Method to establish determination model and method to determine foot accessory

Abstract

To provide method to establish determination model and a method to determine foot accessory.SOLUTION: A method to establish a determination model for determining a foot accessory is to be implemented by an electronic device. The electronic device stores training data sets that correspond to sampled objects. The method includes: for each of the training data sets, grouping entries of sensor data of the training data set into sensor-data groups that correspond to phases of a specific activity; establishing a classification model on the basis of the entries of sensor data that belong to a target one of the sensor-data groups of each of the training data sets; and combining the classification model and a lookup table by using an output of the classification model as an input of the lookup table so as to obtain the determination model.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present disclosure relates to a method for establishing a determination model for determining foot wear and a method for determining foot wear. [Background technology]

[0002] Because people have different foot shapes and walking styles, they need a variety of foot orthoses that fit their feet. Traditionally, people typically try on foot orthoses to determine whether they fit their feet. However, this method can be unreliable, and using an inappropriate foot orthosis can cause discomfort, poor posture, extra strain, and an increased risk of injury.

[0003] Related prior art includes, for example, Chinese Patent Application Publication No. 109602422, US Patent Application Publication Nos. 20230153890, 20240315390, and 20240315390. Summary of the Invention [Problem to be solved by the invention]

[0004] The object of the present disclosure is to provide a determination model for determining foot equipment suitable for a particular type of movement of a subject, and a method for determining foot equipment suitable for a particular type of movement of a subject, which can alleviate at least one drawback of the prior art. [Means for solving the problem]

[0005] The present invention provides a method for establishing a judgment model for determining foot equipment suitable for a specific type of movement of a subject, the method being executed by an electronic device, the electronic device storing a plurality of training data sets corresponding to a plurality of sample subjects, each of the plurality of training data sets being related to one of a plurality of specific types of movement and including a plurality of sensor data entries, the plurality of sensor data entries being generated for a corresponding one of the sample subjects based on detection of the sample subject's movement, and each of the plurality of sensor data entries being at a plurality of time points when the sample subject performs a specific activity, the specific activity having a plurality of periods, the method including, for each of the plurality of training data sets, determining the plurality of sensor data entries of the training data set. The present invention provides a method for establishing a judgment model, the method including: grouping sensor data entries into a plurality of sensor data groups, each corresponding to a period of the specific activity; establishing a classification model for determining one of the plurality of specific types of movement of the subject based on the plurality of sensor data entries belonging to a target in the plurality of sensor data groups of each of the plurality of training data sets and outputting the one of the plurality of specific types of movement as an output; and obtaining a judgment model for determining foot equipment suitable for the subject based on the output of the classification model, the judgment model being configured to associate the plurality of specific types of movement with a plurality of types of foot equipment.

[0006] The present invention provides a method for determining foot equipment suitable for a specific type of movement of a subject, the method being executed by a user device, the user device being electrically connected to a sensor device attached to the subject's lower leg, the sensor device generating a plurality of detection data entries based on detection results, each of which is at a plurality of time points when the subject is performing a specific activity, the specific activity having a plurality of periods, the user device storing a determination model obtained by the method described in claim 1, the method comprising: grouping the plurality of detection data entries into a plurality of detection data groups, each of which corresponds to a period of the specific activity; using the determination model to determine one of the specific types of movement related to the subject as a target type based on the plurality of detection data entries belonging to a target one of the plurality of detection data groups; and using the determination model to determine one of a plurality of types of foot equipment that matches the target type.

[0007] According to the foregoing two aspects of the present disclosure, there is provided a method as set out in each of the accompanying claims.

[0008] Other features and advantages of the present disclosure will become apparent from the following detailed description of the embodiments, taken in conjunction with the accompanying drawings, in which various features may not be drawn to scale. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a block diagram illustrating a user device and a sensor device according to an embodiment of the present disclosure. [Figure 2] FIG. 1 is a flow diagram illustrating a method for establishing a decision model according to an embodiment of the present disclosure. [Figure 3] FIG. 1 is a schematic diagram showing a gait cycle during the stance phase of walking. [Figure 4] 1A and 1B are schematic diagrams showing examples of forward / backward and inward / outward accelerations in three types of motion. [Figure 5]FIG. 1 is a flow diagram illustrating a method for determining a foot attachment according to an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0010] Before describing the present invention in more detail, it should be noted that, where considered appropriate, reference numerals or terminal portions of reference numerals have been repeated among the figures to indicate corresponding or analogous elements, which may optionally have similar characteristics.

[0011] 2, an embodiment of a method for establishing a judgment model for determining a foot attachment suitable for a specific type of movement of a subject (e.g., a human) of the present disclosure is shown. The foot attachment can be realized by, but is not limited to, a pair of shoe insoles, a pair of shoes, a pair of socks, a foot support, etc.

