Field of the Art

The present invention relates in a first aspect to a physical exercise correctness calculation method, wherein said physical exercise is performed by a user connected or close to electronic sensing devices which collect measurements of different variables related to performing said physical exercise, and more specifically to a method which comprises supervising said exercise remotely by means of processing at least part of said variables in a computation device to which said electronic sensing devices are connected, obtaining said correctness calculation depending on the distance measured between at least one of said variables and previously established correctness values, and transmitting at least said correctness calculation to remote equipment for supervision.

A second aspect of the invention relates to a system provided for implementing the method of the first aspect.

Prior State of the Art

Performing physical and mental exercises is common in rehabilitation therapies, training plans, maintaining physical fitness, wellness, etc... Therefore, the support provided by new information and communications technologies has grown exponentially in recent years. Many companies are already applying mobile devices and electronic biometric sensors for performing and controlling exercise sessions, performing remote patient follow-up and physical activity and sports remote control.

Most of these systems base their results on achieving goals, such as: running a certain distance, losing a certain number of kilos, or performing exercise at a certain time. In all these cases, said goal is used to non-subjectively evaluate the result of the session, indicating either a success percentage or a score. Using this value as the only assessment method can entail a serious mistake because other parameters that could directly affect correctly performing the exercise and therefore the patient's health are overlooked. Said parameters could be posture, axis of motion, strength, speed, time, and ignoring them could lead to serious injuries, performing therapy wrong, lack of improvement in rehabilitation therapies, onset or increase of pain and other problems derived from poorly following up on the exercise while being performed. This consideration is even more relevant when monitoring people performing the exercise remotely and/or without supervision.

A system which can assure that the exercise or training is correctly performed, regardless of meeting the established goal, is therefore needed. Furthermore, said correctness must be supervised by a doctor or specialist who can make decisions, adjust parameters or talk with the user for future sessions.

Cases of systems and applications with goal-oriented rehabilitation exercises: - Nintendo Wii with WiiFit [1]: The system is a videogame console with a wireless sensor, in this case a balance board with pressure sensors, where the goal is to perform a set of iterations for an exercise the instructions of which are shown in graphic animation on a television screen or display. The time used in performing the exercises, the number of iterations and the target value achieved are measured, but not the correctness of the exercise.

- GestureTek [2]: The system is based on motion recognition with cameras for performing physical and mental rehabilitation exercises. Games are played using motions of the body in front of one or more cameras connected to a computer. The image is analyzed and biometric motion data of the user can be obtained.

- Nintendo Wii with Stationary Bike [3]: This is a Nintendo system including the

Wii console and a stationary bike for performing exercises while playing with the console.

External references relating to home rehabilitation exercise control, correctness of the exercises and specific training for performing therapies correctly include:

- The effects of mode of exercise instruction on correctness of home exercise performance and adherence [4]: this is an article about the need to receive training or instruction prior to and while performing home exercises, and how said training affects correctly performing exercises. It proposes systems such as courses, home videos, books, guidelines, etc... This article expresses the need to control correctness when performing exercises, but once at home, there is no automatic or controlled supervision.

- Method of exercise prescription and evaluation [5]: This is a patent relating to the automatic supervised prescription of exercise protocols depending on the subject data. This patent proposes a system for, according to the subject data, performing a prescription of exercise protocols in accordance with the subject's needs. It is therefore an automatic prescription of goals which are later modifiable by a supervisor, and displayed to the client in the device that will be used. It does not approach exercise correctness evaluation.

- Identifying Exercise Correctness for Home-Based [6]: This article acknowledges the need to control correctness in performing home exercises in rehabilitation patients, training, etc... in real time while performing the exercise. To that end it proposes an intelligent system based on neural networks which, by means of prior training of the system, indicates if the exercise is being performed correctly while performing it.

Other patents relating to physical exercise analysis include:

- US6210301 : This is a patent for comparing the sensor signals by means of mathematical or statistical signal analysis (neural networks for example). In this case the original or processed signal compared with a previously calculated signal representing the ideal signal values is used.

- US006007459A: This invention proposes a system for communicating to a physical therapy exercise patient with his/her doctor or physical therapist by means of audio, video or data, such that the doctor can supervise said values and report the possible improvements or changes to the patient in real time or at a later time.

- US6682351 : This patent proposes something virtually identical to US6210301 . This time, the patent relates to a specific unit responsible for evaluating the input data of the sensors and it again proposes neural networks or fuzzy logic as calculation methods. Neural networks have the drawback of needing specific training for each exercise and of obtaining non-deterministic results depending on the input values. They furthermore do not provide precise feedback for the patient, rather one or several approximation values of the exercise with respect to the ideal.

