WO2003013355A1

WO2003013355A1 - Detection of central apneas

Info

Publication number: WO2003013355A1
Application number: PCT/AU2002/001028
Authority: WO
Inventors: John William Ernest Brydon
Original assignee: Resmed Limited
Priority date: 2001-08-03
Filing date: 2002-08-02
Publication date: 2003-02-20
Also published as: AUPR679601A0

Abstract

The use of diagnostic processing of signals derived from a respiratory analysis mattress (2) to determine the occurence of central apneas is disclosed. A plurality of sensor strips (5) supply electrical signals arising from localised displacement by a patient's body. The individual body effort (displacement) signals are converted into digital form and provided to a processor (104). The occurence of a central apnea is determined by a relative dimunition of the digitised body effort signals. In one form, dimunition is determined by a reduction in amplitude in excess of a predetermined threshold amount. Alternatively, dimunition is determined by forming a ratio of a reference average movement effort value and a contemporaneous movement effort value, and comparing the ratio against a threshold reference value. Advantageously, a further condition can be that there is also a predominant cyclic signal within the range of respiratory origin present.

Description

DETECTION OF CENTRALAPNEAS

Field of the Invention

The invention relates to respiratory-analysis mattresses and systems, and to methods of use thereof, and particularly the detection of central apneas, which is to be understood through this document to include the central component of mixed apneas.

Central (and mixed) apneas can occur in patients suffering from a range of respiratory syndromes, including the disordered breathing associated with Cheyne Stokes syndrome, anaesthetic induced partial respiratory obstruction and sleep apnea.

Background of the Invention

Sleep apnea is a respiratory syndrome known to be present in about 8% of the adult male human population and 4% of the adult female human population.

The syndrome manifests itself as the repetitive cessation of, or large reduction in, breathing while the patient is asleep - respectively termed apneas and hypopneas. Apneas may be divided further into central apneas, where the cause of the apnea is the failure of the nervous system to activate the muscles responsible for respiration, and obstructive apneas, where the patient tries to breath but is prevented from doing so by the temporary collapse on inspiration of his or her upper airway. The reasons for such collapses are not completely understood but may include a loss of tone in those muscles which hold the airway open plus an anatomical disposition towards a narrow upper airway.

Prior to treatment the syndrome must be diagnosed. Conventionally, this is performed by an overnight study in a specialised sleep clinic, connecting the patient to electrophysical and respiratory measurement equipment to monitor physiological variables such as the electroencephalogram, blood oxygen saturation, heartrate, chest wall movement, and respiratory air flow during the various stages of sleep.

The attachment of the such monitoring equipment requires skilled staff and is often disruptive to the patient's sleep. Furthermore, the recording of all the physiological variables requires considerable computing power and the subsequent analysis, although assisted by computer, still requires considerable attention by the staff.

Monitoring of the patient's sleep in the patient's home traditionally uses a simplified form of the above-mentioned equipment which still may be complex and disruptive to the patient's sleep.

The measurement of less disruptive variables which correlate well with the traditional ones has been pursued as a way of making such sleep studies simpler to perform and less disruptive to the patient.

A respiratory-analysis mattress and systems that achieves this objective is disclosed in (WEPO) International Publication No. WO 98/52467 (J W E Brydon) published on 26 November 1998.

The present invention is directed to providing diagnostic processing, utilising the above-noted mattress and systems, to detect central and mixed apneas.

Disclosure of the Invention

The invention provides a method for determining the occurrence of central apneas, comprising the steps of:

(a) measuring patient body movement at different locations on a movement sensitive sheet or mattress to generate a set of independent body effort signals; and (b) processing said signals to detect a relative diminution, said diminution being indicative that a central apnea is occuring.

The invention further provides a movement analysis system for determining the occurrence of central apneas, comprising: a sensor array for accommodating a patient to be in contact therewith, the array having a plurality of independent like-sensors for measuring respiratory movement at different locations on the patient to generate a set of independent body effect signals; and processing means receiving and processing said signals to detect a relative diminution thereof, said diminution being indicative that a central apnea is occurring In the preferred form, processing of the signals includes deriving a measure of the movement rate of any significantly sized periodic body movements that are present, deriving a measure of movement effort, and determining when the movement effort decreases. During such a period of decreased effort, if the rate exceeds a threshold value it is considered that a central apnea has occurred.

In a yet further preferred form, a decrease in movement effort includes the determination of an average movement effort as a reference, forming a ratio of the reference and a contemporaneous actual movement effort, and comparing the ratio value against the threshold value. The tlireshold value represents a minimum reduced movement effort. In one specific embodiment the ratio is formed as (actual effort)/(reference effort), and the threshold value is 0.2, below which movement effort is considered to have decreased. The rate threshold value can be 30 per minute. Furthermore, the movement rate can be determined as the band-passed portion of the movement displacement signals cross-correlated with a previous set of band-passed signals. The band-passed filtering can be between 0.2 to 20 Hz.

During non-central apneic periods the predominant movement signal will derive from respiration while within central apneic periods it will be die to cardiac activity or low amplitude random body movements.

