate and tidal volume over several minutes. Although this approach has
yielded important insights into health effects of respiratory patterns, we have pointed out some
limitations of this methodology (Wilhelm & Roth, 1998b). We argued that a complete disclosure
system for respiration is essential since respiratory patterns are too complex to be analyzed on-
line. Probably off-line analysis will always be necessary for research applications, since only
such analysis can deal flexibly with artifacts. Furthermore, irregularity in breathing is concealed
by on-line reduction to means or may be falsely rejected as artifact.

In our recent ambulatory studies we recorded raw waveforms of the RC and AB signal over ex-
tended periods and analyzed the data carefully off-line. For this purpose, we developed a com-
puter program (Wilhelm, Grossman, & Roth, 1999; Wilhelm & Roth, 1993) that automatically
detects important features of respiratory waveforms, for example, onset and offset of inspirations
and expirations, and extracts standard parameters such as respiratory rate, tidal volume, inspira-
tory fractional time, and mean inspiratory flow. The waveforms can be displayed together with
data from the activity and posture sensors to determine intervals that are likely to be confounded
by movement artifacts. Visual inspection then decides if manual editing is necessary or if inter-
vals need to be discarded. This detailed supervision of the automatic analysis enhances our con-
fidence into the quality of extracted parameters, but it is certainly quite time consuming since the
data sets recorded are large. More complex computer programs could alleviate this load.

VMLA-038-01 5
The method of volume calibration for the bands most commonly used is rebreathing into a bag of
known volume or into a spirometer. To differentiate the relative contribution of RC and AB to
the independently measured volumes, patients are instructed to breathe predominantly by moving
the rib cage (for a series of 8-10 breaths), and then by moving the abdomen (again, for 8-10
breaths). Least-squares multiple regression analysis (Chadha et al., 1982) then allows assignment
of multiplication factors to the output of RC and AB bands, and the tidal volume curve is then
obtained by adding the two volumes. Unfortunately, some patients have difficulty in separating
rib cage and abdominal breathing movements, so the procedure can take up to 10 minutes. A
simplified procedure has been suggested that assumes a fixed ratio of 2:1 between RC and AB
volume (Banzett, Mahan, Garner, Brughera, & Loring, 1995). This allows calibration of bands
by a few normal breaths, but, of course, tidal volume estimates can be compromised if the ratio
of RC/AB shifts significantly. In any case, coefficients retrieved from the calibrations are only
valid in the posture in which they were performed. Separate calibrations are necessary for
standing, sitting, and supine posture if absolute tidal volumes are to be monitored during an en-
tire day and night, and need to be applied correctly based on a posture signal obtained concur-
rently.

Our experience with the ambulatory IP monitor over the course of making hundreds of long-term
recordings in anxious patients was mixed. Although most recordings were completed satisfacto-
rily by taking great care of all the details involved, a number of procedural and hardware related
problems resulted in a significant loss of data (about 5-8%). The most common problem was
slippage of bands, even though they were thoroughly taped to either the skin or undergarments.
Another common problem was breaking of cables or disconnection of plugs. This is not surpris-
ing, since the technology had been developed more than 20 years ago and has improved little
since. Some data was lost because of invalid calibrations. And of course, patients experienced
some discomfort wearing the two bands, multiple wires, and tapes on their skin. In summary, the
recording and analysis of respiratory IP data with this old standard system is complicated and
error prone. It can be made to work in research studies, but the applicability for clinical practice
is quite low. An improved system is urgently needed if respiratory monitoring is to be applied
more widely, including in clinical settings by practitioners.

VMLA-038-01 6
II. The LifeShirt System

To serve a need for a clinically usable multi-channel monitoring device that includes respiratory
function, VivoMetrics, Inc. (Ventura, California) recently developed the LifeShirt™, a system
with improved inductive plethysmography (IP) measurement technology at its core. The sensor
array of the LifeShirt system is embedded in a sleeveless undergarment made of hand washable,
reusable highly stretchable material that fits snugly and can be worn comfortably for extended
periods. Several sizes are available.