- EP1 195139 proposes a specific method for checking back posture.

- US200300271 18 also proposes a system/method for controlling exercise during rehabilitation. It is based on sensor data acquisition, without the need for supervision. It is fundamentally centered on acquiring data and comparing it with predefined thresholds.

- WO2008010131 proposes a method for calculating the 3D position of the exercise from three sensors located in two jointed parts of the body (elbow, knee, etc.). It is an invention aimed at the graphic representation and calculation of positions and angles of parts of the body in real time.

Current solutions which are implemented in manufacturers, developers, service providers and telephone operators base their results on achieving goals that can be measured by means of sensors, electronic devices, surveys completed by the user or the observation of a specialist. Said goals, such as distance, speed, time, number of repetitions, etc... are a non-subjective manner of evaluating the result of an exercise, session or therapy. This manner of evaluation seeks to inform the user about whether or not he/she has met the goal but not about whether or not he/she has achieved the goal "correctly". In this respect therapists agree that in many cases performing the exercise incorrectly is more harmful than beneficial, regardless of if the therapeutic goal is met.

The problem with said solutions is that they do not evaluate the correctness in performing the exercise while performing it. Therefore the user does not know if he/she is performing the exercise correctly. Furthermore, they are usually autonomous systems, applications which are run locally in the home or in the terminal of the client, without the supervision of a doctor, physical therapist or specialist. There are some methods which are currently used to know if the exercise is being or has been performed correctly.

In some cases, said correction is performed at a later time, and it is frequently determined by means of directly surveying the user, but in this case the problem may have already presented itself in the form of an injury or increased pain. Furthermore, when said evaluation is performed by means of user tests and/or questionnaires, a subjectivity variable is added which prevents truthfully knowing correct exercise performance.

In addition, in the existing solutions for correct exercise performance an initial explanation or instruction of the methodology to follow is given to the user, he/she is given explanatory videos to watch while performing it, or he/she is given a questionnaire at the end of the session, which does not assure correct exercise performance. Nevertheless, these actions are compatible with and complementary to the proposed invention.

It is necessary therefore to have a system for evaluating exercise correctness while performing it or at the end thereof, which displays information in real time or at the end to the user and/or to an external supervisor which allows the correction by the user at the time of performing the exercise or in subsequent sessions. This concept is illustrated below.

In an exercise the goal of which is to lift a weight a repeated number of times, the clinical goal is to strengthen the musculature of the legs and the goal of the exercise is to reach number of repetitions, for example 10. The user then performs the exercise but does so as indicated in Figure 1 , arching his/her back.

Regardless of reaching the goal, the effect of the exercise will be more harmful to health than beneficial. Therefore, the system is required to control not only reaching the goal of 10 repetitions but furthermore reaching this goal as illustrated in Figure 2. Concerning the current systems:

In the case of the Nintendo WiiFit and Stationary Bike products, exercise performance is controlled by means of a digital electronic balance board or a bike which sends the data wirelessly to the console. Said equipment is capable of controlling some values such as weight, pressure, distance, but not if the exercise is being performed correctly, being able to apply pressure on the balance board with a hand or to have incorrect posture on the bike. Furthermore, in no case is the system supervised remotely.

Motion capture technology based on machine vision is used in the GestureTek applications. This technique allows measuring the amount of motion, color changes, rotations, etc... but it does not provide sufficient precision to differentiate body parts or small motions, so it is not possible to perform the correctness follow-up necessary for avoiding injuries and inappropriate uses of the application.

The article entitled "The effects of mode of exercise instruction on correctness of home exercise performance and adherence" shows the need to create guidelines and instructions for correct exercise performance independently by the patient, but a methodology which allows analyzing that the result is correct is not used, so said correctness is left to the interpretation and will of the patient.

The document entitled "Method of exercise prescription and evaluation" presents a system for creating therapies of exercises adapted to each patient's needs. This is done at the beginning of the treatment and allows customizing goals according to patient characteristics, but the follow-up and modifications are performed concerning said goals and not the manner of performing the exercises.

The document entitled "Identifying Exercise Correctness for Home-Based" proposes a solution for identifying if an exercise is being performed correctly at home, removed from a hospital environment, without the supervision of a physical therapist. It is therefore the only solution found that provides a correctness calculation in real time, information to the user and adjustment of the exercise while performing it.

However, activity sensors and a classifier based on training by means of neural networks or a system having similar features are used to that end. This means that the exercise is performed correctly several times by a specialist, and an automatic system is trained. Then, when the patient performs the exercise at home the signal from the sensors is analyzed and compared with the correct performances that the specialist had performed, obtaining a similarity percentage. The system is more reliable if the specialist further performs some incorrect performances of the exercise on purpose and therefore trains the system with positive and negative performances.