Brief Description of the Drawings

A number of embodiments of the invention will now be described, by way of example only, with reference to the accompanying drawings, in which:

Fig. 1 is a schematic overview of a respiratory-analysis system to be described below;

Figs. 2a and 2b are respectively plan and side views of a movement-sensitive mattress forming part of the system; Fig. 3 is a cross-sectional view through the movement-sensitive mattress;

Fig. 4a is a cut-away schematic drawing of the movement-sensitive mattress showing the internal sensor strips;

Figs. 4b and 4c are top views of further embodiments of movement-sensitive mattresses; Fig. 5 illustrates the use of the movement-sensitive mattress to produce a multichannel electrical signal indicating the displacement of the patient's body near the sensor strips;

Fig. 6 illustrates a sequence of displacements of the patient's body associated with normal breathing;

Fig. 7 illustrates a sequence of displacements of the patient's body associated with disordered breathing;

Figs. 8a and 8b are respectively schematic cross-sectional and plan views of one of the sensor strips; Figs. 9a and 9b show the connection means of a sensor strip respectively before and after connection of the sensor strip thereto;

Figs. 10a and 10b show the connection means with the sensor strip attached, but respectively before and after attachment of a rigid pressure plate;

Figs. 11a and l ib show the connection of a coaxial cable to the connection means;

Fig. 12a shows the attachment of the sensor strip to the connection means;

Fig. 12b is a cross-section taken along A - A' in Fig. 12a;

Fig. 13 shows an alternative embodiment in which the sensor strips are connected to a single bus board instead of to individual circuit boards; Fig. 14 shows an alternative means of providing a plurality of sensor strips, in which the sensor strips are integrally formed from a single PVDF sheet;

Fig. 15 shows conductive tracks on the embodiment of Fig. 14;

Fig. 16a shows an alternative to the embodiment of Figs. 14 and 15, in which the sensor strips, are cut from a narrower PVDF sheet and then folded through 90° as shown in the next figure;

Fig. 16b shows the folding of the sensor strips through 90° while remaining integrally connected to a tail strip;

Figs. 17a and 17b show an embodiment which is the same as that of Fig. 16, except that a broader tail strip is used, so that the tail strip can be folded beneath the sensor strips to provide greater support;

Fig. 18 and 19 show conductive strips on the embodiment of Fig. 16;

Fig. 20 shows the metallisation on each side of one of the sensor strips;

Fig. 21 shows the sensor strips wrapped around one edge of a foam sheet;

Figs. 22a, 22b and 22c show movement-sensitive sheets comprising the embodiments of any of Figs. 14 to 20; Fig. 23 shows the movement-sensitive sheet of Fig. 22a mounted on a carrier sheet, which can be in the form of a conventional fitted sheet;

Figs. 24a to 24d show the connection of the sensor strip (via the coaxial cable shown in Figs. 11a and lib, but omitted from Figs. 14a to 14d) to four alternative embodiments of sensor buffers;

Fig. 25 shows the connection of the sensor strips to computing means via strip connection means, sensor buffers, gain stages, and an analog to digital converter;

Figs. 26 and 27 show an alternate arrangement of respiratory movement sensors;

Fig. 28 shows connection of the sensors of Figs. 26 and 27 to computing means; Figs. 29 and 30 show a yet further arrangement of respiratory movement sensors;

Fig. 31 shows connection of the sensors of Figs. 29 and 30 to computing means;

Figs. 32 and 33 show a yet further arrangement of respiratory movement sensors;

Fig. 34 shows connection of the sensors of Figs. 32 and 33 to computing means;

Fig. 35 shows pre-processing means for deconvolving input digital signals to produce output pre-processed digital signals;

Fig. 36 illustrates the deconvolution of a channel by subtraction of a fraction of the signal on that channel from the two adjacent channels in order to sharpen the spatial response of the channels;

Fig. 37 illustrates the calculation of diagnostic signals from the pre-processed digital signals using basic processing means followed by diagnostic processing means;

Fig. 38 shows a graph of respiratory effort versus time for a reference and test period;

Fig. 39 is a flow diagram of the monitoring state; and

Fig. 40 is a flow diagram of the size drop state.

Description of Preferred Embodiments

Fig. 1 gives an overview of a system 101 which measures the body movements of a reclining person and from those measurements determines parameters of his or her respiratory, cardiac and other movement-related functions. The aforesaid parameters can be used to diagnose a range of respiratory disorders, in particular those associated with sleep apnea. The system can be used both in a hospital and in a patient's home.

The system 101 comprises sensor means 102 which generates electrical signals in response to movement of a reclining person, interface means 103 which converts the said signals into a form that can be processed by the computing means 104 (Fig. 1). The computing means 104 processes the said signals to produce the above-mentioned respiratory and movement parameters which are then further combined to produce parameters diagnostic of respiratory disorders associated with various types of sleep apnea. The function of the computing means 104 is determined by the control means 105, which is operated by medical staff who are directing the use of the system.

The aforesaid processing can be in real time, that is at the same time as the said signals are being recorded, or in a review process where the said recorded signals are recalled from storage and processed at some time after their acquisition.

Some or all of the diagnostic parameters can then be displayed using a display means 106, recorded for subsequent review on computer disk by a recording means 107, printed using a printing means 108, transmitted to another location using a transmission means 109 and output to a recording polygraph by polygraph input means 110. Additionally, if a particular preset condition of the diagnostic parameters is met a video camera 111 can be switched on to record moving or stationary video images of the patient's body position and movements. Alternatively, or optionally, a similar or different preset condition can activate an alarm means 112 to indicate to another person the occurrence of the said preset condition. An external Constant Positive Airway Pressure (CPAP) flow generator may optionally be controlled via CPAP control means 113. Sound output means 114 may be used to listen to snore signals, either in real time or on subsequent replay of data.

The system can operate both in a real time mode, producing diagnostic parameters in immediate response to signals from the sensor means 102 or in a retrospective mode wherein the said signals are replayed from a computer disk and diagnostic parameters calculated at the time of replay.