Sensors
IP sensors for monitoring a variety of cardiopulmonary signals are embedded in the shirt. This
ensures their correct and durable placement and makes it easier to set up the sensor array for
multi-channel recording. The sensors, like their predecessors, the Respibands™, consist of a si-
nusoidal arrangement of electrical wires that are excited through an extremely low current, elec-
trical oscillator circuit. The coils have a width of only about 1 inch, which is much less than the
coils embedded in Respibands™. One sensor is sewn into the shirt at the level of the rib cage
(RC) and one at the level of the abdomen (AB). In addition, an important innovation, a single IP
sensor is placed transversely at the level just below or at the xiphoid process and records chest
wall movement resulting from heart activity ("thoracocardiography"). After suppression of respi-
ratory movement with digital filtering and ECG triggered ensemble averaging, this signal reflects
the ventricular volume curve with similar appearance to ventricular volume curves obtained with
automated border detection echocardiography and nuclear angiography (Bloch, Jugoon, & Sack-
ner, 1994; Bloch, Jugoon, & Sackner, 1999; Sackner, Hoffman, Stroh, & Krieger, 1991). As an
optional addition to the shirt, an IP sensor band can be attached around the neck to measure
changes in its cross-sectional area. High pass digital filtering of the neck IP trace suppresses res-
piratory efforts and neck movements and allows display of the higher rate carotid arterial pulses
(Jordan et al., 1984), but only in the upright posture. In the supine or semirecumbent posture, be-
cause of vascular loading, the signal depicts the jugular venous pulse. This postural dependence
of arterial and venous pulse allows non-invasive measurement of central venous pressure using
tilt (Bloch, Krieger, & Sackner, 1991).

VMLA-038-01 7
An electrocardiogram is recorded by means of three electrodes placed directly onto the skin on
the upper chest and on the lateral surface of the abdomen. This standard configuration provides a
single lead for heart rate and ECG waveform determinations. For measurement of body posture
(angle deviation from horizontal) a two-axis accelerometer is placed onto the shirt over the ster-
num. It also serves as an activity monitor, e.g., during walking or running each stride produces a
spike of acceleration that permits focusing in on portions of the tracing in which motion artifacts
might be present. Other peripheral diagnostic devices with digital output may be used as compo-
nents of the LifeShirt system for special purposes. These include pulse oximeter, ambulatory
blood pressure recorder, capnograph, oral thermometer, or digital weighing scale.

Signal Recording and Online Data Reduction

All sensors are attached via secure connectors and a cable tree to a small custom designed inter-
face directly plugged into the port of a mass-produced handheld personal computer (Handspring
Visor™). This compact and lightweight computer serves as a digital recorder and can be worn
on a belt or put into the pocket of a jacket. The screen of this device allows patient input of
symptoms, activities and mood synchronized to the datastream of the physiological signals. An
exchangeable flash memory card provides large storage capacity. The handheld personal com-
puter captures raw waveforms from the LifeShirt for full disclosure storage on a flash memory
card.

ECG R-waves are detected as a pulse with 5 msec resolution and the time of their occurrence is
stored. In later versions of the software, 1 msec resolution will be available. This higher resolu-
tion is preferred for optimal determination of respiratory sinus arrhythmia, but the lower resolu-
tion will suffice for most clinical purposes. The raw ECG waveform is also stored continuously
at 200 points/sec together with the other signals.

Offline Parameter Extraction

The raw waveform signals stored on the flash memory card are processed using VivoLogic™
software that runs on PCs. Currently it is planned to provide expert data processing service at a
central Data Center. For this, the data is uploaded to a personal computer and sent to the Data
Center via the Internet, or the memory card is sent by mail. There the data are subjected to auto-
matic analyses that includes calibration, artifact screening, quality of data control, parameter ex-
traction, making trends and generating reports. Technicians and physicians support the operation

VMLA-038-01 8
of the Data Center. They review the quality of extracted parameters, edit artifactual segments
that cannot be scored automatically, and mark data segments of special clinical interest. This ex-
tracted data is then distributed to the clinician or researcher requesting the service as representa-
tive waveforms, data spreadsheets, or reports. Visual inspection of all raw waveforms is not at-
tempted since this would be unrealistic with the great amounts of data being processed. Instead,
1-min trends of derived parameters are presented as medians with their quartile ranges. Such
analysis minimizes the influence of outliers that might spuriously affect mean values.

Two calibration routines are available for the respiratory inductive plethysmographic sensors. An
automatic calibration software module is activated during a 5-min quiet period of breathing. It
uses the so called "qualitative diagnostic calibration" (QDC) procedure (Sackner et al., 1989) to
accurately set the relative electrical gains of the RC and AB sensors such that obstructive apnea
can be detected as a flat or nearly flat RC+AB trace associated with 180° phase-shift between
RC and AB signals. This calibration method is preferred during sleeping periods. The RC+AB
trace can be converted to absolute tidal volume in ml by rebreathing a few times into a bag with
a fixed, known volume at any time during the recording period. The QDC calibration is based on
the principles of the isovolume maneuver calibration and does not require separate calibrations
for upright and recumbant position of the torso (Adams et al., 1993; Sackner et al., 1989). How-
ever, for upright ambulatory monitoring, the two posture, least squares regression procedure us-
ing the sitting and standing postures with rebreathing into a fixed bag volume for about 20 sec-
onds is preferred for more accurate ambulatory tidal volume estimations (Chadha et al., 1982).