The problems of this system are the need for prior training, the subjectivity in performing the exercise because weight, age, etc., the difficulty in adding new exercises and system precision are not taken into account, the system only showing an exercise correctness percentage, without giving specific information about the motion or posture that is causing the problem. It is therefore a supervised exercise training, wherein to improve the detection it would be necessary to perform non-automatic learning of the system of the different correctness-incorrectness situations in performing the exercises, being almost impossible to cover all the possible incorrect performance scenarios.

Summary of the Invention

It is necessary to find an alternative to the state of the art which allows making up for the detected shortcomings, particularly related to the lack of proposals having a system for evaluating exercise correctness while performing it or at the end thereof, displaying information in real time or at the end to the user and/or external supervisor and which allow the correction by the user while performing the exercise or in subsequent sessions.

To that end, the present invention proposes in a first aspect a physical exercise correctness calculation method, wherein said physical exercise is performed by a user connected or close to electronic sensing devices which collect measurements of different variables related to performing said physical exercise.

Unlike the already known proposals, the method of the invention in a characteristic manner comprises supervising said exercise remotely by means of performing the following actions:

- processing at least part of said variables in a computation device to which said electronic sensing devices are connected;

- obtaining said correctness calculation depending on the distance measured between at least one of said variables and previously established correctness values; and

- transmitting at least said correctness calculation to remote equipment for supervision.

Other aspects of the method of the first aspect of the invention are described according to the attached claims 2 to 14.

The present invention proposes in a second aspect a physical exercise correctness calculation system, wherein said physical exercise is performed by a user connected or close to electronic sensing devices which collect measurements of different variables.

Unlike the already known systems, the system of the second aspect of the invention comprises in a characteristic manner:

- a user computation equipment receiving and processing said measurements of said electronic sensing devices and obtaining said correctness calculation depending on the distance measured between at least one of said measurements and previously established correctness values;

- an application or a graphic interface between said user computation equipment and said user where indications referring to said physical exercise are shown to said user;

- remote computation equipment receiving said correctness calculation from said user computation equipment for the remote supervision by a supervisor;

- an application or graphic interface between said remote computation equipment and said supervisor where at least said correctness calculation is shown.

Other aspects of the system of the second aspect of the invention are described according to the attached claims 16 to 17.

Brief Description of the Drawings

The preceding and other advantages and features will be better understood from the following detailed description of several embodiments with reference to the drawings, some of which were already presented in the prior state of the art section, which must be interpreted in an illustrative and non-limiting manner, in which:

Figure 1 illustrates an incorrect performance of the exercise of lifting a weight.

Figure 2 shows the correct performance of the exercise of lifting a weight.

Figure 3 shows the diagram of the training system proposed in the present invention.

Figure 4 shows a user interface example with an exercise menu for a session, according to a possible embodiment of the present invention.

Figure 5 shows a user interface example while performing an exercise in which different feedback elements for the user are marked according to a possible embodiment of the present invention.

Figure 6 shows the different value sets for a variable of a sensor according to a possible embodiment of the present invention.

Figure 7 shows the correctness value set for a variable of a sensor, said set defined as interpolating extreme values, according to a possible embodiment of the present invention.

Figure 8 shows the correctness calculation value as the distance of the value of an isolated measurement with respect to the closest point of the correctness value subset for a variable of a sensor according to a possible embodiment of the present invention.

Figure 9 shows a numerical example of said correctness calculation value.

Figure 10 shows an example of the calculation of the overall correctness value in the case of having to evaluate more than one variable, said value being the arithmetic mean of the distances calculated for each variable, according to a possible embodiment of the present invention.

Figure 1 1 shows a numerical example in the event that the correctness variables are dependent, said dependent variables being grouped in a vector, according to a possible embodiment of the present invention.

Figure 12 shows the case in which the time is considered another variable of the vector which includes dependent variables, according to a possible embodiment of the present invention.

Figure 13 shows the case which considers different states of the time- dependent correctness variables, performing an interpolation between said states to confirm the correctness space, according to a possible embodiment of the present invention.

Figure 14 shows a sequence diagram of the web services between the different components of the method and system, according to a possible embodiment of the present invention.

Figure 15 shows a therapy configuration example in the server, according to a possible embodiment of the present invention.

Figure 16 shows a graphic interface example in which a user can select the exercise to be performed, according to a possible embodiment of the present invention.