[0012] The method is performed by an electronic device (not shown), which may be implemented as, but is not limited to, a smartphone or a personal computer (e.g., a tablet computer, a desktop computer, a notebook computer, etc.).

[0013] The electronic device stores a plurality of training datasets, each corresponding to a plurality of sample subjects. The sample subjects are people other than the subject. Each training dataset is associated with one of a plurality of specific types of movement. Each training dataset includes, for a corresponding one of the sample subjects, a plurality of sensor data entries generated based on detection of the sample subject's movement and at each of a plurality of time points when the sample subject performed a specific activity. The plurality of sensor data entries are generated by a measurement device connected to an electronic device attached to a corresponding one of the sample subjects in each of the plurality of training datasets on a lower limb (e.g., thigh, calf, or foot). The measurement device can be implemented to include one or more inertial measurement units (IMUs), such as a gyroscope for measuring orientation and angular velocity or an accelerometer for sensing appropriate acceleration, or a magnetometer for measuring magnetic fields or magnetic dipole moments. The connection between the electronic device and the measurement device can be implemented via Bluetooth® wireless technology, Wi-Fi technology, radio frequency identification (RFID), or the like. In a scenario in which the anterior / posterior, superior / inferior, and medial / lateral directions of the lower limbs are each defined as three-dimensional axes in a three-dimensional Cartesian coordinate system, the multiple sensor data entries generated by the measurement device may include the angular velocity of rotation of the axis of the lower limb measured by the gyroscope, the acceleration of the lower limb sensed by the accelerometer and including three acceleration configurations each corresponding to a dimensional axis, and three relative changes in the magnetic field around the lower limb each associated with a dimensional axis.

[0014] The specific activity has multiple phases. In this embodiment, the specific activity is walking, and the specific activity has a stance phase and a swing phase. Note that the stance phase is when the foot under study is in contact with and supported on the floor, and the swing phase is when the foot under study is not in contact with the floor due to movement (i.e., the leg connected to the foot is swinging in the air). The multiple specific types of movement in the multiple training datasets may include types associated with overpronation, pronation, and supination. However, the specific activities and their phases are not limited to the disclosure herein and may vary in other embodiments.

[0015] As shown in FIG. 3 , in one embodiment, the gait cycle during the stance phase of gait can be further divided into a heel-strike subphase, a load-response subphase, a mid-stance subphase, an end-stance subphase, and a pre-swing subphase. The heel-strike subphase is associated with a situation in which the heel of a person's foot first contacts the ground and the person slightly bends their knee to prepare to support their body weight. The load-response subphase is associated with a situation in which the person continues to bend their knee and rotates their foot forward until the entire sole of the foot contacts the ground and supports their body weight. The mid-stance subphase is associated with a situation in which the person gradually straightens their knee and their body weight is propelled forward until the greater trochanter of the person's femur is directly above the middle of the foot. The end-stance subphase is associated with a situation in which the person lifts their heel off the ground. The pre-swing subphase is associated with a situation in which the person presses their toe against the ground and bends their knee to generate forward propulsion. During the stance phase of the gait cycle, the knee flexion reaches its maximum angle in the late stage of the pre-swing subphase.

[0016] In one embodiment, the specific activity is cycling, and the specific activity has a top dead center (TDC) phase, a powering phase, a bottom dead center (BDC) phase, and a recovery phase, where the TDC phase is when the cyclist pedals the bicycle and the bicycle crank passes the top dead center of the crank, the powering phase is when the cyclist applies force (i.e., extends his / her legs to cause a downward stroke) to propel the bicycle forward, the BDC phase is when the cyclist pedals the bicycle and the bicycle crank passes the bottom dead center of the crank, and the recovery phase is when the cyclist lifts his / her legs during the transition from the BDC phase to the TDC phase.

[0017] The method for establishing the decision model includes steps 201 to 203 shown in FIG. 2 and described below.

[0018] In step 201, for each training data set, the electronic device groups the sensor data entries of the training data set into multiple sensor data groups, each corresponding to a specific activity period. In this embodiment, the electronic device groups the sensor data entries of the training data set related to the stance phase into a stance phase group, and groups the sensor data entries of the training data set related to the swing phase into a swing phase group. More specifically, the electronic device uses a machine learning algorithm to obtain the stance and swing phase groups based on features (e.g., directional acceleration change or angular velocity limit) obtained from the sensor data entries and predetermined criteria related to the features for distinguishing the stance and swing phase groups. The implementation of grouping the sensor data entries of the training data set into stance and swing phase groups is well known to those skilled in the art, and therefore, for the sake of brevity, a detailed description thereof will be omitted herein.