Referring to Figs. 2a and 2b, sensor means 102 comprises a movement-sensitive mattress 2 which can rest on top of a conventional mattress 3 on which the patient 1 lies. Fig. 2b shows the movement-sensitive mattress 2 above the conventional mattress 3, but this could alternatively be below the conventional mattress 3. The patient's head may optionally rest on a pillow 4. - 1 -

As shown in Fig. 3, movement-sensitive mattress 2 comprises a sandwich of low density polyethylene foam 7 enclosed by a neoprene envelope 6 constructed in such a way that movements of the patient's body cause stretching of the neoprene envelope 6.

Referring to Fig. 4a, a affixed to the inside surface of the top side of the neoprene envelope 6 are a number of sensor strips 5, arranged in one or more patterns that span most of the patient's body. The patterns may run laterally across the movement- sensitive mattress as illustrated, or vertically from head to toe, or a combination or superposition of both. Electrical signals are conducted from the sensor strips 5 by sensor strip connectors 42.

At a minimum, three sensor strips 5 arranged to be level with the patient's rib cage area are required to obtain useful electrical signals utilised for subsequent processing. A typical range is between three and ten sensors.

In Fig. 4b, six sensor strips 5' are arranged in a spaced-apart configuration. The sensor strips are formed in the same manner as those shown in Fig. 4a, however are substantially shorter than the width of the mattress 2. A signal is taken off from each sensor strip 5'. In the limiting case the signals act as spot strain gauges.

In Fig. 4c, the same six senor strips 5' are connected to a common bus connector 5 a that provides for individual take-off points for each sensor strip.

By the above means a multichannel electrical signal is derived, the channels of which reflect the localised displacement of the patient's body in the vicinity of each of the sensor strips 5, as indicated in Fig. 5. By this means the movement of the body during, for example, respiration may be monitored. This, therefore, provides a means of imaging the displacements of the torso, particularly with regard to respiration, in a reclining patient. By virtue of its many sensor strips, the system is largely insensitive to patient orientation on the movement-sensitive mattress 2.

Different respiratory states of the patient produce different patterns of the aforementioned displacements. Fig. 6 illustrates typical patterns during normal respiration while Fig. 7 illustrates typical patterns associated with disordered breathing. Referring to Fig. 8a, the sensor strips 5 are constructed of a layer of polyvinyledene fluoride (PVDF) film 11, a supporting mylar film 13, an adhesive layer 8 to join together the said films and an adhesive layer 41 to adhere the resulting assembly to the inside surface of the neoprene envelope 6, as shown in Fig. 3.

The PVDF film 11 has the property whereby an electrical charge is generated across the faces of the film 11 when a mechanical strain is applied along the length of the film 11 The electrical charge is conducted from the surface of the layer PVDF film 11 by two conductive, metallised surface layers, a first layer 10 and the second layer 12 which are affixed to opposing faces of the film 11 during its manufacture.

The mylar film 13 acts as a physical support for the PVDF film 11 and regulates the amount of strain applied to the said film when the sensor strip 5 is stretched. The mylar film 13 also has applied on one face a conductive, metallised surface layer 14 which is used to screen the second conductive layer 12 of the PVDF film 11 from external electrical interference. The first conductive layer 10 of the said PVDF film 11 is externally connected to the metallised layer 14 of the mylar film 13 so that the second conductive layer 12 of PVDF film 11 is effectively screened on both sides from electrical interference.

Typical dimensions of each sensor strip 5 are 650 mm long by 12 mm wide. The PVDF film is typically 28 μm in thickness and the mylar film 13 typically 1 mil in thickness. The sensor strip 5 can, for example, be made up from the above-mentioned films by the AMP Corporation of PO Box 799, Valley Forge, PA 19482, USA, as a modification of their standard range of piezoelectric film products.

Referring to Figs. 9a and 9b, electrical charge generated by each of the sensor strips 5 is conducted from the sensor strip 5 by a sensor strip connection means 42. The connection means 42 makes connections to the first conductive layer 10 and second conductive layer 12 of the PVDF film 11 and the metallised layer 14 of the mylar film 13, and, further, electrically connects first conductive layer 10 and metallised layer 14 together for electrical screening purposes. The resultant two electrical paths are connected to a coaxial cable 33 for transmission to interface means 103. As shown in Fig. 8a, the various layers at one end of the sensor strip 5 are staggered in such a way as to provide conductive areas 15 and 16 which will be described below. Sensor strip connection means 42 (see Fig. 9a) comprises a double sided printed circuit board 18 with a contact area 19 that makes electrical contact via conductive adhesive with a conductive area 15 (see Fig. 8a) of the metallised layer 14 of the mylar film 13; a contact area 20 that makes electrical contact with a conductive area 16 of the second conductive layer 12 of the PVDF film 11; and a contact area 21 that makes electrical contact with the first conductive layer 10 of the PVDF film 11 by means of a conducting bridge 30 described below. Most of the surface 18a of the printed circuit board 18 adjacent to the aforesaid electrical contacts to sensor strip 5 is unetched, that is, it remains as conductive copper. This allows the electrical connection between the metallised layer 14 of the mylar film 13 and the first conductive layer 10 of the PVDF film 11, both of which are subsequently grounded. Contact area 20 is electrically isolated from said conductive copper by an etched insulating area 22. This allows contact with the ungrounded second conductive layer 12 of the PVDF film 11.

The electrical signal from the conductive area 16 of second conductive face 12 of PVDF film 11, connected to sensor strip connection means 42 via contact area 20 is conducted from the said contact area to connecting pad 23 via copper track 24 located on the reverse side of printed circuit board 18.

As shown in Fig. 9b, the electrical signal from first conductive layer 10 of PVDF film 11 is connected to the conducting copper top face of contact area 21 of printed circuit board 18 by a conducting bridge 30 constructed from copper tape with conductive adhesive on its contact side. The electrical signal from the conducting copper top face of printed circuit board 18 is conducted to a connecting pad 25 on the printed circuit board 18.