Figure 17 shows a graphic interface example in which once the exercise is selected, the correct sensor placement, the number of sensors connected and the degree of correctness of the sensors, among other variables, are indicated according to a possible embodiment of the present invention.

Figure 18 shows certain flexion and extension angles for a specific knee exercise.

Figure 19 shows the correct performance of a knee rehabilitation exercise in which two moments in time for each of which the goal is different are defined, according to a possible embodiment of the present invention.

Figure 20 shows an incorrect performance of said knee rehabilitation exercise.

Figure 21 shows the exercise and correctness value sets for the sensors associated with said knee rehabilitation exercise, according to a possible embodiment of the present invention.

Figure 22 shows an example of calculating distances of a measured value with respect to the correctness value set associated with the two sensors used in said knee rehabilitation exercise, according to a possible embodiment of the present invention.

Figure 23 shows a server supervision web portal example which includes the results that the client application sends from each user, according to a possible embodiment of the present invention.

Figure 24 shows a flowchart of the method and system which indicates the input and output parameters for each modulus, according to a possible embodiment of the present invention.

Detailed Description of several Embodiments

The present invention defines a method and a system for:

- Defining exercises, sessions and therapies in accordance with user needs, with well-defined goals and correctness parameters.

- Performing either physical or mental goal-oriented rehabilitation and training exercises, which allows correctness calculation level of the exercise while performing it or at the end of the session by means of a data-processing application installed in a terminal with (fixed or mobile) computation capacities based on which the user operates.

- Obtaining and displaying a quantitative correctness value of the exercise to the user to allow correcting the exercise performance in real time or in subsequent sessions. Said value can be displayed to the user graphically/by sound/by vibration.

- Transmitting the goal-related results and correctness parameters for the supervision of said calculations by a specialist locally or remotely (through a web portal or local/remote application).

- Redefining the exercises and/or the parameters during therapy, after the assessment of the specialist, to update the exercises in the terminal of the user if necessary.

The exercises are performed using biometric sensors (accelerometers, gyroscopes, electrocardiographs, pressure meters, pulse oximeters, etc ..) connected to fixed or mobile data-processing equipment of the user (by means of Bluetooth, infrared, Wi-Fi, USB, cable, etc ..) with processing, audio, video and sufficient communication capabilities (personal computer with touch screen, portable computer, mobile telephone, PDA, built-in system or tablet PC) connected to a remote server on the Internet by means of network technology (LAN, Wi-Fi, GPRS, 3G, cable, Bluetooth, ...) for processing, storing, sending and supervising the results of the exercise. It should be pointed out that the present invention is independent of the specific hardware used and of the Internet connection.

The data-processing equipment receives the definition of the exercises of the user, and while performing them it collects and processes the information arriving from the wireless sensors for calculating the goal and the exercise correctness. While performing the exercise, the user can access the information of said correctness, knowing the possible error in the performance and correcting it in real time until the performance is satisfactory. At the end of the sessions, the data are sent to the server for supervision by the specialist.

Each sensor 'S' is capable of sensing a number 'N' of variables transmitted to the data-processing equipment of the user. Each exercise defines, in addition to the goal/goals to be met, the value sets for the variables indicating when the exercise is being performed correctly. The equipment processes said variables and calculates the correctness level for the exercise that is being performed, indicating the correctness percentage as well as the variables that are affecting said assessment (distance to the aforementioned correctness value set).

Time can be considered as another variable within the correctness variables or as a different state at all times of performing the exercise.

The present invention describes a method and system for defining and performing rehabilitation and training exercises by means of using multidisciplinary sensors and data-processing equipment connected to the Internet wherein in addition to calculating the goal, the exercise performance correctness is calculated in real time or at the end of the exercise to inform the user and/or remote equipment of the results, as shown in Figure 1.

The exercises are defined and performed within the framework of a training session, which is a set of exercises programmed for a given time (a day, an hour, etc .). The set of training sessions form therapy. The rehabilitation therapy or training is custom defined for each user depending on his/her medical condition by a therapist or specialist. The patient must be instructed as to which exercises to perform, their intensity, performance correctness and goals to be achieved. The present invention assumes the use of a computer program to make this information reach the patient. Furthermore, the computer program receives inputs from sensors which sense the state (motions, expressions, etc) of the patient and shows the patient relevant information. The exercises making up the training sessions are transmitted to the data- processing equipment of the user over the Internet and are stored locally for their programmed performance.

In conventional rehabilitation or training exercise, a goal is defined as a value that can usually be evaluated quantitatively and therefore the exercise performance (for example walking 10 kilometers) can be considered satisfactory (or not). These values are referred to as target exercise values.

