METHOD AND SYSTEM FOR PROVIDING A TRAINING PROGRAM TO A SUBJECT

[0001] The present application relates to a training method and system, more particularly, to a method and system for providing a training program to a subject.

[0002] At present various devices are known to provide exercises for a subject, e.g. a patient who suffers from loss of motor function as a result of accident or disease. Usually these exercises are efficient in regaining motor control, provided the training is intense and the patient is guided in the therapy. Another example of a subject is an athlete.

[0003] A rehabilitation system is disclosed in application CN 200410056143.0. In the disclosure of the rehabilitation system, during the exercise, the posture of a patient is captured by two cameras. The parameters, such as the range of movement, physical activity level, etc., acquired by the cameras and/or other sensors, are used to evaluate an actual performance of the patient during one exercise. A performance goal of the exercise, such as target level of the exercise, is predefined by a rehabilitation specialist. The specialist may make a more accurate diagnosis and/or set up a more suitable rehabilitation program for the patient, based on a comparison between the actual performance and the target level.

[0004] The rehabilitation program, e.g. target level of an exercise, cannot be adjusted until the patient visits the specialist. The duration may be too long to ensure compliance of the patient. The patient may become de-motivated to do the exercise especially in an unsupervised home rehabilitation program.

[0005] It is therefore an object of this application to provide a method and a system for providing a training program that is adjusted automatically.

[0006] In accordance with one aspect, a method of providing a training program including at least a first exercise and a second exercise is provided, the method comprising the steps of: acquiring one or more parameters associated with the first exercise performance of a subject; and adjusting, based on said one or more parameters, one or more target values of the second exercise to be provided after said first exercise.

[0007] In accordance with another aspect, a system of providing a training program including at least a first exercise and a second exercise is provided, the system comprising: a first unit for acquiring one or more parameters associated with the first exercise performance of a subject; and a second unit for adjusting, based on said one or more parameters, one or more target values of the second exercise to be provided after said first exercise.

[0008] These and other characteristics, features and advantages of the present invention will become apparent from the following detailed description, taken in conjunction with the accompanying drawings, which illustrate, by way of example, the principles of the invention. The description is given for the sake of example only, without limiting the scope of the invention. The reference numerals given below refer to the attached drawings.

[0015] Throughout the above drawings, like reference numerals will be understood to refer to like, similar or corresponding features or functions.

[0016] The present invention will be described with respect to particular embodiments and with reference to certain drawings, but the invention is not limited thereto but only by the claims. The drawings described are only schematic and are non-limiting. In the drawings, the size of some of the elements may be exaggerated and not drawn to scale for illustrative purposes. In the block diagrams of FIG. 1 and FIG. 4, the dashed lines indicate that the element in question may be removed from the system in various embodiments; in the flowcharts of FIG. 3 and FIG. 5, the dashed lines indicate that the step in question may be removed from the process in various embodiments. Where an indefinite or definite article is used when referring to a singular noun, e.g. “a”, “an”, “the”, this includes a plural of that noun unless something else is specifically stated.

[0017] FIG. 1 shows a system 100a of providing a training program to a person P according to an embodiment of the invention. The system 100a comprises a sensing device 10, an analyzer 20a and a controller 30. The sensing device 10 may include a plurality of sensors to detect movement of a person P. The sensors can be e.g. inertial sensors, marker-based or markerless camera systems for motion acquisition. During the time that an exercise is performed by the person P, the sensing device 10 is used to monitor the exercise and capture parameters of the movements of the person P. The parameters are supplied to the analyzer 20a via wired or wireless connection 60 between the sensing device 10 and the analyzer 20a.

[0018] In an embodiment, the parameters may be represented as a set of progress factors associated with an exercise for a specific training, for example, rehabilitation therapy. The values of the progress factors may be associated with the exercise performance of the person P. In a physical activity exercise for upper limbs, the set of progress factors may include range of movement PF_a, speed of movement PF_b, smoothness PF_c, and trunk stability PF_d. An actual performance 25a (i.e. performance level) of the exercise performed by the person P is generated by the analyzer 20a based on analysis of the progress factors PF_a, PF_b, PF_c and PF_d.

[0019] In another embodiment, the system 100a further comprises a weight supplier 40 which may supply a set of weight factors to the analyzer 20a based on characteristics of an exercise. In a physical activity exercise for stretching upper limbs, for example, the range of movement PF_a and trunk stability PF_d are used to evaluate the actual performance 25b of the exercise performed by the person P. Correspondingly, weight factors WF_a and WF_d are set to non-zero values, and weight factors WF_b and WF_C are set to zero. The analyzer 20a selects PF_a and PF_d in accordance with the non-zero weights WF_a and WF_d. The actual performance 25b is thus evaluated from the progress factors PF_a and PF_d.