[0019] In an embodiment in which the specific activity is cycling, the electronic device groups a plurality of sensor data entries associated with a TDC phase into a TDC phase group, a plurality of sensor data entries associated with a powering phase into a powering phase group, a plurality of sensor data entries associated with a BDC phase into a BDC phase group, and a plurality of sensor data entries associated with a recovery phase into a recovery phase group.

[0020] In step 202, the electronic device establishes a classification model to determine one of a plurality of specific types of movement of the subject based on the entries of the plurality of sensor data belonging to a target one in the plurality of sensor data groups of each training data set, and to output one of the plurality of specific types of movement as an output. In this embodiment, the target one in the plurality of sensor data groups is one of a stance phase group and a swing phase group. In an embodiment in which the specific activity is cycling, the target one in the plurality of sensor data groups is one of a TDC phase group, a power delivery phase group, a BDC phase group, and a recovery phase group.

[0021] Note that in this embodiment, the electronic device establishes the classification model directly based on the plurality of sensor data entries belonging to one of the target sensor data groups. In another embodiment, for the plurality of training datasets, the electronic device performs data processing on the plurality of sensor data entries belonging to one of the target sensor data groups to generate a plurality of processed training data entries, each corresponding to the plurality of sensor data entries. Each processed training data entry includes one of a motion trajectory associated with a body part of a lower limb corresponding to one of the sample subjects, a movement velocity associated with the lower limb corresponding to one of the sample subjects, and a sweep area associated with the swing of the lower limb corresponding to one of the sample subjects. The electronic device then establishes the classification model based on the plurality of processed training data entries for each of the plurality of training datasets.

[0022] In some embodiments, each training data set further includes physiological parameters associated with one of the corresponding sample subjects, and the electronic device establishes the classification model further based on the physiological parameters included in each training data set, such as age, sex, weight, foot shape, leg shape, etc.

[0023] The method for establishing a classification model can be carried out by three approaches, which are described below.

[0024] In a first approach, for each specific type of motion, the electronic device generates a reference probability distribution based on entries of multiple sensor data belonging to a target in multiple sensor data groups of each training data set (each training data set is related to a specific type of motion). Then, the electronic device establishes a classification model including the reference probability distribution for each of the multiple specific types of motion. An example of the reference probability distribution is, but is not limited to, a normal distribution.

[0025] In the second approach, each training data set further includes a label indicating one of a plurality of specific types of movement associated with the training data set. For example, one of the plurality of training data sets related to overpronation types includes a label indicating the overpronation type, one of the plurality of training data sets related to pronation types includes a label indicating the pronation type, and one of the plurality of training data sets related to supination types includes a label indicating the supination type. The electronic device uses a machine learning algorithm to establish a classification model further based on the labels included in the plurality of training data sets. Because the implementation of establishing a classification model using a machine learning algorithm is well known to those skilled in the art, a detailed description thereof will not be provided herein for the sake of brevity.

[0026] In the third approach, for each of the plurality of sensor data entries belonging to a target sensor data group in each of the plurality of training data sets, the electronic device acquires inward / outward acceleration and forward / backward acceleration based on the sensor data entry. The inward acceleration represents the acceleration of the medial part of the human lower limb moving toward the longitudinal axis of the human trunk, the outward acceleration represents the acceleration of the lateral part of the human lower limb moving away from the longitudinal axis of the human trunk, the forward acceleration represents the acceleration of the human lower limb in front of the human when the human bends the knee and lifts the leg, and the backward acceleration represents the acceleration of the human lower limb in back of the human when the human straightens the knee and extends the leg. The inward / outward acceleration and forward / backward acceleration can be acquired directly by an accelerometer of a measurement device. Obtaining inward / outward acceleration and forward / backward acceleration using a measurement device is well known to those skilled in the art, and therefore, for the sake of brevity, a detailed description thereof will be omitted herein. See, for example, publications such as "Body CoM acceleration for rapid analysis of gait variability and pedestrian effects on structures" by Chiara Bedon (2022), "Wearable inertial sensors to measure gait and posture characteristic differences in older adult fallers and non-fallers: A scoping review" by Patel et al. (2020), "An innovative approach of using inertial sensor data to detect abnormal gait patterns" by Patterson et al. (2012), and "A review of accelerometry-based wearable motion detectors for physical activity monitoring" by Yang et al. (2010).

[0027] For each specific type of movement, the electronic device determines a reference inward / outward interval based on the inward / outward accelerations and forward / backward accelerations obtained based on multiple sensor data entries of multiple training data sets related to the specific type of movement. More specifically, for each specific type of movement, the interval determination procedure is performed as follows: First, the electronic device selects a desired forward / backward acceleration from the multiple forward / backward accelerations thus obtained for the multiple sensor data entries, where the selected desired forward / backward acceleration is relatively large among the multiple forward / backward accelerations. For example, but not limited to, the multiple desired forward / backward accelerations are in the top 10% of the multiple forward / backward accelerations. Next, the electronic device selects a plurality of desired inward / outward accelerations from the multiple inward / outward accelerations, where the multiple desired inward / outward accelerations are obtained based on a portion of the multiple sensor data entries from which the desired forward / backward accelerations are obtained. Third, the electronic device specifies a range of the desired inward / outward accelerations as a reference inward / outward interval.