Correctness is defined as the (often qualitative) measurement of said performance, i.e., of how well or how poorly the exercise is being performed depending on the dynamic parameters regulating it (speed, length of the step, rest times, etc .). The evaluation of the correctness can be carried out regardless of whether or not the established goal is being achieved. These parameters shall be referred to as exercise correctness parameters.

Therefore, each exercise could be associated with target values and correctness parameters. It must be indicated that the measurement of both parameters does not have to be performed with the same devices. Furthermore, the fact that a parameter is part of the correctness or target set depends on the exercise to be performed. That which is a goal for one exercise can be a correctness parameter for another, and vice versa. Overall progress, assessment and result values of an exercise, session or therapy can be calculated with these values and reported to the different users of the system.

Automatic sensing methods (biometric sensors, cameras, etc), direct feedback from the user (surveys) or the supervision of a specialist or family member can be used to measure the goal. The present invention can use any of these forms or other equivalents for obtaining said assessment, which will be processed and sent to the server equipment.

Algorithms and automatic methods such as biometric sensors or electronic devices with sufficient precision to assure the objectivity in the measurement will be used to measure correctness. Performing the exercise is defined in each of its phases by means of a set of variables which can be measured using the biometric sensors either with values collected directly from the sensor or through mathematic calculations (aggregates) on the unprocessed values from the sensors (calculation of angles from an accelerometer, for example). Different values of the correctness parameters can further be defined depending on the time for each moment of the exercise (at the beginning, during the exercise, at the end, etc.).

The calculation of this correctness is a fundamental part of the invention, because it proposes a novel and very precise system for being able to report both to the patient and, optionally, to remote server equipment of the correctness level in performing the exercise in real time and/or at the end of the exercise.

The first part of the invention consists of defining a set of exercises, sessions and therapies for a specific user depending on his/her medical condition. These exercises can be defined locally by means of a client application intended for this purpose or remotely by means of a web server or remote application. Therapies, sessions and exercises can be defined by the supervisor (doctor, physical therapist, family member, etc ..) with the help of therapy planning tools. The system includes:

- User management

- Therapy, session and exercise management

- Displaying therapy, session and exercise results

As shown in Figure 4, the client application will be run by the user every time he/she has to perform a programmed session. By means of the interface proposed by the application (using the mouse, touch screen, keys or wireless controller) the user will select the programmed exercises and perform them. The exercises usually consist of sets of iterations and can include a rest between sets and/or between exercises.

Once the user selects the exercise, the exercise performance screen containing information about the number of repetitions, sets, posture for performing the exercise, state of the sensors and correct placement thereof will be displayed.

Figure 5 shows the sitting posture of the user for performing the exercise, the correct placement of the sensor above the knee, the number of sets (1 ) and repetitions (10) to be performed, the goal of the exercise (lifting the leg in the direction of the arrow) and exercise correctness (green arrow indicates correct motion). The idea is for the user to imitate the posture of the avatar, but the system must further provide some mechanism for knowing if the user is indeed adopting the indicated posture. It is very important to point out that it does not propose an invention capable of detecting any incorrectness. It is impossible to automatically control all the circumstances for which an exercise is considered poorly performed. Rather it tries to add additional evaluation mechanisms assuming that the user wants, can and knows how to perform the exercise in a reasonable manner. The goal of the exercise is to know the value which allows quickly and objectively knowing if the user has achieved the purpose for which he is performing the exercise. The following can be used to measure said goal in the system:

- biometric sensors: defining the target values

- through surveys at the end of the exercise. These surveys are usually associated with assessment scales translating a qualitative and/or subjective assessment of the user into quantitative data.

- using cameras connected to the data-processing terminal of the user

- by means of the supervision of a therapist, family member, doctor, specialist, etc...

Once the system has processed the target values, it stores them locally and sends them to the remote server for supervision.

The correctness calculation is a fundamental part of the invention and consists of the measurement in real time or at the end of the exercise of the correctness parameters through electronic sensing devices (biometric sensors or the like) connected to the data-processing terminal of the patient by means of wired or wireless connections.

Said calculation allows knowing at all times if the exercise performance is correct. If the suitable algorithm is developed, it could even inform the user about why the performance is incorrect and how to correct it. In addition to the exercise performance, it is very relevant to know if the pre-exercise pre-requisites are complied with: sensor placement, initial posture of the patient, etc. The invention also allows checking the exercise preconditions and knowing, for example, if the sensors are correctly placed, showing the user information about the state of the sensor placement and the posture the user must have by means of the graphic interface.

