FR3100431A1 - Connected sole and associated method of fall detection and gait analysis - Google Patents

Abstract

Connected sole and associated method of fall detection and gait analysis The present technique relates to a removable connected sole comprising a box located at the rear of the sole and comprising a plurality of electronic components arranged to comply with a maximum thickness of the sole. 'rear of the sole of seven millimeters and comprising in particular a fall detection module comprising a microcontroller comprising a processor for: processing temporal variations of the data captured by an acceleration and orientation sensor and a pressure sensor, and deliver processed data relating to the wearer of the sole or the sole, - detect a fall of the wearer of the sole taking into account the processed data, and deliver information relating to the wearer of the sole, - communicate, via a communication module and to a communication entity separate from the sole, the information relating to the wearer of the sole. Figure for the abstract: Fig 1

Description

Connected sole and associated method for detecting falls and analyzing gait

The field of the present technique is that of detecting a person's fall via the use of one or more connected soles worn by this person, in his shoes or slippers. More specifically, the proposed technique concerns a connected sole and the associated fall detection method.

Prior art

There are so-called connected sole techniques integrated into a liner or a shoe, the design of which is carried out globally and integrated with the liner or the shoe. The major drawback of these techniques lies in the fact that these soles are not removable and therefore do not allow optimal use, since a user cannot wear the soles in just any shoes or slippers.

To remedy these drawbacks, there are removable connected insole techniques, such as for example that which is the subject of document WO 2015/189691 which discloses a so-called connected insole, the main drawback of which however lies in the fact that the sole does not is not autonomous and requires a wireless control unit (eg such a mobile) to interact with the user.

Some other connected insole techniques have the main drawback of being very limited in terms of energy autonomy and therefore not very ergonomic.

There is therefore a need for a reliable and ergonomic fall detection technique, through the use of connected insoles that are removable and have increased battery life.

In addition, strong design constraints must be taken into account in the field of connected insoles in order to guarantee user comfort, i.e. maximum thickness, adaptability to all foot sizes/lengths and optimal robustness.

The present technique meets this need by proposing a removable connected sole intended to be inserted into a shoe or slipper of a user, called the wearer of the sole, the sole comprising a box located at the rear of the sole and comprising a plurality of electronic components, the electronic components being arranged to respect a maximum thickness of the back of the sole of seven millimeters and corresponding to:

- at least one long-lasting battery,

- a first pressure sensor, and

- at least one electronic circuit comprising:

- at least one fall detection module for the wearer of the sole,

- at least one acceleration and orientation sensor, and

- at least one communication module,

According to the proposed solution, said at least one fall detection module comprises at least one microcontroller comprising at least one processor for:

- processing data and/or temporal variations of the data captured by said at least one acceleration and orientation sensor and/or said first pressure sensor, and delivering processed data relating to the wearer of the sole and/or to the sole, and/or

- detecting a fall of the wearer of the sole taking into account the processed data, and delivering information relating to the wearer of the sole, and/or

- communicate, via said at least one communication module and to at least one separate communication entity of the sole, the information relating to the wearer of the sole.

Thus, the present technique is based on an entirely new and inventive approach to fall detection via a so-called connected, removable sole, worn by a user in a shoe or slipper, sole in which data processing is implemented. measured by various sensors (such as a pressure sensor and an acceleration and orientation sensor), in order to allow the detection of falls then the communication of information resulting from the detection of falls to a remote entity.

Indeed, the implementation of the processing, in the sole itself, of the sensor data allows an optimization of the energy performance of the connected sole compared to the techniques of the prior art, by reducing the communications (for example the exchanges radio) sensor data to the outside of the sole. This internal treatment in the sole also makes it possible to limit interactions with the wearer of the sole, for better ergonomics, and not to depend on a communication network, or a remote device, to perform data processing.

In addition, the connected sole according to the present technique has better autonomy thanks to the insertion of a long-lasting battery, this insertion being made possible thanks to an optimized arrangement of all the electronic components inside the connected sole, while respecting the thickness constraints of the sole, for optimal comfort of the user wearing this sole.

Furthermore, the positioning, at the rear of the sole, of the box comprising the majority of the electronic components of the sole, makes it possible to overcome the contrary designs of soles of the prior art according to which all the technical elements is located in a restricted space under the arch of the foot. The fact that the battery necessary for the present technique is a long life battery occupying a larger space than the batteries conventionally used has compelled the inventors to design a completely different optimum arrangement for implementing the present technique. This arrangement is partly based, as described below, on a superposition of elements rather than a juxtaposition.