[0028] FIG. 4 illustrates example forward / backward accelerations and inward / outward accelerations for overpronation, pronation, and supination types. In the figure, a minus sign designates forward / backward accelerations as backward accelerations, and a plus sign designates forward / backward accelerations as forward accelerations. Additionally, a minus sign designates inward / outward accelerations as inward accelerations, and a plus sign designates inward / outward accelerations as outward accelerations. For overpronation types, the desired inward / outward accelerations are obtained based on a portion of the sensor data entries based on which the desired forward / backward accelerations (i.e., the largest of the top 10% of forward / backward accelerations) are obtained, and are -16, -15, -14, and -13 m / s, respectively. 2 and the desired inward / outward acceleration range, i.e., -16 to -13 m / s 2is designated as the reference inward interval. Similarly, for the pronation type, the desired inward / outward acceleration is obtained based on the portion of the sensor data entry from which the desired forward / backward acceleration (i.e., the largest of the top 10% of forward / backward accelerations) is obtained, and is -11, -10, and -9 m / s, respectively. 2 and the desired inward / outward acceleration range, i.e., -11 to -9 m / s 2 is designated as the reference inward interval. Similarly, for the supination type, the desired inward / outward acceleration is obtained based on the portion of the sensor data entry from which the desired forward / backward acceleration (i.e., the maximum of the top 10% of forward / backward accelerations) is obtained, and is -4 and -3 m / s, respectively. 2 and the desired inward / outward acceleration range, i.e., -4 to -3 m / s 2 is designated as the reference inward spacing.

[0029] In a scenario in which the stance phase of gait is further divided into a heel strike subphase, a load response subphase, a mid-stance subphase, an end-stance subphase, and a pre-swing subphase, in order to distinguish a plurality of specific types of movements according to entries of a plurality of sensor data collected in the later stage of the pre-swing subphase, a plurality of reference inward / outward intervals for the plurality of specific types of movements are determined by performing the above interval determination procedure based on the entries of the collected plurality of sensor data.

[0030] After determining the plurality of reference inward / outward directions for each of the plurality of specific types of motion, the electronic device establishes a classification model including the plurality of reference inward / outward directions.

[0031] In another embodiment, for each sensor data entry belonging to a target one of the multiple sensor data groups in each training data set, the electronics obtain an angular velocity based on the sensor data entry. For each specific type of motion, the electronics determine an angular velocity criterion (which may be a threshold or interval) based on the angular velocities obtained based on the multiple sensor data entries in the multiple training data sets associated with the specific type of motion. The electronics establish a classification model including the angular velocity criterion determined in response to the multiple specific types of motion.

[0032] In step 203, the electronic device obtains a determination model for determining a suitable foot accessory for the subject based on the output of the classification model, wherein the determination model is configured to associate a plurality of specific types of movements with a plurality of types of foot accessories (e.g., a pair of shoe insoles, a pair of shoes, a pair of socks, a foot support, etc.).

[0033] For example, the electronics further stores a lookup table including a plurality of specific types of motion and a type of foot equipment associated with each of the plurality of specific types of motion, and the electronics combines the classification model and the lookup table using an output of the classification model as an input of the lookup table to obtain a determination model for determining which type of foot equipment is appropriate for the subject.

[0034] In the case of overpronation, a person's foot tilts inward too much when landing while walking, which can cause the arch to collapse, so support insoles that stabilize the arch are suitable for overpronation. In the case of pronation, a person's foot typically tilts inward when landing while walking, so insoles with moderate arch support are suitable for pronation. In the case of supination, a person's foot tilts outward when landing while walking, and high arches require shock absorption, so insoles with excessive strain and cushioning are suitable for supination.

[0035] Referring to Figure 5, an embodiment of a method for determining a foot attachment suitable for a specific type of movement of a subject (i.e., a human) according to the present disclosure will be described. The method is executed by a user device 11 shown in Figure 1. The user device 11 is electrically connected to a sensor device 12 attached to the subject's lower limb. Based on the detection results, the sensor device 12 generates multiple detection data entries, each at multiple time points when the subject performs a specific activity. The user device 11 stores a determination model established by the aforementioned electronic device for determining one of multiple types of foot attachments for the subject (each type of foot attachment is associated with a specific type of movement).