To measure the correctness variables from the sensors (accelerometers, electronic gyroscopes, pressure sensors, etc ..) said electronic devices are placed on the patient's body, next to it (according to the type) by means of a fastening method (belts, pockets, etc .), or in the apparatus that is needed to perform the exercise. These sensors will send the detected values to fixed or mobile data-processing equipment with a wired connection (USB, serial or Ethernet), wireless connection (Bluetooth, ZigBee, Wi-Fi, Xbee, infrared) or through network gateways (router, switch, etc ..) for processing in said equipment which can be locally present or can be remote equipment connected to the Internet or to another computer network.

An SV (value set) set is defined for each variable of each sensor which is used for the correctness calculation and contains all the possible values of said variable. Then an SV subset, called SVe (exercise value subset), is defined containing those possible values that can be generated while performing a specific exercise of each variable. It must be mentioned that each exercise can have different or coinciding sets. Therefore within SV, several SVe sets will be defined for the same variable. A new SVec subset of SVe containing the values which are correct for a specific exercise is defined in the end. This grouping by sets is illustrated in Figure 6.

This separation is very important because it defines two steps in the correctness calculation. To begin with, if a value is outside the SVe set, it means that it will be within the SV set, but outside the possible exercise values, which indicates an impossible value for this specific exercise and the first correctness precondition is not met. This can indicate a poor calibration, misplacement, hardware and/or software errors of the sensor, etc... In any case, the exercise should not be started/continued.

The second step consists of detecting if the value of the variable is at all times within the SVec correctness set, or even belonging to the SVe exercise set, the value is not correct.

An example of these sets is the weight variable of an electronic balance board. The value set of the weight variable will be 0-250 (assuming 250 is the maximum value supported by the electronic device and it does not support negative weights). The values for an exercise will be in the set of [50-150] and the correctness values in the set of [75-95] for a specific user.

If the set was a continuous set, it can be defined by means of a range of values or by means of an initial state and an end state, performing an interpolation of the extreme values in both cases to be able to perform the correctness calculation. Figure 7 shows an example for a set consisting of the speed values comprised between 80 and 120. This set is defined by means of a segment.

The correctness value at a given time is calculated as the distance of the isolated value from its correctness value set. Said distance can be calculated in different ways according to the type of value (absolute value of the difference, modulus, distance to the center of the set, etc .).

Said correctness values will be calculated, normalized between 0 and 1 . This normalized value is obtained from the distance to the set value taking into account the maximum and minimum values to consider the exercise correct. Zero represents the smallest possible distance and it is a perfect correctness value, whereas 1 represents the farthest value, and therefore the most incorrect way to perform the exercise. Figure 8 shows an example where the calculation of the measurement to the closest point of the set is used.

Figure 9 shows an example for a correctness set consisting of the speed values comprised between 80 and 120. In this case the calculation used for the distance is the mathematic distance between a point and a segment. In other words, the correctness value is defined as the distance of the value of the variable to the closest point of the SVec set. The invention does not restrict the specific formula which is used to calculate the distance, any of them (quadratic, Euclidean, Manhattan distance, etc) being valid. A current value of 130 is assumed.

The correctness values (distance) will be normalized so that they are in the range of 0 to 1 , understanding that a distance = 1 means that the value is outside the values allowed for the exercise and 0 is within the allowed and correct values. All the distance values between 0 and 1 refer to possible but incorrect values, being more correct when they approach the value 0.

In the case of having to evaluate more than one variable to calculate exercise correctness, each of the sets defined above will exist for each variable, as shown in Figure 10. During the evaluation of the exercise, the correctness value (distance to the correct value set) of each variable will be calculated independently. Once all the correctness values are calculated, an aggregate value of these results will be calculated because it is the only way to obtain a single correctness value. This aggregation can be calculated using a weighted mean of the correctness values, median, or any other necessary aggregation formula that best represents the overall result. The information can thus be displayed calculated in two formats, one a high- level aggregate and one in detail with a value for each variable which allows knowing to what degree each one affects the overall correctness.

Finally, the correctness variables cannot always be considered independently. For example, if the time and the distance are taken into account as two correctness variables, separately it cannot be known if a time value of 10 seconds or a distance value of 5 meters is correct. In this case, the sets will consist of vectors with N components, one for each parameter. A two-variable vector will indicate that both variables are dependent. A three-variable vector means simultaneous dependence among the three variables (not two to two). The rest of the processing (interpolation, calculation of distance, etc ..) is performed similarly but on this vector space.

In this case, two states have been defined as shown in Figure 1 1 , an initial time state = 0 seconds and distance state = 0 meters and another end time state = 10 seconds and distance state = 5 meters. For the exercise to be correct, for the time 3s it was necessary to run 2.5 meters, but having run 1 meter a value of d=0.25 in correctness is obtained.