Finally, the fall detection is also optimized thanks to the processing not only of the data captured by the various sensors but also of the temporal variations of these sensor data. Indeed, taking into account variations of the sensor data makes it possible in particular to more reliably detect changes in position of the wearer of the sole, and thus to greatly limit false detections of falls.

Fall detection finds applications in particular in the accompaniment of the elderly or in convalescence, at home or in specialized institutions, in the post-operative follow-up of patients, but also in the prevention of occupational risks.

According to a particular aspect, the connected sole also comprises a second pressure sensor located at the front of the sole, the processor further processing data captured by the second pressure sensor to deliver the processed data.

Thus, according to this embodiment, the use of a second pressure sensor, located at the front of the sole, in addition to the first pressure sensor located at the rear of the sole, makes it possible to further optimize the fall detection, thanks to better precision in defining the instantaneous state (or position) of the sole wearer and therefore position variations. Indeed, thanks to these two pressure sensors, this technique makes it possible to measure the pressure exerted by the foot of the wearer of the sole on the two main points of support: at the level of the heel (first pressure sensor) and at the level metatarsus (second pressure sensor).

Additionally, the data from the first and second sensors can be correlated.

Furthermore, according to a particular aspect of the proposed solution, the electronic components further include:

- at least one communication antenna (long range), and

the at least one electronic circuit further comprises:

- at least one battery charging circuit (USB),

- at least one electronic component supply circuit,

- a rear connection cable, and

the sole further comprises, on at least a part:

- a lower support layer on at least a part of the sole separate from the case,

- a front connection cable,

- an upper comfort layer covering at least part of the electronic components and part of the case,

- a cover layer covering at least part of the upper comfort layer.

The present technique also relates to a method for detecting a fall of a wearer of a connected sole as described above, according to the different embodiments, the method comprising the following steps:

- obtaining at least one capture data from the first pressure sensor and/or from the second pressure sensor and/or from the acceleration and orientation sensor,

- processing of the data and/or temporal variations of the data obtained, delivering processed data relating to the wearer of the sole and/or to the sole,

- detection of a fall of the wearer of the sole taking into account the processed data, delivering information relating to the wearer of the sole,

- Communication, via said at least one communication module of the sole and to at least one separate communication entity of the sole, information relating to the wearer of the sole.

Thus, the present technique is based on an entirely new and inventive approach to fall detection via a so-called connected, removable sole, worn by a user in a shoe or slipper, thanks to the implementation, in the sole and no longer in a remote entity as according to the techniques of the prior art, of the processing of the data captured by various sensors (such as a pressure sensor and an accelerometer), in order to allow the detection of a fall and then the communication of information from fall detection to a remote entity. "Information from fall detection" means not only an indication that a fall has been detected but also contextual information, i.e. indicative of the context of the fall (such as the time of the fall, the direction of the fall, the supposed presence of an obstacle, etc.). This contextual information can thus for example be displayed by the remote entity, in parallel with the transmission of a fall alert.

Indeed, the implementation of the processing, in the sole itself, of the sensor data allows an optimization of the energy performance of the connected sole compared to the techniques of the prior art, by reducing the communications (for example the exchanges radio) sensor data to the outside of the sole. This internal treatment in the sole also makes it possible to limit interactions with the wearer of the sole, for better ergonomics, and not to depend on a communication network, or a remote device, to perform data processing.

In one particular aspect, the processing step includes the following sub-steps:

- filtering of the data obtained, delivering filtered data,

- determination of a current state of the wearer of the sole and/or of the sole as a function of the filtered data, and the fall detection step takes into account the determined current state and at least one previously determined current state .

Thus, according to this embodiment, the fall detection implements a determination of a current state of the wearer of the sole, for example "lying down", "sitting", "standing", "walking" or of the sole (e.g. "unworn"), so as to detect a fall by comparison between a current state and one or more previously determined states. It is therefore variations between successive current states, determined from data captured by pressure sensors and an accelerometer, which make it possible to detect a fall, unlike certain techniques of the prior art which are based solely on detection, via an accelerometer, a loss of verticality followed by the detection of an absence of movement.

In addition, the data from the various sensors / accelerometer are filtered, for example via signal processing algorithms, so as to remove noise, to standardize the data captured and therefore to make them usable for the implementation of the fall detection algorithm.

According to a particular feature, the method further comprises:

- a learning phase of the first and/or second sensors for the wearer of the sole delivering ranges of values corresponding to the different possible current states of the wearer of the sole and/or of the sole,

- a step of updating at least one threshold separating two of said possible current states of the sole wearer and/or of the sole, said updating taking account of the ranges of values delivered and delivering at least one detection threshold of fall.