[0036] The user device 11 can be realized by, but is not limited to, a smartphone or a personal computer (e.g., a tablet computer, a desktop computer, a notebook computer, etc.). The sensor device 12 can be implemented to incorporate an IMU and a magnetometer. The electrical connection between the user device 11 and the sensor device 12 can be implemented by Bluetooth® wireless technology, Wi-Fi technology, RFID, etc. The user device 11 can be implemented the same as or different from the aforementioned electronic devices, and the sensor device 12 can be implemented the same as or different from the aforementioned measurement devices.

[0037] The method includes steps 301 to 306 shown below.

[0038] In step 301, the user device 11 groups a plurality of detection data entries into a plurality of detection data groups, each corresponding to a specific activity period. Specifically, in this embodiment, the user device 11 groups a plurality of detection data entries related to the stance phase into a stance phase group, and groups a plurality of detection data entries related to the swing phase into a swing phase group. A target among the plurality of detection data groups is one of the stance phase group and the swing phase group.

[0039] In an embodiment where the specific activity is cycling, the user device 11 groups a plurality of sensed data entries related to the TDC phase into a TDC phase group, a plurality of sensed data entries related to the powered phase into a powered phase group, a plurality of sensed data entries related to the BDC phase into a BDC phase group, and a plurality of sensed data entries related to the recovery phase into a recovery phase group, wherein a target one of the plurality of sensed data groups is one of the TDC phase group, the powered phase group, the BDC phase group, and the recovery phase group.

[0040] In step 302, the user device 11 determines one of a plurality of specific types of motion associated with the object as a target type using a determination model based on the plurality of detection data entries belonging to one of the plurality of detection data groups that is a target.

[0041] In this embodiment, the user device 11 determines the target type using a classification model directly based on the detection data entries belonging to one of the plurality of detection data groups that is the target. In another embodiment, the user device 11 performs data processing on the plurality of detection data entries belonging to one of the plurality of detection data groups that is the target to generate a plurality of processed detection data entries, each corresponding to the plurality of detection data entries. The plurality of processed detection data entries include one of a movement trajectory associated with the body part of the target's lower limb, a movement velocity associated with the target's lower limb, and a sweep area associated with the swing of the target's lower limb. The user device 11 determines the target type using a decision model based on the plurality of processed detection data entries.

[0042] In some embodiments where the classification model is established further based on physiological parameters associated with each of the multiple sample subjects, the user device 11 determines the target type using a determination model further based on physiological parameters associated with the subjects (e.g., age, gender, weight, foot shape, leg shape, etc.).

[0043] The method of determining target type using a classification model can be implemented by three approaches, which are described below.

[0044] In the first approach, the determination model includes multiple reference probability distributions for each specific type of motion. The user device 11 generates a detection probability distribution based on multiple detection data entries belonging to one of the target detection data groups. The user device 11 then selects one of the multiple reference probability distributions that is most similar to the detection probability distribution as a target distribution, and determines the target type corresponding to the target distribution.

[0045] In the second approach, the classification model is established by a machine learning algorithm further based on the label, and the user device 11 inputs the entries of the plurality of detection data belonging to one of the plurality of detection data groups as a target into the classification model, and obtains the output of the classification model as the target type.

[0046] In a third approach, the judgment model includes a plurality of reference inward / outward intervals, each representing a plurality of specific types of motion. For each of a plurality of detection data entries belonging to a target in the plurality of detection data groups, the user device 11 acquires inward / outward accelerations and forward / backward accelerations based on the detection data entries. The user device 11 selects a plurality of target forward / backward accelerations from the plurality of forward / backward accelerations acquired by the plurality of detection data entries, and the selected plurality of target forward / backward accelerations satisfy a statistical characteristic of the plurality of forward / backward accelerations. In this embodiment, the statistical characteristic is that the plurality of target forward / backward accelerations are the largest of the top 10% of the plurality of forward / backward accelerations. Furthermore, the user device 11 selects a plurality of target inward / outward accelerations from the plurality of inward / outward accelerations, and the plurality of target inward / outward accelerations are acquired based on a portion of the detection data entries, where the plurality of target inward / outward accelerations are acquired based on a portion of the detection data entries based on which the plurality of target forward / backward accelerations are acquired. Then, the user device 11 calculates an average of the multiple target inward / outward accelerations, and selects one of the multiple reference inward / outward intervals that includes the average of the multiple target inward / outward accelerations as the target interval. The user device 11 determines a target type corresponding to the target interval.

[0047] As shown in Figure 4, the reference inward interval for the overpronation type was -16 to -13 cm / s. 2 and the reference inward interval for the type of pronation is -11 to -9 cm / s 2 and the reference inward interval for the type of supination is -4 to -3 cm / s 2 In the scenario, the average target inward / outward acceleration is -13 cm / s 2 When calculated, the user device 11 is -16 to -13 cm / s 2The reference inward spacing is determined as the target spacing, and the type of overpronation is determined as the target type.