Given that an exercise may require different correctness or target values according to the time they are performed, there are two ways to take the time variable into account in the calculation. The first approach would consist of considering the time variable an additional value of the vector set defining the correctness space, as shown in Figure 12. The second alternative would be to consider different states according to the time and performing the interpolation between states mentioned above, as shown in Figure 13.

In this second case, the distance is calculated between the current sample and the curve which represents the interpolation between the states. It is considered that while the sample belongs to the curve, the exercise is being performed correctly. This allows further calculating which state of the system the sample is closest to.

· Application example: knee rehabilitation

An implementation of the invention which allows performing knee rehabilitation therapies from the patient's home is defined, said therapy being defined and supervised by a doctor or physical therapist from a hospital, rehabilitation center or clinic.

The elements forming part of the system in this example (but the invention is not limited to them) are:

- Data-processing server equipment connected to the Internet with an administration web portal

- Data-processing terminal for the user connected to the Internet with a touch screen or wireless controller

- Two independent sensors with accelerometer and gyroscope connected wirelessly to the data-processing equipment of the user (Xbee technology in this case) and with a rechargeable batter.

- Two belts for fastening the sensors on the user's leg.

The server has an administration web portal from where the supervisor can access his/her patient's data by means of user name and password. To generate a therapy for a patient, predefined templates can be used and completely new therapies can be created.

The server further has a series of web services (interfaces for data exchange via software) which the data-processing terminal of the patient will use to collect and send the data relating to therapies, sessions, exercises, questionnaires and results. Figure 14 shows the sequence diagram which illustrates the web services (arrows) between the different components of the invention.

In this case, the application will consist of a therapy comprising a knee rehabilitation exercise called sitting knee flexion-extension. It consists of repeating knee movements, extending and flexing the leg while sitting, in an upright position and keeping the femoral area in a horizontal position. By using the touch screen as the interface, as shown in Figure 16, the user chooses Select to start performing the exercise. At this time, he/she must place the sensors on himself/herself correctly (using the belts provided) so that the arrow indicating if the position is correct becomes green again.

Figure 17 shows the number of sensors connected, a necessary condition to be able to start the exercise, the position in which they must be placed (above and below the knee with the sensor on the outer part of the leg), and an end of the green colored arrow indicating that they are correctly placed. It must be indicated that it is reported that the sensors are well placed but it is not known if the sensors have been coupled to the patient's body. It is assumed that the patient has reasonably followed the instructions for performing the exercise. If that were not the case, it would be orange or red colored.

The goal and the correctness in performing a knee rehabilitation exercise in the sitting posture are to be measured. Two sensors (with accelerometer and gyroscope built into each sensor) are placed in the upper and lower parts of the knee respectively and supported on the outer part of the leg. In this case, the same sensors will be used for calculating the goal and the correctness, but this does not have to be the case. For exercise performance, a processing is performed on the data provided by the sensors to obtain a highest level value, which will be used to calculate the goal and the correctness. The information to be inferred from each sensor is the absolute rotation angle with respect to the X axis of each of them. The knee movements are in the same plane (plane XY). To that end, a Kalman filter is used on the accelerometer and a calibration is used on the gyroscope. The angle is obtained by combining both values.

The goal will be to stretch and flex the knee until obtaining certain flexion and extension angles, obtained as the difference between the angles of both sensors. The origin is marked in this case by the sensor itself, and the calculated angle is with respect to the Earth's gravity. The flexion and extension angles of the knee are measured with respect to certain axes the origin of which is located in the knee itself, total extension being 0°, progressively increasing towards flexion (a maximum of 150°). Figure 18 schematically depicts the angle measuring system. The range or interval of extension-flexion angles which a person can perform usually are referred to as range of motion. This range will be measured through the data obtained by two sensing devices placed on the patient's leg.

The correctness parameters are the separate angles of each sensor at all times of the exercise, for which a flexion value and another extension value will be defined.

1 ) S1 is the reported value for the sensor of the upper part and S2 the sensor of the lower part of the same leg, further taking the time variable T at two instants, T1 during flexion and T2 during extension. The sensors do not have an absolute reference value external to them. In other words, the coordinate axis is located in the center of each of them and the angles which are then reported are measured in reference to the X axis of said center. The values of the sensors for each time instant (states) are:

T1 : S1 =0, S2=90

T2: S1 =0, S2=0

2) The goal of the exercise is to start with the knee flexed at 90° and to extend it until reaching zero degrees, once this is done, it must be flexed again to 90°. As can be seen, one of the sensors (S1 ) is kept in the same position while the other sensor (S2) changes its position angle. The flexion angle is obtained by subtracting the absolute angles of each sensor. This resulting angle is the goal to be achieved.