A large number of cardiopulmonary parameters of clinical interest are extracted from the rela-
tively small array of noninvasive sensors. The lung volume signal is processed using a feature
detection algorithm that identifies the beginning and end of inspiration and expiration and ex-
tracts tidal volume, after eliminating periods with movement artifacts. The derivative of this sig-
nal is used to detect peak inspiratory and expiratory flow. Table 1 lists the key respiratory pa-
rameters based on these or similar algorithms and their clinical significance. Although the sig-
nificance of these parameters relates primarily to respiratory or cardiac disorders, we expect that
patients with panic disorder will show abnormalities in many of them such as sighs, apneas, and
irregular respiratory rates and tidal volumes consistent with current theories of respiratory dys-
regulation in this clinical syndrome (e.g., Klein, 1992). Further, anxiety and panic is common in
patients with chronic cardiorespiratory disease (Wise & Taylor, 1990).

VMLA-038-01 9
For thoracocardiography, the chest wall movement signal is processed to filter out respiratory
contributions. Under resting conditions, respiratory movement dominates the waveform with
only approximately 3-5% of its content consisting of oscillations synchronous with the heartbeat.
Thoracocardiographic oscillations are extracted from the waveform by combining a band pass
digital filter to suppress respiratory content and an ECG triggered ensemble average. Computed
parameters include the amplitude of the averaged ventricular volume curve that is a measure of
stroke volume. Cardiac output is obtained by multiplying stroke volume by heart rate. These
hemodynamic parameters cannot be independently calibrated to ml but can provide accurate per-
centage change assessment equivalent to thermodilution measurement of cardiac output (Bloch,
Baumann, Stocker, & Russi, 1997; Sackner et al., 1991). Shape changes of ventricular wall mo-
tion, i.e., outward motion during ventricular systole signify myocardial ischemia. The thoraco-
cardiogram cannot provide continuous recording of left ventricular ejection fraction because the
absolute values of end-systolic and end-diastolic volume cannot be obtained as in nuclear angi-
ograms or echocardiograms. However, a surrogate of this measure, the volume ejected during the
first third of systole divided by the stroke volume (Johnson, Ellis, Schmidt, Weiss, & Cannon,
1975) can be computed as a dimensionless number from the thoracocardiogram. Preliminary ex-
periments in normal subjects indicate that this parameter rises in response to exercise like the
classic parameter of ejection fraction. If it parallels the fall of ejection fraction in coronary artery
patients subjected to mental stress (Goldberg et al., 1996), the LifeShirt system would offer a
tremendous tool for following the ischemic consequences of mental stress in the workplace and
home.

The ECG RR interval information is converted to instantaneous heart rate. Respiratory sinus ar-
rhythmia (RSA) is measured using the peak-to-trough method (Grossman, van Beek, & Wient-
jes, 1990): for each breathing cycle, the shortest RR interval during inspiration is subtracted from
the longest RR interval during expiration. RSA reflects parasympathetic neural control of the
heart. Breath-by-breath RSA values can be normalized by tidal volume and respiratory rate, both
important confounds in the estimation of changes in parasympathetic activity from RSA
(Grossman, Karemaker, & Wieling, 1991; Saul et al., 1991), which is currently neglected in most
clinical studies. The single-lead ECG waveform also allows a rough quantification of cardiac
arrhythmias. However, since the P-wave can be invisible in a single lead, a 12-lead ECG is nec-
essary for accurate classification of arrhythmias.

The rectified and integrated accelerometer signal is used to detect periods of physical activity
and rest. The posture signal is used to adjust RC and AB gains when posture changes from hori-

VMLA-038-01 10
zontal to the upright and vice-versa, and to interpret the vascular pulses derived from the neck
inductive plethysmograph as to the level of central venous pressure. The carotid arterial pulse,
together with the ECG R-wave time, allows computation of systolic time intervals, for example,
the pre-ejection period (PEP, the time from the R-wave to initiation of the upstroke of the carotid
arterial pulse)(Jordan et al., 1984), as well as pulse wave transit time, a measure negatively cor-
related with diastolic blood pressure (Steptoe, Smulyan, & Gribbin, 1976), especially over short
periods of time. Episodic reduction of pulse transit time signifying a transient rise of blood pres-
sure is consistent with microarousals from sleep (Pitson & Stradling, 1998).

The LifeShirt system does not measure EEG, EOG, and EMG, which would be necessary meas-
urements for standard sleep staging (Rechtschaffen & Kales, 1968). However, breathing pattern
analysis allows a rough discrimination among wake state, rapid eye movement (REM) sleep, and
non-REM sleep in most individuals. High values of tidal volume, respiratory rate, and heart rate
along with motion artifacts on the breath waveform traces and high variability of end-expiratory
lung volume are consistent with waking while lower values with sleeping, which usually allows
distinguishing approximate sleep onset. Parameters that are typical of REM vs. NREM sleep in-
clude higher variability of tidal volume, respiratory rate, and f/Vt. Especially good markers of
REM vs. NREM sleep are lower values of %RC/Vt (about 50% reduction from NREM) because
of the accompanying partial thoracic respiratory paralysis (Neilly, Gaipa, Maislin, & Pack, 1991)
and elevated levels of thoracoabdominal discoordination as expressed by high phase angles be-
tween RC and AB compartments.

