PORTABLE MEDICAL DEVICE

Abstract

A portable medical device adapted to be carried by a user, comprising at least one electrically actuated means for performing a therapeutic and/or a diagnostic function on a patient, at least one mobile electric power supply, a control unit having at least one microelectronic controller, at least one electronic memory and at least one multi-channel impact sensor unit. The impact sensor unit is adapted to simultaneously detect linear acceleration along the three space axes and angular rates around the three space axes. The impact sensor unit generates impact sensor data sets, which are electronically transferred to the microelectronic controller via an interface. The device further comprises a housing, in which the electric components of the device are contained. The device is adapted to execute a classification algorithm, which allows to discriminate a first impact sensor data set, said first impact data set being correlated to a first impact event which is detrimental for the device, from a second impact sensor data set, said second impact data set being correlated to a second impact event which is not detrimental for the device.

Claims

1. A portable medical device adapted to be carried by a user, comprising at least one electrically actuated means for performing a therapeutic and/or a diagnostic function on a patient, at least one mobile electric power supply, a control unit having at least one microelectronic controller and at least one electronic memory, at least one multi-channel impact sensor unit, wherein the impact sensor unit is adapted to simultaneously detect linear acceleration (a.sub.x; a.sub.y; a.sub.z) along the three space axes (x; y; z) and angular rates (.sub.x; .sub.y; .sub.z) around the three space axes (x; y; z), wherein the impact sensor unit generates impact sensor data sets, and the impact sensor data sets are electronically transferred to the microelectronic controller via an interface, a housing, in which the electric components of the device are contained, wherein the microelectronic controller is adapted to execute a classification algorithm which allows to discriminate a first impact sensor data set, said first impact data set being correlated to a first impact event which is detrimental for the device, from a second impact sensor data set, said second impact data set being correlated to a second impact event which is not detrimental for the device.

2. Portable medical device according to claim 1, wherein the classification algorithm includes a support vector machine.

3. Portable medical device according to claim 2, wherein the classification algorithm includes a support vector machine using a two-dimensional space and a one-dimensional separation line (hyperplane).

4. Portable medical device according to claim 3, wherein the first dimension of the two dimensional space is defined by the magnitude M.sub.a corresponding to the vector length of the linear acceleration values (a.sub.x; a.sub.y; a.sub.z) measured by the acceleration sensor and wherein the second dimension of the two dimensional space is defined by the magnitude M.sub. corresponding to the vector length of the angular rate values (.sub.x; .sub.y; .sub.z) measured by the rotation sensor.

5. Portable medical device according to claim 1, wherein the impact sensor unit comprises a microelectronic inertial measurement unit (IMU) for detection of linear acceleration and rotation and wherein the microelectronic inertial measurement unit (IMU) is integrated in a single chip.

6. Portable medical device according to claim 1, wherein the at least one electronic memory is adapted to store operating instructions for the microelectronic controller and/or to record status information related to the function of the device.

7. Portable medical device according to claim 6, wherein the at least one electronic memory is further adapted to store sensor data sets generated by the at least one impact sensor unit and/or to store a result obtained by the classification algorithm.

8. Portable medical device according to claim 1, herein the at least one electronic memory comprises a separate electronic impact sensor data memory, in particular a flash memory, and the separate electronic impact sensor data memory is adapted to store impact sensor data sets generated by the at least one impact sensor unit and/or a result obtained by the classification algorithm.

9. Portable medical device according to claim 1, wherein the device further comprises a separate mobile electric power supply, in particular a battery, a rechargeable accumulator or a capacitor, and the separate electric power supply is adapted to supply electric energy to the impact sensor unit.

10. Portable medical device according to claim 1, wherein the impact sensor unit is adapted to activate an inactive microelectronic controller, if the impact sensor unit detects at least one linear acceleration value (a.sub.x; a.sub.y; a.sub.z) or at least one angular rate value (.sub.x; .sub.y; .sub.z) above a predetermined threshold level.

11. Portable medical device according to claim 1, further comprising a real time clock, whereby the device is adapted to record the time of an impact, if the impact sensor unit detects at least one linear acceleration value (a.sub.x; a.sub.y; a.sub.z) or at least one angular rate value (.sub.x; .sub.y; .sub.z) above a predetermined threshold level.