The problem which arises is that a ninety degree angle must be achieved as the target value, then zero and then flexing again to ninety degrees, but it is not assessed if it is achieved correctly. To that end, the value of the angles of each sensor will be considered separately as the correctness parameters, and two correctness value sets will thus be formed.

In the case of performing the exercise as shown in Figure 20, the goal of starting at 90° and reaching 0° can be met, but the exercise is not being performed correctly, because now the values of the angle separately are:

T1 : S1 =-90, S2 = 0

T2: S1 = -90, S2 = -90

Such incorrect performance can lead to problems such as injuries and/or unsatisfactory rehabilitation.

3) To calculate the value exercise set of each variable to be measured, the angle of each sensor in this case, the range of values that each sensor can take (SV set) must be known. In this case, the SV set of each sensor is the same [0°-360°], the SVe exercise subset is [0°-360°] and the correctness subset of each sensor is the range [0,0] for S1 and the range [0,90°] for S2, as illustrated Figure 21.

It can be seen how, by linear interpolation between the value sets of the variables, the correct values for performing the exercise are those which leave the value of sensor S1 at zero, and the values of sensor S2 between 0° and 90°.

Two states of the exercise to be assessed have been defined, state E1 during flexion and state E2 during extension. Therefore, the values defining if the exercise is correct are those which are between both states, these included, obtained by interpolation, resulting in two segments, one for each variable. It proposes using interpolation techniques to construct the curve or hyperplane connecting the expected correctness values throughout successive states. The way to calculate correctness values in a state E (not previously specified) is by using this interpolation function. Given a new value obtained from the sensors, it is checked if the value belongs to the interpolated curve by calculating the distance of the value to the closest point of said curve. If it belongs, it means that the value is correct and the interval of states where this value must be located can further be estimated.

In this case, the variables are independent, so the correctness value of each one can be calculated separately.

In the case of taking an exercise performance value at a given time T, the following values, for example, are obtained: S1 =5° and S2=100°.

The independent distances are D1 = 0.05 and D2 = 0.10, as is illustrated in Figure 22. The aggregate of both values using the arithmetic mean as an aggregation formula would be D = 0.075 which is the correctness value for this moment of the exercise.

For the exercise to be perfectly performed, the distance d must be very close to zero. In this case, if the variables are analyzed separately, it is observed that an error of 0.05% is being made in sensor S1 and an error of 0.10% is being made in sensor S2 with respect to the correct exercise.

In the implementation example of the invention, the final component is the supervision web portal of the server. It contains all the results that the client application sends from each user in order to perform goal and correctness follow-up, therapy update and decision making. As in the case of administration, a user name and password are used by the supervisor to access the data.

The following is shown for each patient: personal data of the patient, warnings received, therapies, sessions, exercises, questionnaires and results. Finally, Figure 24 shows a flowchart of the method and system, indicating the input and output parameters of the proposed invention, where it is already assumed that all the parameters necessary for the performance are calculated:

- Already processed variables of the sensors (or known algorithms)

- Pre-analysis of the exercise, with the correctness variables space

- Necessary time values, iteration values, etc....

Advantages of the invention

- Method and system for defining and remotely supervising therapies, sessions and exercises. Real time follow-up of patients and users.

- Method and automatic system for measuring the correctness in performing exercises which do not require external human supervision or training. It can be used by the patient directly once the exercise is defined.

- Real time control of the values defining exercise correctness.

- Definition of variables affecting the correctness of an exercise and identification of which of the values produces an incorrect result.

- Immediate information to the user of the correctness of his/her exercise, indicating the incorrect values in real time and the possibility of supervision by specialized personnel.

- Summarized information of the target values, correctness parameters and evolution of the therapy during and at the end of it.

- Separation between goal of an exercise and correctness in performing it.

- Applications in games for rehabilitation which allow including more variables in developing the exercise and better feedback for the patient.

A person skilled in the art could introduce changes and modifications in the described embodiments without departing from the scope of the invention as it is defined in the attached claims.

Literature

[1] http://es.wikipedia.org/wiki/Wii_Fit

[2] http://www.gesturetekhealth.com/

[3] http://www.bikerumor.com/2009/10/14/nintendo-wii-stationary- bike-controller/ [4] http://www.crd.york.ac.uk/CRDWeb/ShowRecord. asp?View=Full&ID=120040052 29 [5] http://www.freepatentsonline.com/6626800.html

[6] http://sciencestage.com/d/637208/identifying-exercise-correc tness-for-home- based.html.