As shown in Fig. 10b, conducting bridge 30 and the two other aforementioned sensor strip connections are maintained in a state of intimate connection with their respective contact areas 19, 20, 21 by a non-conducting, rigid pressure plate 31 which bears down on the aforementioned contact assemblies by virtue of two pressure springs

32. Referring to Fig. 11a, to a connecting pad 23 on the printed circuit board 18 is soldered or otherwise electrically attached the inner conductor 35 of a coaxial cable 33. To connecting pad 25 is soldered or otherwise electrically attached the outer screening conductor 34 of the coaxial cable 33.

The coaxial cable 33 is attached to circuit board 18 by a method which simultaneously stress relieves the soldered connections and locates the cable 33. The coaxial cable 33 is located over cable location tongue 27 (as illustrated in Figs. 11a & 1 lb), sourced from circuit board 18 by two parallel slots 28. This arrangement allows a heatshrink sleeve 36 to be pushed simultaneously over the coaxial cable 33 and the cable location tongue 27 so that, on the application of heat, the reduction in diameter of the heatshrink sleeve 36 pulls the coaxial cable 33 into intimate and stable contact with the cable location tongue 27. Adhesive on the interior of the heatshrink sleeve 36 plus its physical grip when shrunk ensure that the coaxial cable 33 is clamped sufficiently for there to be no strain on its internal conductors 34 and 35.

Referring to Figs. 12a and 12b, the circuit board 18 is attached to the interior of the neoprene envelope 6 using a novel arrangement of adhesive that reduces the strain on the electrical connections between the sensor strip 5 and the circuit board 18. The sensor strip 5 is attached to the circuit board 18 using adhesive in location 40; adhesive barrier slot 29 is cut in the said circuit board to prevent adhesive from location 40 straying into contact area 19. Adhesives in the location 40 and subsequently described are all of a high strength cyano-acrylic gel type such as that sold under the registered trademark "Loctite 454". The sensor strip 5 is attached to the neoprene envelope 6 along its length by an adhesive strip 41, for example the transfer adhesive sold under the registered trademark "3M type 9460".

The circuit board 18 is constructed with two strain relief horns 26 which are attached to the interior surface of the neoprene envelope 6 using the above-mentioned cyano-acrylic adhesive applied at locations 37. The function of the strain relief horns 26 is to limit the stretch of the neoprene envelope 6 in the vicinity of the attachment of the circuit board 18 to the sensor strip 5 thus significantly reducing the strain on the aforementioned electrical connections with the strip 5. Optionally, the sensor strip 5 can additionally be stabilised by the application of the said cyanoacrylic adhesive at location 39. The remainder of circuit board 18 is attached to the neoprene envelope 6 using said cyano-acrylic adhesive in at least locations 38.

Referring to Fig. 13, an alternative embodiment 93 combines circuit boards 18 in i parallel on to one long bus board 94 or circuit strip such that the individual connections to strips 5 are conducted in parallel to a single multichannel connector 95 to which is connected a single multicore cable 96 which conducts all the signals from sensor strips 5. Optionally, sensor buffers 43 described below may be located in close proximity to the bus board 94.

Alternative embodiments will now be described.

A die cut part 5 is seen in Fig 14 where a number of sensor strips 5 are cut out from a single sheet of PVDF from which is also formed tail strip 5 A. Separate conductive tracks 17 and 17' on each face of each of the sensor strips 35 (Fig. 15) are formed by selective etching or printing at the metallisation layers of the tail strip, 5 A to conduct the sensor signals to bus connector 42A.

The PVDF film material is normally only producible in strips that can be many metres long but which have a restricted width that may be too narrow to allow the manufacture of a large one-part multistrip assembly 9 in the form visualised in Fig. 14. The one-piece form thus displayed is advantageous from a manufacturing point of view - whereby all the strips 5 are part of a single, die cut sheet and, further, in which the electrical connections from each strip 5 may be conducted from the strip via metallisation on integral tailstrip 5A.

Advantages bestowed by the embodiment of Figs. 14 and 15 are:

(a) cheaper cost of manufacture

(b) more closely matching of sensor electrical characteristics (c) more precise location of sensor strips within the movement sensitive mattress

(d) additionally, the provision of an integral strengthening and location element which stabilises and orients the sensor strips. Fig. 16 shows a further embodiment, which involves the cutting of a relatively narrow (typically 6 cm wide) sheet of PVDF 11 with a pattern illustrated in Fig. 16a, consisting of a number (typically between 4 and 15) of staggered parallel cuts 14 separated by the required width of each sensor strip (typically 1 cm), and of length equal to that required in the aforesaid sensor strips 5, typically 60 cm. The parallel cuts dissect out from the PVDF film, strips 5 whose length is limited only by the length of the said film and not its breadth. Subsequent to the aforesaid dissection, each of the strips 5 is folded into a position 90 degrees from its original orientation at its base 150 via a crease 152 oriented at 45 degrees to the said cuts (Fig. 16b). The residual unfolded element of PVDF sheet 10 serves as an integral tail strip 5 A which conducts the electrical signals away from the said sensor strips to a remote electrical connector. Thus can be achieved the goal of producing a single piece sensor system from a film of restricted width.

Additionally, PVDF sheet 11 may be manufactured with stabilising element 11a (Fig. 17a) consisting of an integral portion of the said sheet which folds underneath and is glued to the sheet and to the folded sensor strips 5 (Fig. 17b). The said 45 degree creasing of the strips 5 is thus immobilised, thereby removing any tendency for the strips

5 to return elastically to their original orientation.

Fig. 18 shows in more detail the electrical connections 17 from each sensor strip

5 along tail strip 5A to bus connector 42A. Normally there will be two separate connections 17 and 17' from each of the sensor strips 5, one from each face.