12. Portable medical device according to claim 1, further comprising a real time clock, whereby the device is adapted to record the time at which the classification algorithm has produced a discrimination result.

13. Portable medical device according to claim 1, wherein the microelectronic controller comprises a user interface adapted to receive user settings and/or to display one or more system conditions.

14. Portable medical device according to claim 1, wherein the device is selected from the group of an ambulatory wound therapy device, in particular a negative pressure wound therapy device, an ambulatory deep brain stimulation device, an ambulatory cardiac stimulator, an ambulatory infusion device, in particular an insulin pump, an ambulatory blood pressure unit, an ambulatory pulse monitor, an ambulatory electrocardiogram device, an ambulatory electroencephalogram device and an ambulatory diagnosis system for determining metabolites circulating in the blood.

15. Method of discriminating a first and a second impact event, comprising the steps providing a portable medical device according to claim 1, detecting the first or a second impact event by the impact sensor unit, executing a classification algorithm, which allows to discriminate a first impact sensor data set, said first impact data set being correlated to a first impact event which is detrimental for the device from a second impact sensor data set, said second impact data set being correlated to a second impact event which is not detrimental for the device recording the result obtained from the classification algorithm, and optionally displaying an alarm message to the user, if the result obtained from executing the classification algorithm indicates, that a first impact event, which is detrimental for the device, has occurred.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0161] Further characteristics, details, and advantages of the invention result from the appended patent claims and from the drawings and the following description of preferred embodiments of the invention.

[0162] The drawings show a negative pressure wound therapy device as preferred (but non-limiting) example of a portable medical device. The invention can be used for many other kinds of portable medical devices as described in detail in the above chapters of the description.

[0163] The drawings show:

[0164] FIG. 1 A schematic drawing of a simple negative pressure wound therapy device including the negative pressure bandage applied to a wound of a patient.

[0165] FIGS. 2 a to e Different views of a typical portable negative pressure wound therapy device to generate a vacuum for medical applications.

[0166] FIG. 3 A schematic drawing of the piping system and of the electronic components of a negative pressure wound therapy device according to a preferred embodiment of the present invention.

[0167] FIG. 4 A schematic drawing of the structure of the electronic control system and of the electric/electronic/eletromechanical components of a portable negative pressure wound therapy device according to a preferred embodiment of the invention.

[0168] FIG. 5 A schematic drawing of the structure of a stand-alone impact sensor unit module according to a preferred embodiment of the invention.

[0169] FIG. 6 Example of a warning message displayed to the user of a portable negative pressure wound therapy device according to a preferred embodiment of the invention.

[0170] FIGS. 7 a-f Signal magnitudes determined from impact experiments (training experiments). The charts show the values of the magnitudes M.sub. and M.sub. in relative units. [0171] 7 a Distribution of signal magnitudes from 102 impact experiments [0172] 7 b Subset pendulum experiment (pool of EXP 1 A/B) [0173] 7 c Subset tilting experiment (EXP 8) [0174] 7 d Subset stairs experiment (pool of EXP 9 A/B) [0175] 7 e Subset chair experiment (EXP 7) [0176] 7 f Subset driving experiment (EXP 8)

[0177] FIGS. 8 a, b Acceleration forces and angular rates detected during a single impact event. From the detected linear acceleration forces along the three axis the magnitude M.sub. was calculated (FIG. 8 a). Rotation magnitude M.sub. was calculated (FIG. 8 b) from the gyration sensor recordings made during the same drop experiment.

DESCRIPTION OF THE PREFERRED EMBODIMENT

[0178] Preferred embodiments of the invention are described by means of a negative pressure wound therapy device. A mobile negative pressure wound therapy device serves as a non-limiting example for a portable medical device of the invention.