[0020] In addition, an exercise may be performed repeatedly by the person P. The number of repetitions of the exercise may be predetermined or determined by a rehabilitation specialist. In another embodiment, the system 100a may further comprise a storage 50. During the time that an exercise is performed repeatedly, values of progress factors for the exercise are stored in the storage 50. In another embodiment, where it is assumed that an exercise is performed repeatedly for ten times, values of progress factors for one or more mid-repetitions of the exercise, five mid-repetitions for example, are stored in the storage 50. When the person P finishes the exercise, each of the progress factors includes a group of values to be stored in the storage 50. The average of one group of values is calculated as an actual value of one progress factor. The actual values are used to evaluate an actual performance 25c of the exercise performed by the person P. That is to say, the actual performance 25c may be obtained on the basis of actual values for each of progress factors PF_a, PF_b, PF_c and PF_d.

[0021] The controller 30 is provided with the actual performance (25a, 25b, or 25c). In the controller 30, the actual performance is compared with a target level of the exercise. An instruction is generated by the controller 30 to adjust the target level of an upcoming exercise, based on the compared result, which will be explained in detail later in the text. The controller 30 provides the upcoming exercise with adjusted target values to the person P through a display (not shown).

[0022] FIG. 2 is an illustration of a training program including a series of exercises which can be monitored by the system 100a as shown in FIG. 1. In this embodiment, the training program can be used for stroke rehabilitation. The training program includes a plurality of sessions, each session further including a series of exercises, and the series of exercises being intended to be sequentially provided to the subject. As shown in FIG. 2, a session of physical activity for upper limbs includes three exercises, i.e. Exercise A for hand movement, Exercise B for wrist rotation and Exercise C for stretching the upper limbs.

[0023] An evaluation of the performance level for each of the exercises in the session of physical activity for upper limbs is made on the basis of four progress factors, i.e. range of movement PF_a, speed of movement PF_b, smoothness PF_c, and trunk stability PF_d. One or more progress factors are selected by the corresponding weight factors, and in one embodiment, these selected progress factors may be used in the evaluation of the actual performance 25b. Here it is assumed that Exercises A, B and C represent Stages 1, 2 and 3. The upper index 1, 2, and 3 in progress factor (PF) or weight factor (WF) shown in FIG. 2 indicates values of Exercises A, B and C corresponding to stage 1, 2 and 3 respectively.

[0024] The following is a table showing a relationship between an evaluation of an actual performance of an exercise, progress factors and weight factors.

TABLE 1
	Progress Factors (PF)	Weight Factors
Exercise	Used in Evaluation	(WF)
Exercise A	PF1a, PF1c	WF1a > 0, WF1c > 0
(hand movement)		WF1b = 0, WF1d = 0
Exercise B	PF2a, PF2b, PF2d	WF2a > 0, WF2b > 0
(wrist rotation)		WF2d > 0, WF2c = 0
Exercise C	PF3a, PF3d	WF3a > 0, WF3d > 0
(stretching the upper limbs)		WF3b = 0, WF3c = 0

[0025] FIG. 3 illustrates a flowchart for a method of providing a training program as shown in FIG. 2 to the person P. Exercises A, B and C are intended to be sequentially provided to the subject. During the time that Exercise A is provided to the subject, a sensing device 10 monitors Exercise A performed by the person P to acquire progress factors, i.e. range of movement PF¹_a, speed of movement PF¹_b, smoothness PF¹_c, and trunk stability PF¹_d (Step S210).

[0026] During the time that the person P performs Exercise A repeatedly, e.g. ten times, values of the progress factors for the five mid-repetitions of Exercise A are stored in the storage 50 (Step S220). Once the person P finishes the Exercise A, average values for each of the progress factors PF¹_a, PF¹_b, PF¹_c and PF¹_d are calculated as actual values based on the stored values of the progress factors from the five mid-repetitions of Exercise A. The actual values are supplied to the analyzer 20a (Step S230).

[0027] Based on characteristics of Exercise A, i.e. hand movement in the session of physical activity for upper limbs, the weight supplier 40 provides a set of weight factors WF¹_a, WF¹_b, WF¹_c and WF¹_d corresponding to the acquired progress factors PF¹_a, PF¹_b, PF¹_c and PF¹_d (Step S240). The weight factors are also supplied to the analyzer 20a, wherein the values of WF¹_a and WF¹_c are non-zero, and the values of WF¹_b and WF¹_d are zero (Step S250).