Thus, according to this embodiment, a sensor learning phase makes it possible, for each sole wearer, to identify the states of the wearer of the sole or of the sole itself (for example when it is not used) according to the data measured by the sensors. This involves, for example, differentiating the "sitting" state from the "standing" state by analyzing the pressure values measured by the sensors located in the sole. Ranges of values are therefore thus determined to each represent a state of the wearer or of the sole.

This continuous learning in real time makes it possible to take into account any physical changes to the wearer of the sole, such as weight gain or loss, a change in balance or gait due to an injury or fatigue (the wearer limps or walks more slowly …) over a more or less long time, without requiring interaction with the user.

The update of at least one threshold then makes it possible to define the various thresholds making it possible to segment the ranges of values previously determined by the learning phase, and in particular the fall detection threshold.

Updating at least one threshold means both fixing a first threshold value whose default value is 0 for example, then updating, continuously or periodically, the value of the threshold according to the activity of the wearer of the sole.

For example, the step of determining a current state of the wearer of the sole and/or of the sole includes a sub-step of analyzing and correlating the filtered data.

Thus, according to this embodiment, the data of the two pressure sensors and those of the accelerometer, as well as the temporal variations of these sensor data are analyzed and correlated so as to be able to determine a current state of the wearer of the sole, or the sole itself, as reliable as possible. In this way, detection of a sudden change in current state makes it possible to detect a drop.

According to a particular aspect, the method also comprises a step of transmitting, via said at least one communication module of the sole and to at least one communication entity distinct from the sole, processed data relating to the wearer of the sole and/or at the sole.

Thus, according to this embodiment, the usable data from the sensors, after filtering, are also transmitted to a remote entity, even when no fall is detected, so as to be able to monitor the activity of the wearer of the sole (number of steps, pressure level, time spent in each state standing, sitting, lying, walking, walking speed, step frequency, division of step phases, etc.). This therefore makes it possible to initiate, if necessary, fall prevention actions: modification of the environment of the insole wearer, for example in the context of home support for an elderly person, by removing potential obstacles, help walking (crutches, canes, walkers, etc.), …. For example, the remote entity can make comparisons of this data on the wearer's activity over long periods (several weeks), in order to precisely implement these fall prevention actions.

According to a particular characteristic of the solution, the steps of obtaining and/or processing and/or detecting and/or filtering and/or determining a current state and/or updating a threshold and /or transmission and/or said learning phase are implemented periodically and/or as a function of a predetermined criterion and/or of a variation of a predetermined criterion.

Thus, according to this embodiment, the sensor measurements are obtained periodically, preferably continuously, so as to also continuously determine a current state of the wearer of the sole and therefore to be able to detect a fall reliably.

The frequency of implementation of these steps can be chosen according to different criteria which may depend, for example, on the wearer of the sole (person with low or high risk of falling, low or high activity, etc.) captured data.

According to a particular aspect, the method further comprises the following steps, implemented within the separate communication entity:

- reception of information relating to the wearer of the sole and/or to the sole, called first wearer information for a first sole;

- search for information relating to the wearer of a second sole distinct from the first sole and previously matched to the first sole and/or the second sole, called second wearer information;

- correlation of the first and second wearer information, delivering a positive or negative fall detection decision,

- generation of an alert when the fall detection decision is positive.

Thus, according to this embodiment, the remote entity intended to generate an alert in the event of detection of a fall of a sole wearer uses the information communicated by the two soles of the same wearer in order to improve the reliability fall detection. Indeed, the correlation of these two sources of information from each of the soles makes it possible to remove the doubt when the information from a sole is not considered reliable enough to generate an alert for example, thus reducing the number of false detections. .

Taking into account the information received from the two soles also makes it possible to obtain contextual information relating to the wearer, and thus for example to obtain indications on the left/right balance of the wearer.

When an alert is generated, it is then processed in such a way as to be broadcast/transmitted to other predetermined entities, such as for example one or more mobile telephones of one or more previously identified persons (for example the members of staff supervising the wearer of the sole in an application to support an elderly person in a specialized establishment, one or more family members in an application to support an elderly person at home, etc.).

The present technique also relates to a computer program product comprising program code instructions for implementing a fall detection method as described above, when it is executed by a processor.

Presentation of figures

Other aims, characteristics and advantages of the proposed technique will appear more clearly on reading the following description, given by way of a simple illustrative example, and not limiting, in relation to the figures, among which:

illustrates an example of a connected insole presenting the main elements making it up, seen from above, according to one embodiment of the proposed technique;

illustrates an example of an electronic circuit included in a connected sole as illustrated in FIG. 1;

illustrates a side view of the connected sole example as illustrated in FIG. 1;

illustrates a side view of another example of a connected sole;

presents the main steps of a fall detection method, according to an embodiment of the proposed technique.