[0179] A schematic representation of a simple negative pressure wound therapy device 1 is shown in FIG. 1. The therapy device is in fluid communication with a wound 2 of a patient to be treated. The device 1 comprises a container 3 for collecting the fluids (such as blood, instillation fluids and the like) that are sucked from the wound 2. Wound therapy devices of this type are known in the prior art. The container (or canister) 3 is typically made of a solid material, such as a plastic material. It is usually a disposable article designed for single use. Conveniently, the container 3 can be detachably mounted to the housing part 4 of the device, which contains the electrical components of the apparatus. The container 3 can be evacuated by the electrically actuated suction pump 5. A connection (not shown) is provided for a suction line 6 that leads to the wound such that vacuum communication can be established between the suction pump 5, the container 3, and the suction line 6 that leads to the wound. A filter or air/liquid-separator 7 located within the fluid-pathway between the container 3 and the suction pump 5 is used to prevent exudate from being sucked into the pump 5. A negative pressure wound therapy device typically comprises additional components such as a control system for controlling activity of the pump and means for interacting with the user, such as a touch-screen display or control buttons. These components are not shown in FIG. 1.

[0180] In some embodiments, the portable negative pressure wound therapy device does not have a container for receiving the drained body fluids. Instead, the body fluids can be contained, for example, in the dressing. This is achieved by providing absorbent layers (not shown in FIG. 1). Such negative pressure wound therapy devices, which do not make use of a separate solid exudate canister are typically used for treating less exudating wounds, for example surgical wounds.

[0181] FIGS. 2 a to e show a typical example of a portable device 1 for the provision of the vacuum for medical applications. The device 1 comprises a first housing part 4 in which a vacuum-producing device in the form of an air suction pump as well as the electrical components for the device are accommodated, including batteries or preferably rechargeable batteries. A recharging connection for the batteries is designated by reference symbol 8. Moreover, the device 1 comprises a second housing part in the form of a container 3 for receiving body fluids, in particular, for receiving wound exudates suctioned away from a wound. Preferably, the entire second housing part is constituted as a disposable single-use item. In its upper region, a connection gland 9 for a suction tube is provided that may, for example, lead to a wound dressing sealing the wound when the device 1 is used in the vacuum therapy of wounds and there it can, for example, communicate with the wound space through a port to apply and maintain a vacuum to the wound space and to suction away wound exudates into the container. For this purpose, the container 3 communicates with the suction pump 5.

[0182] It can also be seen from FIG. 2 d on the side 10 of the second housing part 3 facing the body, a grip recess 11 is formed in the shape of an opening extending right through the second housing part 3. In this way, the device 1, or only its second housing part 3, can be gripped and handled with one hand.

[0183] In the preferred embodiment shown, a manually operable element 12 is provided near this grip recess 11 on the upper side of the device 1, for example, in the form of a pushbutton that acts on locking and back-gripping means (not shown). In the joined condition of the two housing parts 3 and 4, the locking or back-gripping means are in a locked condition holding the two housing parts 3, 4 together by positive action. Only upon operation of the operating element 12, the lock is released so that the housing parts 3, 4 can be separated from each other.

[0184] FIG. 3 shows the nature of the piping system and of the electric components (including electronic and/or electromechanical components) of a negative pressure wound therapy device according to a preferred embodiment of the invention. According to this preferred embodiment, the device comprises an inventive microelectronic impact sensor, wherein the impact sensor is adapted to detect an acceleration impact acting on the device. Apart from the inventive feature, the device is similar to negative pressure wound therapy device of the type exemplified in FIG. 2. In contrast to the very basic system shown in FIG. 1, the device of FIG. 3 includes additional components (known from the art) such as the air rinsing pathway of the fluid system. Based on FIG. 3 one embodiment of the inventive device is explained. FIG. 3 shows the previously described or a similar device for providing a vacuum for medical applications in a purely schematic representation, wherein relevant reference symbols are used for the corresponding components. However, FIG. 3 shows only the components that are relevant for the following description. FIG. 3 depicts a schematically indicated wound to be treated with a vacuum with a vacuum-tight wound dressing 13, to which the suction tube 6 emanating from the container 3 leads. From the container 3, a further tube section 14 leads outwardly through the already mentioned filter 7. If the container 3 or the first housing part 4 is put into its operating position on the first or basic housing part 4 of the device 1, the tube section 14 is connected to a further tube section 15 within the first housing part 4 that leads to the intake side of the suction pump 5. When the suction pump 5 operates, a vacuum is applied to the container 3 and to the suction tube 6 via tube sections 14, 15.