[0028] The analyzer 20a, based on the actual values of the progress factors and the weight factors, generates an actual performance (i.e. actual performance level) of Exercise A (Step S260). Since WF¹_b and WF¹_d are zero, PF¹_a and PF¹_c corresponding to non-zero weights WF¹_a and WF¹_c, are selected to be used in the evaluation of the actual performance level of Exercise A. Then, the actual performance level is supplied to the controller 30 (Step S270). The controller 30 compares the actual performance level with a target level of Exercise A (Step S280a). In this embodiment, the target level for each exercise is determined by a set of target values of the progress factors associated with the exercise. Preferably, initial target values are predefined or predetermined by a specialist.

[0029] Based on the comparison result, if the difference between the actual performance level and the target level of Exercise A exceeds a threshold value, the controller 30 generates an instruction to adjust the target level of upcoming Exercise B and/or Exercise C (Step S290a). In particular, if the actual performance level shows the person P performs Exercise A very well and the instruction then indicates to adjust the target level of the next exercise (i.e. Exercise B), target values of progress factors PF²_a and PF²_c of Exercise B may be increased based on the selected progress factors PF¹_a and PF¹_c of Exercise A. On the other hand, if the actual performance level shows the person P performs Exercise A poorly and the instruction also indicates to adjust the target level of Exercise B, the target values of the progress factors PF²_a and PF²_c of Exercise B may be decreased based on the selected progress factors PF²_a and PF¹_c of Exercise A. Since a target level of an exercise is determined by target values of progress factors of the exercise, the target level of Exercise B will be adjusted by updating the target values of progress factors thereof.

[0030] In an embodiment of the step for adjusting the target level of Exercise B, PF²_a and PF²_c of Exercise B are updated by replacing the initial target values of PF²_a and PF²_c of Exercise B with the actual values of PF¹_a and PF¹_c of Exercise A respectively. In an alternative embodiment, the initial target values of PF²_a and PF²_c of Exercise B are updated by multiplying them by a coefficient which depends upon the actual performance level of Exercise A.

[0031] Then, Exercise B, with adjusted target values, is provided to the person P. During the time that the person P performs the next exercise (i.e. Exercise B), a similar process, i.e. Step S210-Step S290a, is performed. The acquired progress factors PF²_a, PF²_b and PF²_d are selected to be used in an evaluation of the actual performance level of Exercise B, and whether the target level of the next Exercise C is adjusted depends upon a comparison between the actual performance level of Exercise B and the target level of Exercise B that has been adjusted on the basis of progress factors of Exercise A.

[0032] In an embodiment of the step for adjusting the target level of Exercise C, according to the compared result, part of target values of PF³_a, PF³_b and PF³_d of Exercise C are updated. As an example, target values of PF³_a and PF³_b of Exercise C are replaced by the actual values of PF²_a and PF²_b of Exercise B, while the target value of PF³_d of Exercise C is not updated. Then, Exercise C, with adjusted target values, is provided to the person P.

[0033] According to the embodiment described above, the initial target values of PF²_a and PF²_c of Exercise B are updated based on the actual values of PF¹_a and PF¹_C of Exercise A. Further, the initial target values of PF³_a and PF³_b of Exercise C are updated by the actual values of PF²_a and PF²_b of Exercise B. When the person P starts to perform Exercise C, all target values of the progress factors of Exercise C, except PF³_d, are updated. Since at least some of the target values of progress factors of Exercise C are different from the initial target values thereof, the target level of Exercise C is not the initial target level of Exercise C.

[0034] FIG. 4 shows a system 100b of providing a training program to a person P according to another embodiment of the invention. In comparison with the system 100a illustrated in FIG. 1, the system 100b comprises an analyzer 20b. In addition to analyzer 20b, sensing device 10, controller 30, weight supplier 40 and storage 50 in the system 100b, the system (?) can adopt the same or similar devices as the system shown in FIG. 1. The detailed description of these same or similar devices is omitted herein.

[0035] As illustrated in FIG. 4, in addition to an evaluation of the actual exercise performance by the analyzer 20a in FIG. 1, the analyzer 20b further comprises a constructor to create a coefficient matrix for updating target values of upcoming exercises. A process performed by the system 100b is illustrated in FIG. 5. The process may be described according to an embodiment to provide the training program as shown in FIG. 2.