Detailed Description of Embodiments of the Invention

The general principle of the proposed technique is based both on an optimized design of a connected sole and on an optimized method for detecting a fall of a user wearing such a sole, so as to provide a sole whose autonomy in energy is greatly increased compared to the techniques of the prior art and a fall detection of increased reliability also.

The field of application of the proposed technique is that of fall detection and monitoring of a person's walking activity, for example in the context of home support for elderly people, support for elderly people in specialized establishment (accommodation establishment for dependent elderly people (or EHPAD), senior residence, autonomy residence, hospital) and follow-up for the staff of these establishments, post-operative follow-up of a patient by health professionals, or also the prevention of risks at work (such as musculoskeletal disorders linked to physical activities such as walking or carrying loads, for example in logistics).

To do this, the present technique proposes a method for detecting the fall of a wearer of connected sole(s), a method embedded in the sole(s) themselves, it that is to say in which all the treatments are implemented in the sole itself so as to reduce the energy consumption of the sole by greatly limiting, or even eliminating, data communications to the outside (which are frequent according to the techniques of the prior art).

Furthermore, the fall detection method is based on observing the current state of the wearer of the sole and on a change between two successive current states.

As described below, the fall detection method according to the present technique also takes advantage of the use of two connected soles for the same wearer, so as to be able to correlate the information resulting from the fall detection of each of the soles and provide even more reliable fall detection.

In addition, the optimization of the energy autonomy of the sole is based, in addition to the reduction of exchanges, also on the choice of the battery itself, in this case a long-lasting battery, unlike the techniques of the prior art.

However, the choice of a long-lasting battery generates architectural constraints of the sole itself, because such a battery occupies more space than a battery conventionally used according to the prior art, the constraints of comfort, size and especially the thickness of the sole remaining for their part of the same level of requirement.

Thus, the proposed technique is also based on a particularly and advantageously optimized arrangement of the various electronic components present in a connected sole in order to accommodate such a long-lasting battery while limiting the thickness of the sole.

A removable connected insole will now be described in detail, according to an embodiment of the present technique, in relation to FIGS. 1 to 3 (a and b).

Fig 1 illustrates an example of a removable connected sole 1, according to an embodiment of the present technique, intended to be inserted into a shoe, a liner, a boot, etc. of a user, subsequently called wearer, in order to detect a possible fall of this wearer, but also to follow up on his walking/resting activity.

According to this example, the sole 1 comprises on the one hand a box 10, located at the rear of the sole, embedding a certain number of electronic components necessary on the one hand for the fall detection itself (sensors, processor of calculation, etc.) and on the other hand to the operation of the sole (battery, connectors, communication module to the outside).

As already indicated above, the choice of the battery is guided by the need for a sole with increased autonomy in terms of energy, therefore a long-lasting battery 101, for example able to operate for six months, compared to the batteries autonomy limited to a few weeks used currently.

The box 10 also includes a first pressure sensor 1020, the measurements of which are intended to be used in the fall detection method described below.

Finally, the box 10 also includes an electronic circuit 103 (or PCB for "Printed Circuit Board" in English) comprising:

- a fall detection module for the wearer of the sole, comprising at least one processor U6 for implementing the fall detection method,

- an acceleration and orientation sensor 1031, the measurements of which are intended to be used in the fall detection method described below, and

- a communication module 1032, comprising in particular a long-range flexible antenna 104 connected to the electronic circuit 103 (positioned near the edge of the case 10 for radiation reasons), this communication module 1032 allowing the feedback of information relating to the wearer of the sole 1 (fall detected or monitoring of its activity) to a predetermined remote entity.

According to this embodiment of the proposed technique, the sole 1 also comprises a second pressure sensor 1021 located at the front of the sole, the measurements of which are also intended to be used in the fall detection method described below. . This second pressure sensor 1021 is connected to the electronic circuit 103 by a front connection layer 105, both to power the sensor and to conduct the data captured to the processor executing the fall detection method.

Thus, the presence of two pressure sensors makes it possible to increase the reliability of fall detection, as well as the monitoring of the wearer's activity, by measuring the pressure exerted by the foot on the two support points which are the heel and metatarsus.

The measurements of these two pressure sensors 1020 and 1021, linked to those measured with the acceleration and orientation sensor 1031, are processed and analyzed and provide a fusion of data to an on-board fall detection algorithm, described below. After. This data fusion allows to locate the state of the wearer in the sole ("sitting", "standing", "walking", "lying" ...) or the state of the sole ("used", "not used" ), this state then making it possible to detect a fall.