[0048] In one embodiment in which the classification model includes an angular velocity criterion, for each detection data entry belonging to one of the multiple detection data groups that is a target, the user device 11 obtains an angular velocity based on the detection data entry. The user device 11 selects multiple target angular velocities for the detection data entry from the multiple angular velocities thus obtained, where the selected target angular velocities satisfy a statistical characteristic of the angular velocities. In this embodiment, the statistical characteristic is, but is not limited to, the multiple target angular velocities being the largest of the top 10% of the multiple angular velocities. The user device 11 then calculates an average of the multiple angular velocities and selects one of the angular velocity criteria that matches the average of the target angular velocity (e.g., the average of the target angular velocity is within the interval of the angular velocity criterion) as the target criterion. The user device 11 determines a target type corresponding to the target criterion.

[0049] In step 303, the user device 11 uses the judgment model to determine one of a plurality of types of foot attachments that matches the target type (hereinafter also referred to as the target foot attachment).

[0050] In step 304, the sensor device 12 further generates, based on the detection results, a plurality of fitting data entries, each of which corresponds to a plurality of time points when the subject wearing the target foot attachment was performing a particular activity.

[0051] In step 305, the user device 11 determines whether the target foot attachment is suitable for the subject based on the plurality of fitting data entries and the plurality of detection data entries. If it is determined that the target foot attachment is suitable for the subject, the method flow proceeds to step 306. On the other hand, if it is determined that the target foot attachment is not suitable for the subject, the method flow ends.

[0052] Specifically, the user device 11 obtains first biomechanical data based on multiple detection data entries and second biomechanical data based on multiple fitting data entries, compares the first biomechanical data with the second biomechanical data, and determines whether the target foot garment is suitable for the subject based on the comparison. Each of the first biomechanical data and the second biomechanical data includes at least one of the subject's lower limb sway area, the subject's postural stability, the ratio of one leg to the other leg, and the symmetry of the subject's two legs. In a scenario where the specific activity is walking, the lower limb sway area represents the degree of sway of the lower limb when walking or stepping. The smaller the lower limb sway area of ​​a person, the less the person sway, thereby enabling the person to walk better. The postural stability represents a person's ability to consistently maintain the same movement. The higher the postural stability, the more consistent (i.e., less fluctuation) the person's movement, thereby indicating a better gait. Therefore, when the user device 11 determines based on the comparison result that the sway area of ​​the subject's lower limbs has decreased after the subject wears the target foot attachment, the user device 11 determines that the target foot attachment is suitable for the subject. Since obtaining biomechanical data based on the entry of multiple detection data or multiple fitting data is well known to those familiar with the relevant art, a detailed description thereof will be omitted herein for the sake of brevity. See, for example, publications such as "Construct Validity of A Wearable Inertial Measurement Unit (IMU) in Measuring Postural Sway and the Effect of Visual Deprivation in Healthy Older Adults" by Luca Ferrari, Gianluca Bochicchio, Alberto Bottari, Alessandra Scarton, Francesco Lucertini, and Silvia Pogliaghi, 2024.

[0053] In step 306, the user device 11 designates the multiple detection data entries as a new training dataset for the target type. If one foot attachment that matches the target type among the multiple types of foot attachments is suitable for the subject, it means that the judgment result made by the user device 11 in step 302 (i.e., the judgment of the subject's target type based on the multiple detection data entries) was appropriate and correct, so the multiple detection data entries may be useful as material for training a judgment model or material for enriching a training database (e.g., merging the new training dataset with the multiple training datasets originally stored in the electronic device).

[0054] In some embodiments of this method, steps 304 through 306 may be omitted.

[0055] In summary, in the method for establishing a judgment model and the method for identifying foot equipment, the entries of multiple sensor data in the training dataset are first grouped into sensor data groups according to the timing of specific activities, and then a classification model is established based on a target one of the multiple sensor data groups. In this way, a specific type of movement of a subject can be identified using the classification model. Computational resources for establishing the classification model can be saved, and the robustness of the classification model can be improved. Furthermore, the classification model is combined with a lookup table to obtain a judgment model, and the subject's foot equipment can be appropriately and reliably identified using the judgment model.

[0056] For purposes of explanation, numerous specific details have been set forth above to facilitate a thorough understanding of the embodiments. However, it will be apparent to one skilled in the art that one or more other embodiments may be practiced without these specific details. Furthermore, in the description of "one embodiment" or "an embodiment" herein, all references to "one embodiment" or "an embodiment" accompanied by ordinal or other designations should be understood to encompass specific implementations of the present invention having particular aspects, structures, or features. Furthermore, although multiple variations may be incorporated into a single embodiment, drawing, or description thereof, this is for the purpose of streamlining the description and for the purpose of understanding the multifaceted aspects of the present invention. Furthermore, one or more features or specific embodiments of one embodiment may, where appropriate, be combined with one or more features or specific embodiments of other embodiments in the implementation of the present disclosure.