[0036] As illustrated in FIG. 5, the analyzer 20b obtains training data acquired by the sensing device 10 during the time that a person P performs Exercise A (Step S205). Then, the constructor generates the coefficient matrix by using the training data, based on a linear approximation model (Step S208).

[0037] On the supposition that progress factors of each exercise are represented as elements of a matrix, the matrix of stage m and the matrix of stage m−1 can be expressed as:

[PF^m_a,PF^m_b,PF^m_c . . . PF^m_n]=Coff_n×n[PF^m-1_a,PF^m-1_b,PF^m-1_c . . . PF^m-1_n]  (1)

where the index (e.g. m and m−1) indicates the stage of the exercise, n is the number of progress factors associated with a specific exercise, and Coff_n×n represents the coefficient matrix.

[0038] In this embodiment, a matrix of Exercise A and a matrix of Exercise B can be shown as:

[PF²_a,PF²_b,PF²_c,PF²_d]=Coff_n×n[PF¹_a,PF¹_b,PF¹_c,PF¹_d]  (2)

[0039] Coff_n×n is a 4×4 matrix, thus expression (2) is further represented as:

$\begin{array}{cc}\left(\begin{array}{c}{\mathrm{PF}}_a^2\\ {\mathrm{PF}}_b^2\\ {\mathrm{PF}}_c^2\\ {\mathrm{PF}}_d^2\end{array}\right)=\left[\begin{array}{cccc}{\mathrm{Coff}}_{11}& {\mathrm{Coff}}_{12}& {\mathrm{Coff}}_{13}& {\mathrm{Coff}}_{14}\\ {\mathrm{Coff}}_{21}& {\mathrm{Coff}}_{22}& {\mathrm{Coff}}_{23}& {\mathrm{Coff}}_{24}\\ {\mathrm{Coff}}_{31}& {\mathrm{Coff}}_{32}& {\mathrm{Coff}}_{33}& {\mathrm{Coff}}_{34}\\ {\mathrm{Coff}}_{41}& {\mathrm{Coff}}_{42}& {\mathrm{Coff}}_{43}& {\mathrm{Coff}}_{44}\end{array}\right]\left(\begin{array}{c}{\mathrm{PF}}_a^1\\ {\mathrm{PF}}_b^1\\ {\mathrm{PF}}_c^1\\ {\mathrm{PF}}_d^1\end{array}\right)& \left(3\right)\end{array}$

[0040] In accordance with a matrix algorithm, in order to obtain the values of each element in the coefficient matrix, the training data at least includes five groups of progress factors for one exercise, wherein each group of the progress factors can be acquired by sensing device 10 from one repetition of the exercise performed by the person P. In an embodiment, the person P performs the exercise ten times, and the training data is obtained from one or more mid-repetitions of the exercise, for example five mid-repetitions.

[0041] Using the training data, the coefficient matrix Coff_n×n is formed based on a linear approximation model, for example minimizing RMSE (root mean square error). After the coefficient matrix Coff_n×n is obtained from the training data, a similar process, i.e. Step S210-Step S280a, is performed. The acquired progress factors PF_a and PF_c are selected to be used in an actual performance level evaluation of Exercise A, and whether the target level of Exercise B is adjusted depends upon a comparison between the actual performance level of Exercise A and a target level of Exercise A. Steps S210-S280a may be the same or similar steps as those shown in FIG. 3. The detailed description of these steps is omitted herein.

[0042] Based on the comparison result, if the difference between the actual performance level of Exercise A and the target level of Exercise A exceeds a threshold value, the controller 30 generates an instruction to adjust the target level of upcoming Exercise B (Step S290b). In Step S290b, target values of the progress factors PF²_a, PF²_b, PF²_c and PF²_d of Exercise B are obtained in accordance with Expression (3). As compared with Step S290a in FIG. 3, each of the actual values of progress factors PF¹_a, PF¹_b, PF¹_c and PF¹_d of Exercise A makes contributions to target values of the progress factors of Exercise B. That is to say, the target values of the progress factors of Exercise B are adjusted by a combination of actual values of progress factors PF¹_a, PF¹_b, PF¹_c and PF¹_d of Exercise A. In other words, even though PF¹_b and PF¹_d of Exercise A are not selected to be used for an evaluation of the actual performance level of Exercise A according to non-zero weights, each one of the progress factors of Exercise A, including PF¹_b and PF¹_d, exerts an influence on the target values of the progress factors of Exercise B.