[0185] Moreover, a pressure sensor 17 for measuring the pressure is provided in the tube section 15 between container 3 and suction pump 5. Its signals are sent to an electronic control unit, collectively identified by reference symbol 18, which performs open-loop and closed-loop control of the device 1 in total. The electronic control unit 18 comprises a microelectronic controller and at least one electronic memory. Also shown is the charging connection 8 for rechargeable batteries that are located in a compartment 19 and a connection 20 for a schematically indicated power supply unit 21. Reference symbol 22 indicates a display unit, preferably having a capacitive switch membrane (touchscreen). A user may control operation of the device via said touchscreen. The electrical connection to the electronic control unit 18 is indicated via electrical lines 23. The suction pump 5 is controlled by the electronic control unit 18 in which, by means of the signals of the pressure sensor 17, a pressure and vacuum closed-loop control is implemented with known open-loop and closed-loop control mechanisms (set point/actual value control mechanisms), so that the pressure value corresponding to the currently selected program is controlled in the tube section 15.

[0186] Also shown is an additional rinsing or aeration tube 24 that, only shown by way of example, leads through the container 3 and just like the suction tube 6 leads to the wound dressing 13. When the container 3 is attached in its intended assembly position on the first housing part 4, this rinsing tube 24 communicates with a tube section 25 provided in the first housing part 4 in which an electromagnetically operated valve 26 is provided that can be actuated by the electronic control unit 18 and connects the tube section 25 with the atmospheric air when it is open, so that an air current toward the wound via the rinsing tube 24 can be generated.

[0187] The device 1 and its electronic control unit 18 also feature a data interface 27, preferably a USB interface, by means of which the electronic control unit 18 or its method of operation can be programmed. In addition, device 1 comprises a speaker 28 which is connected to the control unit 18. The speaker can be used to generate acoustic alarm signals.

[0188] According to the invention, the portable negative pressure wound therapy device further comprises a multi-channel impact sensor unit 29, which is capable of communicating with electronic control unit 18. As explained above, the electronic control unit 18 comprises a microcontroller. The multi-channel impact sensor unit 29 has three linear acceleration sensor channels as well as gyration sensors having three channels for angular rotation. The IMU motion sensor LSM9DS0 from ST Microelectronics, United Kingdom, can be used. Advantageously, the LSM9DS0 additionally comprises a magnetic field sensor, which can be used to detect magnetic fields that could harm the electronic components of the device.

[0189] In the example shown in FIG. 3, the multi-channel impact sensor unit 29 communicates directly, by means of an interface (not shown), with the electronic control unit 18. In this embodiment (herein called an integrated impact sensor), the impact sensor unit 29 is an integrated component of the control system 18 of the device. Also, the impact sensor data are processed by the electronic control unit 18 and the sensor raw data and/or the processed data storage is controlled by the electronic control unit 18. The integrated impact sensor unit 29 is usually (but not necessarily) provided with electric energy by the main power supply system of the device.

[0190] More preferably (unlike the example shown in FIG. 3), the multi-channel impact sensor unit is a component of a stand-alone impact sensor unit module. A stand-alone impact sensor unit module is explained in more detail in FIG. 5 below.

[0191] Interaction of the multi-channel impact sensor unit 29 with the control unit 18 is also shown in FIG. 4 below.

[0192] Multi-channel impact sensor unit 29 can conveniently be mounted to the same main board, on which electronic control unit 18 is sitting.

[0193] Proceeding to FIG. 4, communication of the control unit with peripheral components is shown schematically. One peripheral component communicating with the controller is the integrated multi-channel impact sensor unit (termed IMU in FIG. 4). Instead of an integrated multi-channel impact sensor unit a stand-alone impact sensor module can be used likewise.

[0194] The control unit includes elements/functions such as the microelectronic controller (microcontroller), a protection circuit for the power supply, a fuel gauge for the power supply, a charging circuit for the power supply, power converter circuits, a speaker controller and switching circuits for controlling the suction pump and the electromagnetic valve.