In another, simpler configuration, one face of each of the strips 5 is comiected in common and that single common connection is conducted to bus connector 42 A along with the single connections from the obverse sides of each individual strip (Fig. 19a).

In a further simplification, all or some of the top faces of the said sensor strips may be connected in common and, separately, all the obverse faces may be likewise connected in common, to give just two electrical connections to the assembly (Fig. 19b). Such a simplification no longer allows the signal from each of the said sensor strips to be recorded separately, rather the output electrical signal is the sum of all the individual responses.

Optionally, further conductive layers or films may be applied over the entire area of the PVDF film 11 to shield the aforesaid connections from external electrical interference.

Figs. 20a and 20b show in more detail the design of the conductive metallisation 13 on the top surface (Fig. 20a) and conductive metallisation 14 on the obverse (Fig. 20b) which collect the strain-generated charge from the strip and conduct it to the bus connector 42A. Electrical charge is only conducted from the faces of the sensor strips 5 when each opposing face is metallised with conductive layers which overlap. In order to limit the area of sensitivity to that of the strip itself, the opposing metallisation patterns are staggered in the region 15 & 16 of said 45 degree crease and, thereafter, on tail strip 5 A. This renders the region of the said crease and the tail strip insensitive to any strains that maybe imposed thereon. This is important because the creasing causes disproportionate strain to be experienced at the crease.

In all the above-mentioned sensor configurations the separate option exists (Fig.

21) to curl sensor strip 5 round the edge of mounting foam sheet 7B, allowing the tail strip 5A and bus connector 42A to be located away from patient contact. This advantageously removes any difference in stiffness that may be felt by the patient when lying on tail strip 5A or bus connector 42A and further protects the said bus connector and associated wires from potential physical damage.

When placing a said movement sensitive mattress on top of a conventional mattress, if the lateral dimensions of the two mattresses differ then the patient on the bed may well be discomfited. By making full use of the inherent thinness of the above- mentioned sensor assemblies the complete movement sensitive mattress may be mounted in an assembly less than 3 mm in thickness, allowing its easy and comfortable location on a range of conventional mattress sizes.

The thin movement sensitive mattresses described above can be regarded as a movement-sensitive sheet.

It may be desirable that the sensor strips 5 be enclosed in a waterproof envelope.

However, the use of a sheet of neoprene or similar rubber to effect the waterproofing function can be clammy and uncomfortable. In order to improve comfort in this regard the construction of Fig. 23 a is used. Movement-sensitive mattress or sheet 2' is constructed as described above with sensor strips 5 connected to tail strip 5A and bus connector 42A, the assembly thereof being sandwiched between thin neoprene or other suitably waterproof, flexible sheet 6A. The said sheet 6A is, however, perforated with holes 6B that allow the assembly to "breathe" - that is, to facilitate the diffusion of humidity from the area in contact with the patient to the conventional bedding beneath the said movement sensitive mattress or sheet. The location of the sensor strips 5 are optionally located by web components 6C.

In a further simplification to the above-mentioned design for increased patient comfort, the sensor strips 5 may be enclosed by the waterproof envelope 6 A only in the immediate vicinity thereof (Fig. 22b). In this configuration gaps 6D between the enclosed sensor strips 5 facilitate the above-mentioned diffusion of humidity away from the subject.

Fig. 22c shows an arrangement of six sensors 5' arranged at the edge margin of the mattress 2. Lateral strain elements 162, acting to channel vertical body displacement to the respective sensor 5' are provided. If preferred, a series of alternating slits 160 can be provided to decouple or isolate adjacent strain elements 162. In another form, channelling of the lateral strain to each sensor 5' may be achieved by use of a substrate material (not shown) in which the lateral (left-to-right) stiffness is significantly greater than the longitudinal (head-to-toe) direction.

In another implementation which improves the ability to locate accurately the sensor strips 5 and aids subject comfort, the movement-sensitive mattress or sheet 2' is mounted on a carrier sheet 7' typically made of cotton or an equivalent porous bed sheeting material or net (Fig. 23). The mounting method for the construction may be permanent, whereby movement-sensitive mattress or sheet 2' is permanently bonded to carrier sheet T, or removable, whereby movement-sensitive mattress or sheet 2' is attached to carrier sheet T by fastenings such as haberdashers' press studs or "Velcro"™ hook and loop material. The construction of carrier sheet T can, advantageously, follow the form of a conventional "fitted" bedding sheet whereby an elasticated border (not shown) holds the carrier sheet T on to a conventional mattress 3.

Referring to Figs. 24a to 24d, electrical signals from each sensor strip connection means 42 are conducted to a respective sensor buffer 43 via the coaxial cable 33 (not shown in Figs. 24a to 24b). The sensor buffer 43 can be of the form where an operational amplifier 51 operates as a charge amplifier (as shown in Fig. 24a), balancing charge received from sensor strip 5 in response to the patient's movement, against charge built up on a capacitor 52 from operational amplifier output 54. This design is commonly used in such situations and referenced in "Piezo Film Sensors Technical Manual O/N: 6571" published by the AMP Corporation of PO Box 799, Valley Forge, PA 19482, USA. This Technical Manual also indicates the necessity of using silicon diodes 55 (as shown in Fig. 24b) to protect the input of the operational amplifier 51 against high voltage transients produced if sensor strip 5 is subjected to a large impulsive force. The action of the diodes 55 is to clamp the input voltage of the operational amplifier 51 to approximately the operational amplifier supply voltages, +V and -V as indicated in the Fig. 24b.