III. Applications and Limitations

The LifeShirt is a sophisticated system for measuring a variety of clinically important respiratory
parameters. It is much improved compared to any previously available ambulatory respiratory
monitoring systems in terms of ease of use and patient comfort. In addition, it has the capability
of recording a variety of parameters related to cardiac function and can obtain posture and motil-
ity information important as control parameters for unsupervised monitoring of patients. All the
sensors and electronics are state-of-the-art technology and are integrated within a shirt and a
small handheld computer device. The system is a logical extension of ECG Holter monitors and
may open a new era in ambulatory monitoring for both clinical practice and scientific research.

One limitation of the system for research applications is that in its current implementation the
raw data is processed in a central Data Center, which, while using quite sophisticated software,

VMLA-038-01 11
limits the possibilities for detailed inspection of the data by the researcher. However, one could
argue that with the overwhelming amount of data this kind of simplification is necessary and de-
sirable. It remains to be seen if the automatic artifact screening algorithms will be efficient and
tidal volume will be measured accurately with posture changes. Furthermore, once the device is
in widespread use, the Data Center will have to prove that it can keep up with incoming data
streams.

Another limitation of the technology at the current stage is that although most of the extracted
parameters have been validated within the laboratory, the usefulness of several of the parameters
in ambulatory settings has not been demonstrated. Certain parameters may only be reliably ob-
tainable during a standardized sitting or supine resting period. In any case, such standardized as-
sessment periods should be part of the 24-hr monitoring, since they can minimize confounds
from physical activity and movement artifacts and help detect any physical dysfunction that may
be buried in a noisy signal. Another way the system reduces false judgement is by providing a
wide variety of loosely correlated parameters from respiratory and cardiac systems. A clinical
expert can utilize this redundancy to better ascertain a specific diagnostic decision.

The LifeShirt system can be used for ambulatory monitoring of physiologic parameters during
wake, sleep, and activity states. In addition, reports of mood and symptoms can be gathered
automatically and reliably, overcoming the limitations of retrospective reports or paper and pen-
cil diaries. Intervals when patients reported specific symptoms can receive special attention in
the inspection of physiological parameters. The system will undoubtedly be used with great en-
thusiasm in several areas of clinical research. In addition, it is user-friendly enough to be used in
clinical practice, and even by healthy individuals interested in having feedback about their bodily
functions, for example, during mental or sports performance. One apparent area for application
of this new technology is the monitoring of breathing patterns over extended periods in respira-
tory disorders such as chronic obstructive pulmonary disease, pulmonary emphysema, restrictive
lung disease, or asthma. In clinical studies, this can help elucidate their phenomenology and
physiological mechanisms, and monitor the effects of pharmacological treatment. In clinical
practice, this can enhance patient care.

Another important application is the determination of the best dosages of drugs balancing effi-
cacy and potentially dangerous side effects. For example, drugs with respiration suppressing side
effects such as opiates and sedatives are commonly prescribed for chronic cancer related pain.
Often these drugs are underdosed to avoid any risk of respiratory arrest. By monitoring respira-

VMLA-038-01 12
tory parameters while titrating the dosage, patient safety can be assured even at higher doses.
Cardiologists will also benefit from the broad cardiorespiratory capabilities of the system. With
the addition of a pulse oximetry option, the LifeShirt system becomes a reasonable sleep study
recorder and can provide analysis of breathing patterns as an aid in classifying apneas. However,
it cannot replace clinical polysomnography since current standards for sleep onset detection and
sleep staging are not met, which would require the addition of an EEG, EOG, and neck EMG
lead, and an accelerometer attached to the leg.

The LifeShirt technology vastly expands the scope of self-report and physiological measurement
and will likely contribute to a refined assessment of anxiety disorders and other psychiatric syn-
dromes. We already mentioned the recent evidence from our and other laboratories indicating a
prominent role of self-reported and physiologically measured breathing irregularities in panic
disorder. This user-friendly technology may help overcome the limitations of current diagnostic
standards that solely rely on potentially biased retrospective reports of patients with anxiety, en-
tirely neglecting physiological measurement, when many of the defining symptoms are likely of
physiological origin. We have argued that this situation is unsatisfying, since most other medical
diagnoses (e.g., diabetes) rely on both symptom self-report and systematic biomedical measure-
ments (e.g., the glucose tolerance test) to confirm a diagnosis (Wilhelm & Roth, 2001).