[0195] The peripheral electric/electronic components include a source of electric energy, preferably a rechargeable battery, the suction pump, preferably a membrane pump and at least one electronic memory typically including both ROM and RAM chips. In up-to date devices, the peripheral electric/electronic components may additionally include, for example, a vacuum valve for a rinsing fluid pathway (as exemplified in FIG. 3), interfaces such as a USB connector, electric connectors, such as a connector for the battery charger, a light sensor, a pressure sensor for sensing the pressure within the closed fluid-system of the device, a speaker for signalling warnings to the user, one or more LEDs for signalling warnings and/or status information to the user, a display, preferably a touchscreen display for interacting with the user, a backlight for the display, buttons, such as an on/off button. More advanced peripheral components such as a wireless interface for communication of the control unit with a remote computer may optionally be present (box with dotted line in FIG. 4).

[0196] The control unit is the central instance for controlling and monitoring function of the device.

[0197] According to the invention the peripheral components include a multi-channel impact sensor unit (IMU), which may be an integrated impact sensor unit or preferably a stand-alone impact sensor module, each of which communicate with the microcontroller via an interface. The microcontroller receives impact sensor data (i.e. sensor data from any of the six channels of the multi-channel impact sensor unit). If an integrated impact sensor unit is used, the microcontroller receives the sensor raw data from the 6 channels. The raw data are processed as explained in detail above. If a stand-alone impact sensor module is used, the microcontroller receives the processed data including the classification results and may, for example, display warnings to the user or lock the device.

[0198] The stand-alone impact sensor module shown in FIG. 5 comprises at least one multi-channel impact sensor unit (IMU) and further a microelectronic controller (in the following called impact sensor module controller; in FIG. 5 the impact sensor module controller is shortly termed pController) cooperating with the IMU. The impact sensor module controller functions independent of the microelectronic controller of the control unit. The module further comprises a data interface, such as a RS232 interface, a memory, such as a flash memory and a real-time clock. The sensor, the interface, the memory and the real-time clock are in direct electronic communication with the impact sensor module controller. A separate battery or accumulator supplies electric energy to the components of the module thus making the module independent of the charging status of the main battery of the device. A particular advantage of using a stand-alone impact sensor module is, that the module can function independently of the working status of the device. The stand-alone impact sensor module can be active even if the device is switched off, as described in connection with the preferred embodiments herein. Thus, the stand-alone impact sensor module may perpetually report detrimental impacts that were acting on the device, for example also during storage of the device or during transport of the device.

[0199] FIG. 6 shows an example of a warning message that is signalled to the user by means of a LCD-display. Such a message can be displayed after the controller of the device has classified the data received from the impact sensor unit as detrimental for the device (first class). The alarm message according to this example informs the user that an impact was acting on the device and furthermore that the device is locked. Unlocking of the device can only be accomplished by a service technician authorised by the supplier/manufacturer. Before unlocking the device the service technician will perform a careful inspection of the device, optionally in combination with a prophylactically exchange of those components, which are typically prone to deficiency after a detrimental impact. Hereby the service technician can also utilise information about the strength of the impact, which can be stored in the memory of the device.

Example 1

[0200] In order to establish an exemplary training data set for a portable medical device (here: a mobile negative pressure wound therapy apparatus), the signal magnitudes (linear acceleration as well as rotation) were determined from a plurality of impact experiments (in total 102 experiments). Signal magnitudes are given as relative values. During the experiments the device (portable negative pressure wound therapy unit VivanoTec) was subjected to impacts, which are usually detrimental for the device (first class of impacts) as well as to impacts, which are usually not detrimental for the device (second class of impacts).

[0201] The medical device was subjected to the following types of impacts. The number of experiments is shown in brackets:

[0202] EXP 1A Horizontal crash of display against a wall, while device is swinging from its carrying belt. This kind of crash occurs very often, if a user carries the device around his/her neck (5).

[0203] EXP 1B Horizontal crash of backside/container against a wall, while device is swinging from its carrying belt. Same as 1A this kind of crash occurs very often, if a user carries the device around his/her neck (4).