The use of the protection diodes 55 in the above-mentioned configuration does however have a drawback, namely the reverse leakage current of the diodes 55 flows into the virtual earth 46 of the operational amplifier 51 which results in a compensating offset voltage at the output 54 of the operational amplifier 51.

Two solutions to this problem are presented, and shown in Figs. 24c and 24d respectively. The input 47 of the sensor strip 5 to the operational amplifier 51 in the above-mentioned charge amplifier configuration is a virtual earth 46, that is the negative feedback of the operational amplifier 51 acts to maintain the voltage at the input 47 at zero. In practice the input voltage at input 47 may be a small number of millivolts because of constructional imperfections within the operational amplifier 51. Notwithstanding this latter voltage, the input impedance of such a virtual earth is very low (because the operational amplifier acts to drain away charge in order to maintain the virtual earth) - some tens of ohms at the most, therefore an external impedance can be placed between the virtual earth point 46 and ground 99 and, providing said impedance is larger than about 1000 ohms, that is, large relative to the virtual earth impedance, the functioning of the charge amplifier is unaffected. This allows a combination of parallel 56 and serial 57 impedances to replace the above-mentioned reverse biased diodes 55 connecting the virtual earth 46 to the above-mentioned operational amplifier supply rails ( ±V). Whereas the voltage on the sensor strip 5 produced by the accumulation of charge due to a large impulsive force applied thereto may be large - of the order of 100 volts - the effective source impedance of the sensor strip 5 is also very large - up to 10 12 ohms. Hence the addition of even a fairly large impedance (by electronic standards) of 1 Mohm across the sensor strip 5 dramatically reduces the open circuit voltage that can occur across the strip 5. As an additional precaution, a small series resistor 57 can be placed in series with the output of the sensor strip 5 to limit any residual current flow into the operational amplifier 51 input under overload conditions. During non-overload operation these components are effectively invisible to the charge amplifier function - parallel resistor 56 is much greater than the input impedance of the above-mentioned virtual earth and series resistor 57, which is typically 1 kohm, is effectively zero compared with the 10 12 ohms source impedance of the sensor strip 5. One additional advantage of this configuration is that parallel resistor 56 supplies bias current to the operational amplifier 47 input, thus relieving DC feedback stabilisation resistor 53 of any magnitude constraints

(in the above-mentioned conventional charge amplifier, feedback resistor 53 is limited in magnitude because increasing its value increases the output offset voltage of the amplifier).

As an alternative to the above embodiment of Fig. 24c, silicon diodes 46 can be used back to back between the virtual earth 46 and ground 99 (as shown in Fig. 24d). Under non-overload conditions the voltage across the diodes 46 is insufficient for them to conduct, hence they are invisible to the charge amplifier circuit. Under overload conditions one of the diodes 46 will conduct if the voltage increases above about 0.5 volts, thus limiting the overload voltage applied to the input 47 of operational amplifier 51. Optionally, a parallel resistor 47a of about 1 Mohm can be placed in parallel with the diodes 46 to provide bias current for the operational amplifier inputs, thereby relieving the above-mentioned magnitude constraint on DC feedback resistor 53.

DC feedback stabilisation resistor 53 in conjunction with feedback capacitor 52 forms a highpass filter with an effective - 3 dB frequency of approximately 0.1 Hz. Signal components below this value, being largely due to thermoelectric and slow semiconductor drift effects are, therefore, attenuated. This technique is referenced in the aforementioned "Piezo Film Sensors Technical Manual O/N: 6571"

The outputs 54 of the charge amplifiers 43 are then passed through a further gain stage 44 (see Fig. 25) which comprises a low pass filter with a -3 dB frequency point of approximately 100 Hz. Referring to Fig. 25, the outputs of gain stages 44 are input to a multichannel

Analog to Digital Converter (ADC) 45 which has at least as many inputs as there are sensor strips 5. The ADC converter 45 transforms each of the inputs to a numerical digital signal 58 for subsequent processing and storage with a precision of at least 12 bits at a rate of approximately 200 samples per second.

The digital output signals 58 of the ADC 45 are input to computing means 104 which processes the inputs and which stores the digital outputs to computer disk 107 for subsequent retrieval.

Optionally one or more external electrical inputs 48,49 are provided to permit the recording and subsequent processing of signals derived from the movement-sensitive mattress 2. Such signals are, typically, the output from an oximeter (not shown) attached to the finger or ear of the patient, and the output from a pressure transducer (not shown) connected to a mask on the patient's face or nasal prongs inserted in the patient's nares in order to detect respiration.

External electrical inputs 48,49 are connected to combination buffer amplifiers and low pass filters 50 the outputs of which are connected to the inputs of the ADC 45 in parallel with the above-mentioned sensor strip gain stages 44 for similar conversion to digital outputs 45 but at sampling rates typically lower, say at 50 Hz.

Yet further embodiments can be utilized, as follows:

As an alternative to the above embodiment Figs. 2a and 2b, the movement sensitive mattress 2,5 may be replaced, as shown in Fig. 26, by an array of electrode assemblies 205 each assembly being located about a different part of the patient's torso. The assemblies 205 each consist of two electrodes 208, 208' the resistance between which is measured via connections 207, 207' by resistance measuring means 209 shown in Fig. 28. The outputs of measuring means 209 are input to the multichannel Analog to Digital Converter 45 for subsequent processing identical to that used in the abovementioned PVDF sensor based embodiment.

As another alternative to the embodiment of Figs. 2a and 2b, the movement sensitive mattress 2, 5 may be replaced, as shown in Fig. 29, by an array of electrical coil assemblies 210 each assembly being located about a different part of the patient's torso. The said assemblies 210 each consist of a coil of conductive material 211, typicaly made of fine wire, wound typically once round the patient's torso, the inductance of the said coil being measured via connection means 212, 212' by inductance measuring means 213 shown in Fig. 31. The outputs of measuring means 213 again are input to a multichannel Analog to Digitial Converter 45 for subsequent processing identical to that used in the abovementioned PVDF sensor based embodiment.