[0057] Although the present disclosure has been described in connection with exemplary embodiments, it is understood that the disclosure is not limited to the disclosed embodiments, but is intended to cover various configurations falling within the broadest spirit and scope, and to embrace all such modifications and equivalent arrangements. [Explanation of symbols]

[0058] 11 User equipment 12 Sensor device Steps 201-203, 301-306

Claims

1. 1. A method for establishing a judgment model for determining foot equipment suitable for a specific type of movement of a subject, the method being executed by an electronic device, the electronic device storing a plurality of training data sets corresponding to a plurality of sample subjects, each of the plurality of training data sets being related to one of a plurality of specific types of movement and including a plurality of sensor data entries, the plurality of sensor data entries being generated for a corresponding one of the sample subjects based on detection of the sample subject's movement, and each of the plurality of sensor data entries being at a plurality of time points when the sample subject is performing a specific activity, the specific activity having a plurality of periods; The method comprises: for each of the plurality of training data sets, grouping the plurality of sensor data entries of the training data set into a plurality of sensor data groups each corresponding to a time period of the particular activity; establishing a classification model for determining one of the plurality of specific types of motions of the subject based on entries of the plurality of sensor data belonging to a target in the plurality of sensor data groups of each of the plurality of training data sets and outputting the one of the plurality of specific types of motions as an output; and A method for establishing a judgment model, comprising: obtaining a judgment model for determining a foot accessory suitable for the subject based on the output of the classification model, the judgment model being configured to associate the plurality of specific types of movements with a plurality of types of foot accessories.

2. the specific activity is walking and has a stance phase and a swing phase; 2. The method of claim 1, wherein grouping the plurality of sensor data entries of the training data set into a plurality of sensor data groups includes grouping the plurality of sensor data entries of the training data set related to the stance phase into a stance phase group and grouping the plurality of sensor data entries of the training data set related to the swing phase into a swing phase group, and wherein in establishing the classification model, a target one of the plurality of sensor data groups is one of the stance phase group and the swing phase group.

3. the specific activity is riding a bicycle and has a top dead center (TDC) phase, a power delivery phase, a bottom dead center (BDC) phase, and a recovery phase; 2. The method of claim 1, wherein grouping the plurality of sensor data entries of the training dataset into a plurality of sensor data groups includes grouping the plurality of sensor data entries related to the top dead center period into a top dead center period group, grouping the plurality of sensor data entries related to the power supply period into a power supply period group, grouping the plurality of sensor data entries related to the bottom dead center period into a bottom dead center period group, and grouping the plurality of sensor data entries related to the recovery period into a recovery period group, and wherein in establishing the classification model, a target one of the plurality of sensor data groups is one of the top dead center period group, the power supply period group, the bottom dead center period group, and the recovery period group.

4. For each of the plurality of specific types of motion, generating a reference probability distribution based on entries of the plurality of sensor data belonging to a target one of the plurality of sensor data groups of each of the plurality of training data sets, wherein the plurality of training data sets are associated with the specific types of motion; The method according to any one of claims 1 to 3, wherein establishing the classification model comprises establishing the classification model including the reference probability distribution for each of the specific types of motion.

5. each of the plurality of training data sets includes a label indicating one of the plurality of particular types of motion to which the training data set relates; 4. The method of claim 1, wherein establishing the classification model comprises establishing the classification model using a machine learning algorithm based on the labels included in each of the plurality of training datasets.

6. For each of the plurality of sensor data entries of each of the plurality of training data sets, obtaining an inward / outward acceleration and a forward / backward acceleration based on the sensor data entry; and determining a plurality of reference inward / outward intervals for the plurality of specific types of motions based on the inward / outward accelerations and the forward / backward accelerations respectively acquired for the plurality of sensor data entries of each of the plurality of training data sets; The method according to any one of claims 1 to 3, wherein establishing the classification model comprises establishing the classification model including the plurality of reference inward / outward intervals.

7. further comprising: for each of the plurality of training data sets, performing data processing on the plurality of sensor data entries belonging to a target one of the plurality of sensor data groups to generate a plurality of processed training data entries, each corresponding to the plurality of sensor data entries; wherein each of the plurality of processed training data entries includes one of: a motion trajectory associated with a lower limb body part corresponding to one of the sample subjects; a movement velocity associated with a lower limb corresponding to one of the sample subjects; and a sweep area associated with a swing of a lower limb corresponding to one of the sample subjects; 7. The method of claim 1, wherein establishing the classification model comprises establishing the classification model based on entries of the plurality of processed training data for each of the plurality of training data sets.