[0204] EXP 2A Crash to carpet from height of 100 cm with empty exudate container (10). EXP 2B Crash to steel plate from height of 100 cm with empty exudate container (3).

[0205] EXP 3 Crash to steel plate from height of 100 cm with full 800 ml exudate container (10).

[0206] EXP 4 Same as EXP 2B (10).

[0207] EXP 5 Crash to steel plate from height of 50 cm with empty exudate container (10).

[0208] EXP 6 Crash to steel plate from height of 25 cm with empty exudate container (10).

[0209] EXP 7 Patient sitting down to a chair (10).

[0210] EXP 8 Device was overturned (90 turn) while standing on a table (10).

[0211] EXP 9A Patient walking/jumping down a stair in a fast mode (5).

[0212] EXP 9B Patient walking down a stair in a slow mode (5).

[0213] EXP 9B Patient was driving in a car including driving on a dirt road (10).

[0214] FIG. 7 a shows the distribution of signal magnitudes from 102 impact experiments, of which 53 (round dots) belong to the first class of impacts and 49 (triangles) belong to the second class of impacts. Sensor data were recorded by means of an inertial sensor module mounted to the main board of a portable negative pressure wound therapy device (VivanoTec negative pressure unit from Paul Hartmann AG, Germany). For the experiments the IMU motion sensor LSM9DS0 (iNEMO inertial module) from ST Microelectronics, United Kingdom was used. This inertial module device has three acceleration channels, three angular rate channels and three magnetic field channels. Data from the magnetic field channels were not recorded. From the detected acceleration forces and angular rates the magnitudes M.sub. and M.sub. were calculated. The chart given in FIG. 7a shows each impact event mapped according to its maximum values of the magnitudes M.sub. (x-axis) and M.sub., (y-axis) observed during the impacts.

[0215] Experiments 2A, 2B, 3, 4, 5, 6 belong to the first class of impacts, which are usually detrimental for the device (dropping to a ground type of impacts). The events resulting from experiments 2A, 2B, 3, 4, 5, 6 are represented by the round dots. The other experiments 1A, 1 B, 7, 8, 9, 10 shown belong to the second class of impacts, which are usually not detrimental for the device. During such kinds of impacts, which occur regularly, when the portable device is used outside of a medical facility, may also be accompanied by significant acceleration and rotation movements. However, a portable medical device is usually constructed to withstand such impacts without being harmed. The events resulting from experiments 1A, 1 B, 7, 8, 9, 10 are represented by the triangles.

[0216] The events corresponding to the first class (round dots) form a clearly recognisable group. Likewise, which the events corresponding to the second class (triangles) form another coherent group. The first and the second classes do not overlap. It is possible to separate the classes by a linear classifier, namely hyperplane H shown in FIG. 7h. Hyperplane H should is chosen with maximum margins to both classes of events. Preferably a support vector machine algorithm is used to construct the hyperplane. Any new impact event detected by the impact sensor system of the portable device can be mapped into the diagram and will fall to either side of the hyperplane resulting in a classification detrimental or not detrimental.

[0217] The experiments show that the first and the second class of impacts are clearly discriminable by a simple classification approach based on plotting the magnitudes M.sub. and M.sub. (as explained more in detail above) in a two-dimensional plane. Hyperplane (marked as H in FIG. 7a) can be generated using a standard support vector machine approach.

[0218] If a linear separation of the two classes should not be possible (which was not observed in the experiments done by the inventors) known Kernel methods can be used to establish a separator.

[0219] In order to collect training data sets, it is necessary to repeatedly perform experiments of the type described in this example. Training experiments will have to be done for each type of device, because classification depends on device specific parameters like stability, weight, material and shape of the device. Reliability of classification increases with the number of training experiments.