The ambulatory setting has some advantages over the laboratory for assessing the biology of
anxiety. Although the laboratory provides control of stimulation and motor activity, it is not typi-
cal of a person’s ordinary life. The laboratory environment is novel and can induce fear, so any
measurements made in it may be poorly generalizable to more natural settings (this is for exam-
ple important in "white coat" hypertension). The laboratory window of observation is limited in
time and infrequent events like spontaneous panic attacks are easy to miss. Using ambulatory
monitoring technology, physiological reactions related to anxiety or other problems can be re-
corded when and where they happen, of course also within the laboratory. Patients with a more
situationally-bound type of anxiety (for example, specific or social phobia) would best be as-
sessed by first reviewing the computer diary information and identifying situations that were
anxiety provoking, and then by examining the physiological records for these periods. In con-
trast, patients with more generalized anxiety (for example, post-traumatic stress disorder, gener-
alized anxiety disorder) would best be assessed by looking for a generally elevated activation
profile. The LifeShirt also provides information about the patients’ motor and speech behavior,
from which avoidant behavior can be registered. In other words, all three emotional response
systems - language, physiology, and motor behavior - are recorded.

VMLA-038-01 13
Ideally, the physiological activation profile of an individual with anxiety would be compared to
that of a reference control population and various patient populations, and the most reactive or
most abnormal physiological systems could be identified. Such reference norms are analogous
to those developed for standard clinical questionnaires like the State-Trait Anxiety Inventory
(Spielberger, Gorsuch, & Luchene, 1970). The reason these norms for anxiety disorders do not
yet exist is that multi-channel monitoring devices were not widely available or clinically appli-
cable in the past. (Of course, for medical diagnoses such as chronic obstructive pulmonary dis-
ease or cardiac disease, such norms do already exist.) Ideally, physiological testing would meet
psychometric standards comparable to those of currently accepted clinical questionnaire assess-
ment measures. But even if this is not achievable, the testing profile will be an important source
of information to the patient and clinician complementing the information gained from self-
reports of patients.

We envision a time when new patients complaining of anxiety and stress will routinely be
monitored to complement their self-report in determining the severity and nature of their symp-
toms and what treatment is appropriate, and then will be monitored again during treatment to
complement self-report data in evaluating its success. In the future, success of treatments could
also be judged in terms of their ability to reverse abnormal physiological patterns. In addition,
treatments could make therapeutic use of physiological measures obtained during this kind of
assessment (Meuret, Wilhelm, & Roth, 2001). The LifeShirt software could be modified rela-
tively easily to convert it into a portable biofeedback device that could help patients normalize
specific physiological functions.

Physiological recording is particularly relevant as an interface between psychiatry and general

medicine. Anxiety, and panic disorder in particular, has recently emerged as a potentially im-
portant risk factor for coronary heart disease (see Rozanski, Blumenthal, & Kaplan, 1999). Since
6 of the 13 diagnostic symptoms of a panic attack are also cardinal features of coronary heart
disease, it is not surprising that panic disorder is 30-50 times more common in noncardiac chest
pain patients than it is in the overall population (Carter et al., 1994). Because many symptoms of
panic disorder mimic those of coronary heart disease, differentiating these disorders and learning
how they may influence each other is imperative for clinical practice. For example, complaints of
palpitations are common in both disorders, which suggests that monitoring of stroke volume and
other cardiac parameters should be beneficial. In addition, undetected breathing disturbance
during sleep may contribute to the symptomatology of these and a wide variety of other disor-
ders. Many unresolved issues leave our current understanding of the anxiety-health relationship

VMLA-038-01 14
incomplete. The comprehensive cardiorespiratory assessment achievable with the LifeShirt
should help advance this understanding.

Acknowledgment

Preparation of this manuscript was supported by grant NIH/MH56094 and the Department of
Veterans Affairs.

VMLA-038-01 15
References

Adams, J., Zabaleta, I., Stroh, D., & Sackner, M. (1993). Measurement of breath ampli-
tudes: comparison of three noninvasive respiratory monitors to integrated pneumotachograph.
Pediatr Pulmonol, 16(4), 254-8.

Alpers, G. W., Wilhelm, F. H., & Roth, W. T. (2000). Psychophysiological measures can
index change in behavior therapy. Psychophysiology, 37, S8.

Alpers, G. W., Wilhelm, F. H., & Roth, W. T. (submitted). Psychophysiological measures

during exposure in driving phobics. Journal of Abnormal Psychology.

Anderson, D. E., Coyle, K., & Haythornwaite, J. A. (1992). Ambulatory monitoring of

respiration: inhibitory breathing in the natural environment. Psychophysiology, 29, 551-557.

Askanazi, J., Silverberg, P. A., Foster, R. J., Hyman, A. I., Milic-Emili, J., & Kinney, J.
M. (1980). Effects of respiratory apparatus on breathing pattern. J Appl Physiol, 48(4), 577-80.