8. each of the plurality of training data sets further comprising a physiological parameter associated with a corresponding one of the sample subjects; The method according to any one of claims 1 to 7, wherein establishing the classification model comprises establishing the classification model based on the physiological parameters of each of the plurality of training data sets.

9. 1. A method for determining a foot garment suitable for a specific type of movement of a subject, the method being performed by a user device, the user device being electrically connected to a sensor device attached to the subject's lower leg, the sensor device generating a plurality of detection data entries based on detection results, each entry corresponding to a plurality of time points when the subject is performing a specific activity, the specific activity having a plurality of periods, the user device storing a determination model obtained by the method of claim 1; The method comprises: grouping the plurality of sensory data entries into a plurality of sensory data groups, each group corresponding to a time period of the particular activity; determining one of the specific types of motions associated with the object as a target type based on an entry of the plurality of detection data belonging to one of the plurality of detection data groups that is a target using the determination model; and A method for determining foot equipment, comprising: using the determination model to determine one of a plurality of types of foot equipment that matches the target type.

10. the decision model includes a plurality of reference probability distributions, each of which corresponds to a specific type of motion; The method comprises: generating a detection probability distribution based on the plurality of detection data entries belonging to one of the plurality of detection data groups that is a target; selecting one of the plurality of reference probability distributions that is most similar to the detection probability distribution as a target distribution; The method of claim 9 , wherein determining a target type comprises determining the target type corresponding to the target distribution.

11. the determination model includes a plurality of reference inward / outward intervals for each of the plurality of specific types of movement; The method comprises: For each of the plurality of detection data entries belonging to one of the plurality of detection data groups that is a target, obtaining inward / outward accelerations and forward / backward accelerations based on the detection data entries; selecting a plurality of target forward / rearward accelerations from the plurality of forward / rearward accelerations obtained by the entry of the plurality of detection data, wherein the selected plurality of target forward / rearward accelerations satisfy a statistical characteristic of the plurality of forward / rearward accelerations; selecting a plurality of target inward / outward accelerations from the plurality of inward / outward accelerations, the plurality of target forward / rearward accelerations being obtained based on a portion of the detection data entries from which the plurality of target forward / rearward accelerations were obtained; calculating an average of the plurality of target inward / outward accelerations; selecting one of the plurality of reference inward / outward intervals as a target interval, the reference inward / outward interval including an average of the plurality of target inward / outward accelerations; The method of claim 9 , wherein determining a target type comprises determining the target type corresponding to the target interval.

12. further comprising: performing data processing on the plurality of detection data entries belonging to one of the plurality of detection data groups that is a target; and generating a plurality of post-processing detection data entries respectively corresponding to the plurality of detection data entries; wherein each of the plurality of processed detection data entries includes one of a motion trajectory associated with a limb portion of the subject's lower limb, a movement velocity associated with the subject's lower limb, and a sweep area associated with a swing of the subject's lower limb; The method according to any one of claims 9 to 11, wherein determining a target type comprises determining the target type using the determination model based on the plurality of entries of processed detection data.

13. The method according to any one of claims 9 to 12, wherein determining the target type comprises determining the target type using the determination model further based on physiological parameters associated with the subject.

14. the specific activity is walking and has a stance phase and a swing phase; The method according to any one of claims 9 to 13, wherein grouping the plurality of detection data entries into a plurality of detection data groups includes grouping the plurality of detection data entries related to the stance phase into a stance phase group, and grouping the plurality of detection data entries related to the swing phase into a swing phase group, and in determining a target type, one of the targets in the plurality of detection data groups is one of the stance phase group and the swing phase group.

15. the specific activity is riding a bicycle and has a top dead center (TDC) phase, a power delivery phase, a bottom dead center (BDC) phase, and a recovery phase; The method of any one of claims 9 to 13, wherein grouping the plurality of detection data entries into a plurality of detection data groups includes grouping the plurality of detection data entries related to the top dead center period into a top dead center period group, grouping the plurality of detection data entries related to the power supply period into a power supply period group, grouping the plurality of detection data entries related to the bottom dead center period into a bottom dead center period group, and grouping the plurality of detection data entries related to the recovery period into a recovery period group, and wherein in determining a target type, one of the targets in the plurality of detection data groups is one of the top dead center period group, the power supply period group, the bottom dead center period group, and the recovery period group.

16. the sensor device further generates a plurality of fitting data entries based on the detection results, the plurality of fitting data entries being at a plurality of time points when the subject wearing one of a plurality of types of foot attachments that matches the target type performs the specific activity; The method comprises: determining whether one of the plurality of types of foot attachments that matches the target type is suitable for the subject based on at least the plurality of fitting data entries; and The method of claim 9, further comprising: when it is determined that one of the plurality of types of foot attachments that matches the target type is suitable for the subject, designating the plurality of detection data entries as a new training data set for the target type.

Images