[0220] The experiments also demonstrate the usefulness of the classification approach based on a plurality of sensor signals (i.e. from the six sensor channels) in combination with calculating the acceleration and rotation vectors (i.e. magnitudes M.sub. and M.sub.). As can be seen from diagram FIG. 7a a discrimination of the first and the second class of impacts would not be possible, if only linear acceleration sensor date would be used: The acceleration magnitudes M.sub. corresponding to experiments EXP 1A, 1 B, 8 (see FIGS. 7b and 7c, which belong to the first class, i.e. not detrimental) overlap with the magnitudes M.sub. corresponding to experiments EXP 2A, 2B, 3, 4, 5, 6 (belonging to the first class, i.e. detrimental). Only the second dimension, i.e. the rotation vectors, allows for a linear discrimination. Similarly, it would not be possible to find a linear separator dividing the first group from the second class by using the rotation vectors alone.

[0221] FIGS. 7 b-f show subsets of the results for the different types of not detrimental impacts. Data sets of the pendulum experiments EXP 1 A/B (FIG. 7b) as well as of the stairs experiments EXP 9 A/B (FIG. 7d) were pooled, respectively. FIG. 7c shows the data sets from the tilting experiment EXP 8, where an apparatus was overthrown. This group of EXP 8 is characterised by relatively similar rotational magnitudes M.sub., whereas the acceleration magnitudes M.sub. exhibit a comparatively stronger variability. The subset shown in FIG. 7e represents the chair experiment EXP 7. Data from subset driving experiment EXP 8 can be seen in FIG. 7f.

[0222] All different data subsets form clearly discernible groups, when the magnitude of the rotation rates M.sub. is plotter against the magnitude of the linear acceleration magnitudes M.sub.. The experiments shown herein demonstrate, that a classification algorithm can be used to assign impact sensor data sets to specific user actions. Classification and detection of user action could be used, for example, to warn a user if he enters a condition, which is not recommended during the type of therapy applied.

Example 2

[0223] FIGS. 8 a, b shows an example of processed impact sensor data recorded by means of an inertial sensor module mounted to the main board of a portable negative pressure wound therapy device (VivanoTec negative pressure unit from Paul Hartmann AG, Germany). For the experiments the same IMU motion sensor as the one used for the experiments described in example 1 was used (LSM9DS0 inertial module from ST Microelectronics). As in example 1, data from the magnetic field channels were not recorded.

[0224] The data corresponding to FIG. 8 a, b were received while the device was subjected to a drop test. During the test the device was dropped to a soft ground (rubber plate) from a height of 1.0 m above ground. A maximum magnitude M.sub. value >11 (relative units) was observed. A drop of this kind in many cases does not leave any damages visible from the outside to the device. However, the acceleration occurring during the drop imposes stress to internal mechanical or electronic components, such as electric contacts or tube connections, which may lead to a malfunction during future therapy cycles. Accordingly, a drop of this category has to be classified as an impact event, which is detrimental for the device.

[0225] The data corresponding to FIG. 8 b shows magnitude M.sub. reflecting the overall gyration/rotation forces that the device was exposed to during the same experiment (drop to a soft ground) already described under FIG. 8 a. A maximum magnitude M.sub. value of >42 (relative units) was observed. A comparison of FIG. 8 a with FIG. 8 b shows that the peak of acceleration magnitude M.sub. coincides with the peak of gyration magnitude M.sub.. This indicates that the crash entails both strong linear acceleration and strong rotation forces, which occur simultaneously.

[0226] FIG. 8a also shows the incidents occurring during the impact experiment: The free fall phase of the experiment, during which gravity approaches towards zero (therefore minimising magnitude M.sub., if the device is dropped vertically) is indicated by a first circle. The free fall ends with the first impact to the ground leading to the acceleration peak. After the drop, the height of the fall can be estimated, using the recorded acceleration values and a standard function analysis algorithm. The algorithm is adapted to detect the begin of the free fall (t.sub.ff; first downward turning point of the acceleration function) and the begin of the impact (t.sub.im, first upward turning point of the acceleration function). The height of the fall can then be calculated using the formula

s (height)=g (gravity)(t.sub.imt.sub.ff).sup.2

[0227] For the estimation of the height of the fall the air resistance is disregarded. In the case shown in FIG. 8 a the time span of the free fall (taken from the diagram) is about 0.33 s leading to an estimated drop height of 1.07 m [9.81 m/s.sup.2(0.33 s).sup.2=1.07 m], which is a good approximation to the actual value of 1.00 m.