Furthermore, the electronic circuit 103 also comprises a rear connection layer 107 making it possible to position and connect the battery 101, as well as connector elements to allow the analysis of the programs embedded in the processor executing the fall detection method in particular. These connection elements are also provided to allow recharging of the battery, at the rear of the sole 1, via a charging circuit 106. For example, these connection elements are in the form of USB communication means.

There is also provided in the electronic circuit 103 a supply circuit for all the electronic components.

As already indicated, one of the challenges with which the inventors were confronted was to arrange the various aforementioned components in an optimal manner in order to be able to respect the constraints of size, and in particular of thickness, of the finalized sole, that is to say thicknesses of acceptable finished products, while offering a comfortable sole. These acceptable thicknesses can be different at the front and the back of the sole, as well as at the level of the arch of the foot. Thus, a maximum thickness of seven millimeters must be respected at the back of the foot, while the thickness of the front of the sole can be much thinner, for example four millimeters maximum.

Furthermore, due to the energy saving constraint imposed to obtain increased autonomy of the sole, the fall detection method is more complex than those of the state of the art, because it implements, in the sole, all the processing necessary for the detection of a fall whereas the techniques of the prior art are content to obtain the sensor data and to transmit it to a remote entity, before any processing, so that this remote entity executes the fall detection method.

This greater complexity of the fall detection process therefore generates the need to integrate a larger long-lasting battery, thus posing a technical challenge compared to the existing one in terms of the arrangement of electronic components in the sole. The state of the art in the design of connected insoles generally positions all the technical elements, in a superimposed manner, under the arch of the foot, in a small space, which cannot accommodate a battery as used in the present technical. The space under the arch of the foot also cannot accommodate the long-range flexible antenna necessary for the communication of information relating to the sole or the wearer.

In the case of this technique, the antenna and the battery cannot enter this space without causing an increase in the thickness of the sole at the level of the arch of the foot (up to 9 millimeters) which is unbearable for the wearer. . It was therefore necessary to rethink the layout of the technical elements to insert the long-life battery, insert the electronic circuit and the various capture/connection layers, protect most of these elements in a robust plastic case, while maintaining low thicknesses of the order of four to five millimeters to also allow the addition of comfort and sole covering materials.

To do this, and as illustrated more particularly in Fig 2 and 3a / 3b, the various technical elements making up the sole are partly superimposed at the rear of the sole, where the box comprising most of the components is therefore mainly located. electronics.

Fig 2 illustrates the arrangement of the various electronic components making up the electronic circuit 103, and in particular the processor U6, the acceleration and orientation sensor 1031 and the communication module 1032 not shown in Fig 1.

Fig 3a, on the other hand, better illustrates the superimposition of the various electronic components and other layers of comfort, support and protection essential to obtaining a comfortable removable connected insole.

Thus, a covering layer A is provided which, according to the embodiment illustrated in FIG. 3a, covers the upper surface of the entire sole, and has a relatively small thickness. There is also an upper comfort layer B which also, according to the embodiment illustrated in Fig 3a, covers the upper surface of the entire sole, and has a greater thickness than the covering layer A.

At the level of the front of the sole, which notably comprises the second pressure sensor 1021 and the front connection layer 105 connecting it to the box 10, which each have a relatively small thickness C, there is provided a lower support layer D , of thickness which may be substantially equivalent to or less than that of the upper comfort layer B.

As regards the rear of the sole, which includes the main electronic components embedded in a box 10, a thickness substantially equivalent to that of the lower support layer D present at the front of the sole is occupied by the box 10 and the flexible antenna 104.

For example, the dimensions of the components implemented in a connected sole according to a particular embodiment of this technique are as follows:

- long-lasting battery 101: about 1 millimeter thick,

- first and second pressure sensors 1020, 1021: 30 millimeters in diameter and 0.2 millimeters thick,

- electronic circuit 103: 30*20*2 millimeters,

- flexible antenna 104: 80*20*0.2 millimeters,

- flexible rear connection cable: 30*77*0.2 millimeters,

- flexible front connection cable: 30*130*0.2 millimeters,

- cover layer A: 0.5 millimeters thick,

- upper comfort layer B (e.g. EVA-type foam): 3.5 millimeters thick,

- lower supporting layer D (e.g. EVA-type foam): between 1 millimeter and 3 millimeters thick.

These dimensions correspond to an example of the composition of a connected sole according to an embodiment of the proposed technique, given by way of illustration only.

Fig 3b illustrates for example another embodiment in which the sole is thinner at the front than at the rear, for better comfort, thus taking advantage of the fact that the front of the sole has few components. , and components having relatively small thicknesses.