Banzett, R., Mahan, S., Garner, D., Brughera, A., & Loring, S. (1995). A simple and reli-
able method to calibrate respiratory magnetometers and Respitrace [see comments]. J Appl
Physiol, 79(6), 2169-76.

Bloch, K., Baumann, P., Stocker, R., & Russi, E. (1997). Noninvasive monitoring of car-
diac output in critically ill patients with thoracocardiography. Am J Respir Crit Care Med,
155(1), p26-31.

Bloch, K., Jugoon, S., & Sackner, M. (1994). Thoracocardiographic-derived left ven-
tricular systolic time intervals. Chest, 106(6), 1668-74.

Bloch, K., Jugoon, S., & Sackner, M. (1999). Inductance cardiography (thoracocardio-
graphy): a novel, noninvasive technique for monitoring left ventricular filling. J Crit Care, 14(4),
177-85.

Bloch, K., Krieger, B., & Sackner, M. (1991). Noninvasive measurement of central ve-
nous pressure by neck inductive plethysmography. Chest, 100(2), 371-5.

Carter, C., Maddock, R., Zoglio, M., Lutrin, C., Jella, S., & Amsterdam, E. (1994). Panic
disorder and chest pain: a study of cardiac stress scintigraphy patients. Am J Cardiol, 74(3), 296-
8.

VMLA-038-01 16
 Chadha, T., Watson, H., Birch, S., Jenouri, G., Schneider, A., Cohn, M., & Sackner, M.
(1982). Validation of respiratory inductive plethysmography using different calibration proce-
dures. American Review of Respiratory Disease, 125(6), 644-649.

Goldberg, A. D., Becker, L. C., Bonsall, R., Cohen, J. D., Ketterer, M. W., Kaufman, P.
G., Krantz, D. S., Light, K. C., McMahon, R. P., Noreuil, T., Pepine, C. J., Raczynski, J., Stone,
P. H., Strother, D., Taylor, H., & Sheps, D. S. (1996). Ischemic, hemodynamic, and neurohor-
monal responses to mental and exercise stress. Experience from the Psychophysiological Investi-
gations of Myocardial Ischemia Study (PIMI). Circulation, 94(10), 2402-2409.

Grossman, P., Karemaker, J. K., & Wieling, W. (1991). Prediction of tonic parasympa-
thetic cardiac control using respiratory sinus arrhythmia: The need for respiratory control. Psy-
chophysiology, 28, 201-216.

Grossman, P., van Beek, J., & Wientjes, C. (1990). A comparison of three quantification
methods for estimation of respiratory sinus arrhythmia. Psychophysiology, 27(6), 702-14.

Johnson, L., Ellis, K., Schmidt, D., Weiss, M., & Cannon, P. (1975). Volume ejected in
early systole. A sensitive index of left ventricular performance in coronary artery disease. Cir-
culation, 52(3), 378-89.

Jordan, C., Henke, K., Stone, A., Brandwayn, L., Belsito, A., & Sackner, M. (1984).
Measurement of systolic time intervals during exercise using inductive plethysmography. Int J
Clin Monit Comput, 1(3), 137-46.

Klein, D. F. (1992). False suffocation alarms, spontaneous panics, and related conditions.
An integrative hypothesis. Archives of General Psychiatry, 50, 306-317.

Konno, K., & Mead, J. (1967). Measurement of the separate volume changes of rib cage
and abdomen during breathing. J Appl Physiol, 22(3), 407-22.

Martinez, J. M., Papp, L. A., Coplan, J. D., Anderson, D. E., Mueller, C. M., Klein, D. F.,
& Gorman, J. M. (1996). Ambulatory monitoring of respiration in anxiety. Anxiety, 2(6), 296-
302.

Meuret, A. E., Wilhelm, F. H., & Roth, W. T. (2001). Respiratory biofeedback assisted
therapy for panic disorder. Behavior Modification, 25(4), 584-605.

VMLA-038-01 17
 Neilly, J., Gaipa, E., Maislin, G., & Pack, A. (1991). Ventilation during early and late
rapid-eye-movement sleep in normal humans. J Appl Physiol, 71(4), 1201-15.

Pitson, D., & Stradling, J. (1998). Autonomic markers of arousal during sleep in patients
undergoing investigation for obstructive sleep apnoea, their relationship to EEG arousals, respi-
ratory events and subjective sleepiness. J Sleep Res, 7(1), 53-9.

Rechtschaffen, A., & Kales, A. e. (1968). A manual of standardized terminology, tech-

niques and scoring system for sleep stages of human subjects. Washington, DC: Public Health
Service, U.S. Government Printing Office.

Roth, W. T., Wilhelm, F. H., & Trabert, W. (1998). Voluntary breath holding in panic
and generalized anxiety disorders. Psychosomatic Medicine, 60(6), 671-679.