[0043] Then, Exercise B, with adjusted target values, is provided to the person P. During the time that the person P performs next Exercise B, a similar process from Step S210-Step S290b is performed. Target values of the progress factors PF³_a, PF³_b, PF³_c and PF³_d of Exercise C are updated by multiplying actual values of progress factors PF²_a, PF²_b, PF²_c and PF²_d of Exercise B with the coefficient matrix Coff_n×n generated on the basis of training data acquired in Exercise A. So, the target level of Exercise C is adjusted according to the actual values of progress factors of Exercise B.

[0044] The system of providing a training program to a subject and a method performed by the system should not be limited to embodiments mentioned above. It will be apparent to those skilled in the art that the various aspects of the invention claimed may be practiced in other examples that depart from these specific details.

[0045] In an alternative embodiment of the system 100a, the analyzer 20a may be removed from the system. Accordingly, the step for performing an actual performance evaluation of Exercise A (Step S260) and the step for supplying the actual performance level (Step 270) are not performed. The comparing step 280a as shown in FIG. 3 or FIG. 5 may be modified as step 280b accordingly.

[0046] After actual values for each of the progress factors PF_a, PF_b, PF_c and PF_d are supplied to the controller 30 (Step 230) and the weight factors, provided by the weight supplier 40, are also supplied to the controller 30 (step 240), the step 280b may be performed in other embodiments as described below.

[0047] In an embodiment, the controller 30 generates an instruction to update target values of Exercise B. The controller 30 selects corresponding actual values of the progress factors of Exercise A according to non-zero weights provided by the weight supplier 40. The target values of Exercise B are then replaced by the corresponding actual values of the progress factors of Exercise A. In another embodiment, the controller 30 compares actual values of the progress factors of Exercise A with target values of Exercise B first. In still another embodiment, actual values of the progress factors of Exercise A provided with non-zero weights are compared with target values of Exercise B. Whether the target values of Exercise B are adjusted depends upon the comparison result. For example, PF¹_a and PF¹_c of Exercise A are compared with PF²_a and PF²_c, respectively, of Exercise B. Based on the comparison results, if the difference between PF¹_a of Exercise A and PF²_a of Exercise B exceeds a threshold value, while the difference between PF¹_c of Exercise A and PF²_c of Exercise B does not exceed a threshold value, only PF²_a of Exercise B is replaced by PF¹_a of Exercise A. That is to say, each of the target values of Exercise B may be adjusted by the result of the comparison between the actual values of the progress factors of Exercise A and the corresponding target values of Exercise B.

[0048] In addition, the order of the steps should not be limited to the procedure shown in FIG. 3 and FIG. 5. In another embodiment of the method of providing a training program to a subject, the step for generating weight factors (i.e. Step 240) and supplying the weight factors to the analyzer (i.e. Step 250) for example, may be performed before the step for supplying the actual values of progress factors to the analyzer (i.e. Step 230).

[0049] Moreover, the coefficient matrix described in system 100b may be created based on training data associated with each one of the exercises. That is to say, a new coefficient matrix may be generated using training data acquired at the start of an upcoming exercise. The new coefficient matrix is used to adjust target values of the upcoming exercise. In an alternative embodiment, a coefficient matrix may be shared by exercises included in one session. The new coefficient matrix is generated using training data acquired at the start of the exercise in the next session performed by a subject.

[0050] Additionally, as described in the embodiment mentioned above, the coefficient matrix is formed based on a linear approximation model. However, other mathematical models in a neural network for example may be applied as well to generate the coefficient matrix based on the training data.

[0051] Further, the operation of updating target values of upcoming exercises based on progress factors acquired in an exercise may be performed at the end of the exercise or at start of the upcoming exercise alternatively.

[0052] Additionally, the analyzer 20a, the controller 30 and the weight supplier 40 are described as separate modules in the embodiments as shown in FIGS. 1 and 4. However, a person skilled in the art may understand that the functions of these modules can be implemented by a processor and a readable medium on which a software program including a set of instructions is recorded. When the instructions are executed, the processor will be enabled to perform any one of the methods described in the above-mentioned embodiments.

[0053] It should be noted that the above described embodiments are given for describing rather than limiting the invention, and it is to be understood that modifications and variations may be resorted to without departing from the spirit and scope of the invention as those skilled in the art readily understand. Such modifications and variations are considered to be within the scope of the invention and the appended claims. The protective scope of the invention is defined by the accompanying claims. In addition, any of the reference numerals in the claims should not be interpreted as a limitation to the claims.