We will now describe the method for detecting a fall of a wearer of a connected sole as described previously, in relation to Fig 4.

As already indicated above, the fall detection method for the wearer of a connected sole is based on the measurements captured by the various sensors present in the sole (two pressure sensors and an acceleration and orientation sensor according to the embodiment described above). To do this, a first step of obtaining 21 at least one capture datum from the first pressure sensor D _CP1 and/or from the second pressure sensor D _CP2 and/or from the acceleration and orientation sensor D _ACC is implemented.

These "raw" data, directly captured by the sensors, or temporal variations of these data, are then processed, during a processing step 22 to deliver processed data D _processed relating to the wearer of the sole and/or to the sole, such as the state of the sole ("used", "not used") or the current state of the wearer ("sitting", "lying", "walking", "standing" ...) This is detailed below.

Thanks to these processed data D _processed relating to the wearer of the sole and/or to the sole, the method then implements a step 23 of detecting a fall of the wearer of the sole taking into account the processed data to deliver information ( I _U ) relating to the wearer of the sole, which information ( I _U ) is transmitted, during a communication step 24, for example via the communication module of the sole, to at least one separate communication entity of the sole .

According to the application of this technique, the remote entity corresponds for example to a remote server, which then retransmits an alert to a smartphone of a previously identified person, such as supervisory staff in a specialized establishment for the elderly or injured or recovering, or a relative of an elderly person...

In addition, during this detection step 23, the information ( I _U ) is analyzed (therefore in the sole), so as to allow in particular:

- to obtain contextual information to determine the causes of the detected fall,

- to follow the walking activity (number of steps, level of pressure, time spent standing, walking, sitting, lying down) of the wearer of the insole.

As indicated, the raw data measured by the sensors (pressure or acceleration and orientation) must be processed in order to be usable to detect a fall or to monitor the wearer's activity and in particular be filtered in order to to remove the noise. It is a question here of applying known signal processing techniques, which can be chosen for example according to the type of sensor or the precision required to be able to exploit these sensor data. This filtering can also be followed by data normalization, in particular with a view to correlating data from different sensors and analyzing data variations.

Thus, the objective being to be able to determine the current state of the wearer of the sole, or of the sole itself, the processing of sensor data must take into account not only the data captured but temporal variations of these data.

Indeed, the state of the person is very important to achieve reliable detection of the fall of an insole wearer. Thus, at a given moment, it is very difficult to differentiate a person on the ground following a fall from a person lying in a bed or an armchair. On the other hand, it is well understood that a person in a fall does not pass through the "sitting" position before being in the "lying" position but goes from a "standing" state to a "lying" state in a very short time. short.

This is why the step of determining a current state of the wearer of the sole (or of determining the current state of the sole) comprises a sub-step of analyzing and correlating the filtered data, to define a succession, over time, of the current states of the wearer (or of the sole) and detect any non-conformity in this succession, such as the sudden transition from a current "upright" state to a current "lying down" state, without going through a common "sitting" state. This solution thus makes it possible to make fall detection more reliable compared to the techniques of the prior art which are based solely on data from an accelerometer, detecting a fall by detecting a loss of verticality followed by an absence of movement, a situation that can however arise when a person sits down and not only when they fall.

The fall detection step of the present technique therefore takes into account the current state of the wearer (or of the sole) and at least one previously determined current state (of the wearer or of the sole). Thus, for example, a fall is detected when the states determined, for the wearer of the sole, over a period of a few seconds preceding the current moment, corresponded to a "standing" state and the current state corresponds, suddenly therefore, to a "lying down" state.

However, this simple detection algorithm presents a technical problem related to use and in particular to the precision necessary to determine a current state reliably. Indeed, it is necessary to be able to segment the analysis results of the data measured by the sensors, whether they are the filtered data or the correlations of filtered data coming from several different sensors, or even the temporal variations of such data, in order to define distinct, easily identifiable "regions" of activity corresponding to the various possible current states. In addition, it is also necessary to be able to precisely define thresholds making it possible to distinguish one state from another state, and in particular the threshold making it possible to decide on the "lying down" state of the wearer, this state most often being that resulting from a fall.

To do this, it is necessary to learn the sensors used in the sole, in order to differentiate the pressure levels measured, for example between a "sitting" state and a "standing" state. This learning is of course necessary per insole wearer, and for the two insoles used, the sensor measurements observed not being similar for each foot of a person. In addition, the pressure levels in particular (measured by the pressure sensors) may vary over time, depending on elements specific to the wearer, such as the way he walks which may vary according to his fatigue, an injury, any discomfort, etc., his balance, his weight… or even external elements that can impact walking (ground with obstacles, slippery ground, etc.).