As a yet further alternative to the above embodiment of Figs. 2a and 2b, the movement sensitive mattress 2, 5 may be replaced, as shown in Figs. 32 and 33, by an array of sealed tube assemblies 214, partially inflated with liquid or gas located beneath the upper torso of the patient on the surface of the bed or within or underneath the mattress, each said tube being connected to pressure measuring means 216 that measures its internal pressure and produces an electrical output 217. The outputs of measuring means 214 again are input to a multichannel Analog to Digital Converter 45 via interface means 218 for subsequent processing identical to that used in the abovementioned PVDF sensor based embodiment.

Referring to Fig. 35, digitised signals 58 resulting from movements of the sensor strips 5 in the movement-sensitive mattress 2 are input to pre-processing means 59

(forming part of the computing means 104) to produce pre-processed digitised signals 60.

The pre-processing means 59 acts both temporally on each individual channel of the digitised signals and spatially on two or more of the digitised signals in concert.

The pre-processing means 59 acts on each channel of digitised signals 58 firstly to equalise the gains of each channel, that is, to remove the variation in amplitude and phase response of each sensor strip 5 relative to the other sensor strips 5, and secondly and optionally to deconvolve the signal of each sensor strip 5 from the effects of adjacent strips 5 (as shown in Fig. 36). Thus processed, the signals are output as pre-processed digitised signals 60.

The above-mentioned deconvolution comprises the subtraction from at least each adjacent channel 63 adjacent to the channel 62 being deconvolved, of a precalculated fraction of the signal measured in said channel 62 such as to remove from said adjacent channels 63 any signal contribution due to physical pressure 61 exerted on the sensor strip 5 corresponding to the channel 62 being deconvolved. The effect of this procedure is to localise or "sharpen" the spatial response for each channel.

Basic Processing

Referring to Fig. 37, and so far as is relevant for the present invention, the pre- processed digitised signals 60 are then separately input, in parallel, to two basic processing means 64, 66 (again forming part of computing means 104) whose function is to extract particular features from the digitised signals, the features subsequently being used in combination to obtain a diagnosis. Some of said basic processing means act temporally on each individual channel of the said digitised signals while others act spatially in concert on two or more of the said pre-processed digitised signals.

Other aspects of the basic processing are described in the aforementioned document WIPO International Publication WO 98/52467.

Respiratory Effort

Basic processing means 64 acts on one or more of pre-processed digitised signals

60 to produce a basic derived signal 65 which is a measure of the integral over time of the patient's movement, regardless of polarity. The processing means 64 firstly derives a signal that is a measure of the patient's instantaneous movement effort ER and is calculated as:

N

Respiratory effort ER(t) = mod,s.(t) ι=l where N is the number of sensor strips 5, and si(t) is the signal derived from the z^'th sensor strip 5 as a function of time. S[ therefore corresponds to the displacement of the zth sensor strip 5.

An alternative calculation is:

1=1 The instantaneous movement effort signal is then further processed within basic processing means 64 to produce the basic derived signal 65, which is a measure of the integral over a fixed time, a complete breath, or the summed separate integrals over the inspiratory and the expiratory phases, of the sum total of the patient's movement, regardless of polarity. The basic derived signal 65 is a measure of the patient's total respiratory effort TR for the breath and is calculated as:

Total respiratory effort TR = J mod s _{ (t) .dt breath '=1 where N is again the number of sensor strips, and mod sχ(t) is the modulus (amplitude) of the signal derived from the z^'th sensor strip 5 as a function of time.

An alternative calculation is

Cyclic Calculation

A further basic processing means 66 can act on pre-processed digitised signals 60 to produce a further basic derived signal 67 which is a measure of the rate of the prevailing predominant cyclic movement signal. Outside periods of central apnea the predominant signal is respiration; within a central apnea it is either the heartbeat or residual low amplitude body movement. This basic derived signal is calculated by correlating the spatial "shape" of the sensor pattern at any given time with the "shapes" of the sensor pattern in past time; the first occurrence of a good correlation (with a coefficient greater than a preset value, typically 0.9) indicates at what time previously a similar pattern occurred, that is, the current respiration rate.

First, each channel of the aforesaid preprocessed digitised signals 60 is passed through a wide band pass digital filter of 0.1 to 20 Hz and, optionally, the sampling rate of the said filtered signals is decimated down to approximately 20 Hz for subsequent computing convenience. At each sampling point in time, the current spatial set of filtered

N sensor signals, ^S«(t) is cross-correlated against the sets sampled at previous times to

give a correlation function:

N

C(t -mT) = Sn(t). ∑S«(t -mT)

the maximum value of m for which correlation function C exceeds the aforementioned threshold is the period separating the present from previous, similar phases of the signal, typically, the breath to breath interval. This measurement is performed at each sampling period, generating many estimates of rate per cycle. The individual estimates of the maximum value of, m, can optionally be low passed filtered to diminish the effect of transient signal artefacts.

Cardiac rate may also thus be measured if the amplitude of the cardiac signal dominates that of respiration, for example during episodes of central apnea.

Diagnostic Processing

The basic derived signals 65 and 67, representing total respiratory/movement effort and wide-band cyclic rate, are input to diagnostic processing means 84 (forming part of computing means 104) which acts to produce a diagnostic signal 85 relating to the detection of central apneas.

Detection of Central Apneas

Fig. 38 shows a plot of the derived instantaneous movement effort signal with time. There are two consecutive periods shown: the "Reference period" and the "Testing period". It is not necessary for the periods to be contiguous, however. The Reference period is typically 30 seconds in duration, and the Test period is typically 10 seconds in duration.