Rozanski, A., Blumenthal, J., & Kaplan, J. (1999). Impact of psychological factors on the
pathogenesis of cardiovascular disease and implications for therapy. Circulation, 99(16), 2192-
217.

Sackner, J., Nixon, A., Davis, B., Atkins, N., & Sackner, M. (1980). Non-invasive meas-
urement of ventilation during exercise using a respiratory inductive plethysmograph. I. Am Rev
Respir Dis, 122(6), 867-71.

Sackner, M., Hoffman, R., Stroh, D., & Krieger, B. (1991). Thoracocardiography. Part 1:
Noninvasive measurement of changes in stroke volume comparisons to thermodilution. Chest,
99(3), 613-22.

Sackner, M., Watson, H., Belsito, A., Feinerman, D., Suarez, M., Gonzalez, G.,
Bizousky, F., & Krieger, B. (1989). Calibration of respiratory inductive plethysmograph during
natural breathing. J Appl Physiol, 66(1), 410-20.

Saul, J. P., Berger, R. D., Albrecht, P., Stein, S. P., Chen, M. H., & Cohen, R. J. (1991).
Transfer function analysis of the circulation: unique insights into cardiovascular regulation.
American Journal of Physiology, 261(4 Pt 2), H1231-45.

Spielberger, C. D., Gorsuch, R. L., & Luchene, R. E. (1970). State-Trait Anxiety Inven-
tory. Palo Alto: Consulting Psychology Press.

VMLA-038-01 18
 Stein, M. B., Millar, T. W., Larsen, D. K., & Kryger, M. H. (1995). Irregular breathing
patterns during sleep in patients with panic disorder. American Journal of Psychiatry, 152, 1168-
1173.

Steptoe, A., Smulyan, H., & Gribbin, B. (1976). Pulse wave velocity and blood pressure
change: calibration and applications. Psychophysiology, 13(5), 488-93.

Tobin, M., Jenouri, G., Lind, B., Watson, H., Schneider, A., & Sackner, M. (1983). Vali-
dation of respiratory inductive plethysmography in patients with pulmonary disease. Chest,
83(4), 615-20.

Tobin, M. J., Guenther, S. M., Perez, W., & Mador, M. J. (1987). Accuracy of the respi-
ratory inductive plethysmograph during loaded breathing. J Appl Physiol, 62(2), 497-505.

Watson, H., Poole, D., & Sackner, M. (1988). Accuracy of respiratory inductive plethys-
mographic cross-sectional areas. J Appl Physiol, 65(1), 306-8.

Wilhelm, F. H., Alpers, G. W., Meuret, A. E., & Roth, W. T. (2001). Respiratory patho-
physiology of clinical anxiety outside the laboratory: Assessment of end-tidal pCO2, respiratory
pattern variability, and transfer function RSA. In J. Fahrenberg (Ed.), Progress in ambulatory as-
sessment. (pp. 313-343). Göttingen: Hogrefe & Huber.

Wilhelm, F. H., Gerlach, A. L., & Roth, W. T. (2001). Slow recovery from voluntary hy-
perventilation in panic disorder. Psychosomatic Medicine, 63(4), 638-649.

Wilhelm, F. H., Gevirtz, R., & Roth, W. T. (2001). Respiratory dysregulation in anxiety,
functional cardiac, and pain disorders: assessment, phenomenology, and treatment. Behavior
Modification, 25(4), 513-545.

Wilhelm, F. H., Grossman, P., & Roth, W. T. (1999). Analysis of cardiovascular regula-
tion. Biomed Sci Instrum, 35, 135-140.

Wilhelm, F. H., & Roth, W. T. (1993). Exam 2.0: A program to visualize, edit and ana-
lyze large vectors of data, with application to biomedical engineering. Talk presented at the 1st
Matlab Conference, Boston, USA.

Wilhelm, F. H., & Roth, W. T. (1996). Ambulatory assessment of clinical anxiety. In J.

Fahrenberg & M. Myrtek (Eds.), Ambulatory assessment: Computer-assisted psychological and

VMLA-038-01 19
psychophysiological methods in monitoring and field studies. (pp. 317-345): Göttingen: Hogrefe
& Huber.

Wilhelm, F. H., & Roth, W. T. (1997). Acute and delayed effects of alprazolam on flight
phobics during exposure. Behaviour Research and Therapy, 35(9), 831-41.

Wilhelm, F. H., & Roth, W. T. (1998a). Taking the laboratory to the skies: ambulatory
assessment of self- report, autonomic, and respiratory responses in flying phobia. Psychophysi-
ology, 35(5), 596-606.

Wilhelm, F. H., & Roth, W. T. (1998b). Trusting computerized data reduction too much:
a critique of Anderson's ambulatory respiratory monitor. Biological Psychology, 49(1-2), 215-9;
discussion 221-2.