Thus, the present technique proposes a learning phase of the various sensors implemented in the sole, in order to deliver ranges of values corresponding to the various possible current states of the sole wearer and/or of the sole.

In addition, according to the proposed technique, this learning phase is advantageously implemented continuously when a sole is used, that is to say it allows continuous adaptation of the criteria making it possible to detect a fall, such as the thresholds making it possible to distinguish between two states of the wearer, in order to provide the best possible level of reliability of the fall detection method, whatever the possible changes in behavior of the wearer of the sole, or even the possible external conditions aforementioned.

Thus, the proposed technique therefore also provides for a step of updating at least one threshold separating two possible current states of the sole wearer and/or of the sole, this updating taking account of the ranges of values delivered by the phase learning, and delivering at least one fall detection threshold.

This step of updating the thresholds separating different current states first makes it possible to set a first value of these thresholds (which can for example be set to a default value when the sole is not used), for example when the wearer has just put on his shoes in which connected insoles are inserted and begins to walk. The update then makes it possible to modify this first value, as the insole is used, and therefore as the various measurements obtained from the various sensors progress.

The various steps previously described of the fall detection method (obtaining sensor data, processing the data obtained, filtering the sensor data, determining a current state, updating a threshold, falling detection) are advantageously implemented periodically and/or according to a predetermined criterion and/or a variation of a predetermined criterion. The learning phase is also implemented periodically and/or according to a predetermined criterion and/or a variation of a predetermined criterion.

Thus, the present technique therefore provides for obtaining data from sensors on a regular basis, for example according to a predetermined frequency (ranging from a few times to several hundred times per second, etc.), which can of course be different depending on the use cases ( according to the wearer, according to pre-identified fall risks, etc.).

Similarly, the fall detection itself can be implemented periodically, the frequency of its implementation being able for example to depend on the data of the sensors: thus, if the analyzed data of the sensors presents anomalies, for example very large variations in pressure over time, the fall detection method can adapt its calculations so as to perform all the steps necessary for detecting a fall more frequently. On the contrary, when a "stable" activity of the wearer is detected, for example a current "sitting" or "lying" state (not following a fall) of long duration, the fall detection method can perform the steps less often necessary for the detection of a fall, for example in order to save energy and to sustain the autonomy of the sole.

Thus, the method according to the proposed technique allows a very great adaptability of the fall detection of a connected insole wearer, by continuously adjusting the different thresholds making it possible to segment the wearer's activity into different states.

As already indicated, one of the advantages of this technique lies in the use of two soles for the same wearer, and therefore in taking into account the fall detection carried out in each of the two soles. This taking into account is carried out at the level of the distinct and remote entity, to which the information relating to the wearer or the sole is communicated, for each of the soles worn. As already indicated, such an entity corresponds for example to a remote server, a computer or a smartphone, intended to receive data from the connected soles indicating that a fall has been detected (after analysis, in the sole(s), of the sensor data) and then issue an alert.

To do this, this distinct entity searches for information relating to a wearer received from two distinct soles, denoted first and second soles, or left sole and right sole, previously paired, this information being for example denoted first and second wearer information. Then, a step of correlating the first and second wearer information is implemented to deliver a positive or negative fall detection decision.

For example, when the information from the left and right soles of a wearer indicate that a fall has been detected, then the fall detection decision is positive and an alert is generated by this separate entity. This alert is for example then transmitted to one or more previously identified recipients.

On the other hand, when the information from the left and right soles of a wearer are not consistent, or appear contradictory, i.e. when one sole has detected a fall and the other sole has not, the separate entity must take a decision to fall detection or not.

For example, this detection decision can take into account predetermined criteria, such as for example weighting elements applied to the results from the two soles, these weighting elements being from a prior analysis of the wearer's activity (such as for example monitoring of its left/right balance).

Or, this detection decision can be made if the information received from the sole that detected a fall is itself above a previously determined threshold.

The fall detection decision can also be deferred and taken only when several successive pieces of information, over a predetermined period, coming from the same sole correspond to a fall detection, even if, over this same predetermined period, the other sole does not detect no fall.

Furthermore, another advantage of the use of two connected insoles lies in the monitoring of the wearer's activity, such as for example the monitoring of the left/right balance, in particular in the context of rehabilitation or prevention, thus making it possible to correct the attitude of the wearer of the soles.

To do this, the fall detection method therefore also comprises a step of transmitting, still via the communication modules of the soles, and to at least one communication entity distinct from the soles, the processed data relating to the wearer and/or to the sole. This transmission step can advantageously be implemented periodically and/or according to a predetermined criterion and/or a variation of a predetermined criterion, as already described below for the other steps of the method.