In its simplest form, the diagnostic processing to detect a central apnea consists of detecting the drop in movement effort associated with the cessation of respiratory drive. In an alternative implementation this drop in movement effort is considered in combination with the occurrence of cyclic movement signals with a rate that is too fast to be of respiratory origin (i.e. during the Test Period). Thus, for such a detection to be positive, the amplitude of any respiration present has to have fallen below the amplitude of the ballistocardiogram and/or the residual noise of the measurement system. The only situation in which this occurs is during a central apnea.

The Reference period is utilized to derive an average movement effort measure, and a threshold is derived therefrom. Referring now to Fig. 39, a monitoring process state is described. In step 250, a ratio (Size Drop Ratio) is formed of average movement effort

(from derived signal 65) during the Test period and the average movement effort during the reference period. By average is to be understood to include the mean value, as well as other measures of representative size such as the median, mode, and mean of the 10% to 90% population of sizes within the period.

In step 252, the Size Drop Ratio is compared against a threshold value, typically 0.2. Once the Size Drop Ratio falls below the Threshold, a SIZE DROP is deemed to have started. As shown in step 254, the Reference period is frozen in time and the start of the Test period likewise is frozen.

Referring now to Fig. 40, relating to the SIZE DROP state, in step 250 the end of the Test period is extended, with that time progressively increasing its length. The Size

Drop Ratio continues to be measured between the extended Test period and the frozen Reference period until the ratio increases above the Size Up Tlireshold (typically 0.4) at which time the SIZE DROP is deemed to have ended.

In step 252, the Wideband Rate is calculated (i.e. signal 67). An alternative is to count the zero crossings in the past time correlation signal and divide by two.

In step 254, it is determined whether the Wideband Rate is higher than a High Rate Threshold (typically 30 per minute), the dominant signal is presumed not to be respiratory. If respiration is present this rate will return to a value typically below 30 per minute.

If the High Rate Threshold is exceeded during a SIZE DROP an "Akinetic Event" is diagnosed (a flag is set in step 266). Such an Event is considered to be synonymous with a central apnea, for the reason that only the effective total loss of respiratory movement can produce such a scenario.

In step 268, the end of an Event is determined when the Size Drop Ratio reduces below the Size Drop Threshold.

At this stage, various diagnostic records are made, including the duration of the apnea and the separation time between its start and that of any previously diagnosed central apnea. This data may have relevance in the cardiological determination of the severity of heart failure. The Size Drop Ratio can also be recorded.

A further condition can be that the Reference signal average size is above a No Signal Threshold. This prevents operation when a patient is not lying on the mattress.

It is understood that various alterations and modifications can be made to the techniques and arrangements described herein, as would be apparent to one skilled in the art.

Claims

CLAIMS:

1. A method for determining the occurrence of central apneas, comprising the steps of: (a) measuring patient body movement at different locations on a movement sensitive sheet or mattress to generate a set of independent body effort signals; and

(b) processing said signals to detect a relative diminution, said diminution being indicative that a central apnea is occuring.

2. A method as claimed in claim 1, comprising the further step of:

(c) processing said effort signals to detect the occurrence of the predominant cyclic signal; and wherein step (b) determines that a central apnea is occurring only if, additionally, the cyclic signal is not within the range of respiratory origin.

3. A method as claimed in claim 1 or claim 2, wherein said relative diminution detection is performed by: processing said effort signals to detect the occurrence of a decrease in amplitude; and determining whether the decrease in amplitude exceeds a threshold value.

4. A method as claimed in claim 1 or claim 2, wherein said relative diminution detection is performed by: determining an average movement effort as a reference; forming a ratio of said reference movement effort and contemporaneous actual movement effort; and comparing said ratio value against a threshold value, said threshold value representing a minimum reduced respiratory effort.

5. A method as claimed in claim 4, wherein said ratio is formed as (actual effort)/(reference effort), and said threshold value is 0.2, below which movement effort is considered to have decreased.

6. A method as claimed in claim 5, when movement effort is determined as a summation function of said movement displacement signals.

7. A method as claimed in any one of claims 2 to 6, wherein said respiratory origin range is 2-30 per minute.

8. A method as claimed in any one of claims 2 to 7, wherein said predominant cyclic signal is determined as the band-passed portion of said respiratory displacement signals cross-correlated with a previous set of band-passed signals.

9. A method as claimed in claim 8, wherein said band-pass filtering is between 0.2 to 20 Hz.

10. A method as claimed in any one of claims 2 to 7, wherein said predominant cyclic signal is determined from the number of zero crossings of the past time correlation signal divided by two.

11. A movement analysis system for determining the occurrence of central apneas, comprising: a sensor array for accommodating a patient to be in contact therewith, the array having a plurality of independent like-sensors for measuring body movement at different locations on the patient to generate a set of independent body effect signals; and processing means receiving and processing said signals to detect a relative diminution thereof, said diminution being indicative that a central apnea is occurring.

12. A system as claimed in claim 11, wherein said ratio is formed as (actual effort)/(reference effort), and said threshold value is 0.2, below which movement effort is considered to have decreased.

13. A system as claimed in claim 12, when movement effort is determined as a summation function of said movement displacement signals.

14. A system as claimed in any one of claims 11 to 14, wherein said rate threshold value is 30 per minute.

15. A system as claimed in any one of claims 11 to 15, wherein said movement rate is determined as the band-passed portion of said movement displacement signals cross-correlated with a previous said of band-passed signals.

16. A system as claimed in claim 15, wherein said band-pass filtering is between 0.2 to 20 Hz.