Wilhelm, F. H., & Roth, W. T. (1998c). Using minute ventilation for ambulatory estima-
tion of additional heart rate. Biological Psychology, 49(1-2), 137-50.

Wilhelm, F. H., & Roth, W. T. (2001). The somatic symptom paradox in DSM-IV anxi-
ety disorders: suggestions for a clinical focus in psychophysiology. Biological Psychology, 57(1-
3), 105-140.

Wilhelm, F. H., Trabert, W., & Roth, W. T. (2001a). Characteristics of sighing in panic
disorder. Biological Psychiatry, 49(7), 606-614.

Wilhelm, F. H., Trabert, W., & Roth, W. T. (2001b). Physiological instability in panic
disorder and generalized anxiety disorder. Biological Psychiatry, 49(7), 596-605.

Wise, M., & Taylor, S. (1990). Anxiety and mood disorders in medically ill patients. J
Clin Psychiatry, 51 Suppl, 27-32.

VMLA-038-01 20
Table 1: Respiratory parameters extracted from the rib cage and abdominal plethysmograph signal of the LifeShirt
system and their clinical significance. Other recorded signals are described in the text.
Parameter Significance
Respiratory rate (f) Highly sensitive but non-specific sign of respiratory dysfunction.
Tidal volume (Vt) Variability helps differentiate between restrictive (less) and obstructive pulmonary
diseases (more) as well as acute anxiety (more). Elevated at baseline in panic disor-
der.
Minute ventilation Equivalent to tidal volume times respiratory rate. Reflects metabolic activity, e.g.,
increased with exercise. Increased (hyperventilation) in acute anxiety, bron-
chospasm, acute pulmonary edema, and cerebrovascular accidents. Decreased (hy-
poventilation) with obstructed breathing, narcotic and sedative overdoses.
Peak inspiratory flow Reflects respiratory center drive. Normally rises with increased metabolic activity.
Similar to mean inspiratory flow, but less dependent on accurate estimation of the
onset and end of inspiration. Elevated at rest in restrictive lung disease because of
increased neural afferent stimulation from intrapulmonary receptors. Elevated in
obstructive pulmonary disease because of increased activity of pulmonary stretch
receptors in chest wall. Decreased with narcotic and sedative overdoses.
Ventilation/peak inspi- Measure of appropriateness of ventilation to respiratory drive. Indirect correlation
ratory flow with intensity of breathlessness, e.g., values low during resistive loading and dy-
namic pulmonary hyperventilation as in acute bronchospasm and in chronic pulmo-
nary emphysema. Helps distinguish organic from psychogenic breathlessness.
Fractional inspiratory "Duty cycle" (Ti/Tt, ratio of inspiratory to total breath time). Low values may reflect
time severe airways obstruction since expiratory time is prolonged. Low values also occur
in normal speech. Values increase during snoring.
Sigh frequency Number of tidal volumes/min that are 2.5 times baseline volume. Increase in count
reflects intensity of anxiety. Elevated at baseline in panic disorder.
Respiratory rate/tidal "Rapid shallow breathing index" (f/Vt), reflects the work of breathing.
volume ratio
Peak/mean inspiratory Reflects presence of upper airway flow limitation during inspiration or expiration.
and expiratory flow
%RC/Vt Percent contribution of the rib cage excursions to tidal volume. Higher in women
than men and during acute hyperventilation. Increased variability with respiratory
muscle fatigue or dysfunction.
Phase relation Measure of thoracoabdominal coordination during breathing. Increased in severe
airways obstruction. Increased during respiratory muscle fatigue or dysfunction.
Apnea & hypopnea Diagnostic components of sleep apnea/hypopnea syndrome and periodic breathing.
detection Hypopnea defined as <50% baseline Vt for >10 sec. Apnea defined as <25% base-
line Vt for >10 sec.
Apnea & hypopnea Phase relation between thorax and abdomen classifies apnea/hypopnea events into
classification central, mixed, and obstructive (1800) types.
Magnitude of periodic Magnitude of oscillations of periodic breathing & Cheyne-Stokes respiration cycles
Vt oscillations associated with apnea or hypopnea.
Cycle time of periodic Cycle length of periodic breathing & Cheyne-Stokes respiration. Longer cycle time
Vt oscillations points to a basis in chronic heart failure, shorter times to idiopathic central sleep ap-
nea syndrome.
Magnitude of changes Elevated in Cheyne-Stokes respiration and periodic breathing.
in end-expiratory lung
volume
Forced expiratory volume Requires a voluntary breathing maneuver. Reflects severity of airway obstruction. Similar to
in 1 sec (FEV1.0) % vital capacity expired in one sec, which can also be measured.

VMLA-038-01 21
 VMLA-038-01 22

View publication stats