This transmission step makes it possible in particular to transmit information relating to the wearer of the soles, excluding detection of a fall, in order to have information for monitoring his activity. This information is useful in applications for monitoring walking activity, for example in post-operative or post-traumatic monitoring of a patient.

For example, if the information received from activity monitoring shows a left/right imbalance of the wearer of the insoles, this can make it possible to implement fall prevention actions.

Thus, the proposed solution makes it possible to provide a removable connected sole with great energy autonomy and total independence (in particular eliminating interactions with the wearer of the sole) and a highly reliable fall detection method because it is continuously adaptive to any changes in the behavior of the wearer and taking into account the data provided by the two soles worn.

Claims (10)

Sole (1) connected removable intended to be inserted into a shoe or slipper of a user, said wearer of said sole, characterized in that it comprises a box (10) located at the rear of said sole and comprising a plurality of electronic components, said electronic components being arranged to respect a maximum thickness of the rear of said sole of seven millimeters and corresponding to:
- at least one long-life battery (101),
- a first pressure sensor (1020), and
- at least one electronic circuit (103) comprising:
- at least one fall detection module of said wearer of said sole,
- at least one acceleration and orientation sensor (1031), and
- at least one communication module (1032),
said at least one fall detection module, comprising at least one microcontroller comprising at least one processor (U6) for:
- processing data and/or temporal variations of the data captured by said at least one acceleration and orientation sensor and/or said first pressure sensor, and delivering processed data relating to said wearer of said sole and/or to said sole, and/or
- detecting a fall of said wearer of said sole taking into account said processed data, and delivering information relating to said wearer of said sole, and/or
- communicating, via said at least one communication module and to at least one separate communication entity of said sole, said information relating to said wearer of said sole. Connected sole according to claim 1, further comprising a second pressure sensor (1021) located at the front of said sole, said processor further processing data sensed by said second pressure sensor to output said processed data. Method for detecting a fall of a wearer of a connected sole according to claims 1 or 2, characterized in that it comprises the following steps:
- Obtaining (21) at least one capture datum of said first pressure sensor ( D _CP1 ) and/or of said second pressure sensor ( D _CP2 ) and/or of said acceleration and orientation sensor ( D _ACC ) ,
- processing (22) of said data and/or temporal variations of said data obtained, delivering processed data ( D _processed ) relating to said wearer of said sole and/or to said sole,
- detection (23) of a fall of said wearer of said sole taking into account said processed data, delivering information ( I _U ) relating to said wearer of said sole,
- Communication (24), via said at least one communication module of said sole and to at least one separate communication entity of said sole, of said information ( I _U ) relating to said wearer of said sole. Fall detection method according to claim 3, characterized in that said processing step comprises the following sub-steps:
- filtering said data obtained, delivering filtered data,
- determination of a current state of said wearer of said sole and/or of said sole according to said filtered data,
and in that said drop detection step takes into account said determined current state and at least one previously determined current state. Fall detection method according to claim 3 or claim 4, characterized in that it further comprises:
- a learning phase of said first and/or second sensors for said wearer of said sole delivering ranges of values corresponding to the various possible current states of said sole wearer and/or of said sole,
- a step of updating at least one threshold separating two of said possible current states of said sole wearer and/or of said sole, said updating taking account of said ranges of values delivered and delivering at least one detection threshold of fall. Fall detection method according to claim 4 or claim 5, characterized in that said step of determining a current state of said wearer of said sole and/or of said sole comprises a sub-step of analyzing and correlating said filtered data. Method for detecting a fall according to any one of Claims 3 to 6, characterized in that it also comprises a step of transmitting, via the said at least one communication module of the said sole and to at least one communication entity distinct from the said sole , of said processed data relating to said wearer of said sole and/or to said sole. Fall detection method according to any one of Claims 3 to 7, characterized in that the said steps of obtaining and/or processing and/or detecting and/or filtering and/or determining a current state and/or updating of a threshold and/or transmission and/or said learning phase are implemented periodically and/or as a function of a predetermined criterion and/or of a variation of a criterion predetermined. Fall detection method according to any one of Claims 3 to 8, characterized in that it further comprises the following steps, implemented within said separate communication entity:
- reception of said information relating to said wearer of said sole and/or to said sole, referred to as first wearer information for a first sole;
- search for information relating to said wearer of a second sole distinct from said first sole and previously matched to said first sole and/or said second sole, said second wearer information;
- correlation of said first and second wearer information, delivering a positive or negative fall detection decision,
- Generation of an alert when said fall detection decision is positive. Computer program product comprising program code instructions for implementing a fall detection method according to any one of claims 3 to 9, when executed by a processor.