Method and system for automatically evaluating movement smoothness of infant

Abstract

A method for automatically evaluating the smoothness of voluntary movements of an infant, according to one embodiment of the present invention, comprises the steps of: obtaining a two-dimensional video of the infant who is moving voluntarily; obtaining position coordinate data of joints of the infant for each frame of the video by using a deep learning-based pose prediction model; and quantifying the smoothness of the voluntary movements of the infant on the basis of the position coordinate data of the joints.

Description

Method and system for automatically evaluating infant movement smoothness

[0001] The present invention relates to a method and system for automatically evaluating the smoothness of movement of an infant.

[0002] Cerebral palsy is a general term for motor dysfunction caused by brain lesions during the immature period before, during, or after birth. Primary symptoms include impaired muscle control and motor function, as well as delayed motor development. To minimize potential future disabilities, early detection of cerebral palsy or motor developmental delay and early, appropriate therapeutic intervention are crucial.

[0003] Traditionally, the General Movement Assessment (GMAA), developed by Pretzel, has been recommended as a method for predicting the risk of cerebral palsy or motor developmental delay in infants, particularly premature infants. However, conventional GMAA methods rely on videotaped images of infants' voluntary movements, which are then directly evaluated and scored by trained, certified evaluators. This approach can be time-consuming, depending on the evaluator's proficiency, and is highly dependent on the evaluator's skills. Furthermore, quantitative assessments are difficult.

[0004] Although engineers have attempted to quantitatively assess voluntary movement in the past, no successful assessment method has been developed that can be used clinically, and it remains at the research level.

[0005] Meanwhile, the development of new parameters significantly associated with the development of later motor developmental delays in infants is crucial as a preliminary step toward the development of automated motor developmental delay prediction programs. Discovering new parameters associated with the risk of cerebral palsy or motor developmental delay and quantifying them using an automated, machine learning-based approach can provide additional information that can inform clinical decision-making.

[0006] In other words, there is a need for an effective automatic movement assessment method and system that can predict cerebral palsy or motor development delay in infants at an early stage by quantifying new parameters associated with cerebral palsy or motor development delay in infants using a quantitative method based on machine learning.

[0007] The technical task of the present invention is to resolve the problems of the prior art described above. The technical task to be achieved by this embodiment is not limited to the technical task described above, and other technical tasks can be inferred from the following examples.

[0008] As a technical means for achieving the above-described technical task, a method for automatically evaluating the smoothness of voluntary movements of an infant according to one embodiment may include the steps of: obtaining a two-dimensional video of the infant moving voluntarily; obtaining position coordinate data of joints of the infant for each frame of the video using a deep learning-based pose prediction model; and quantifying the smoothness of voluntary movements of the infant based on the position coordinate data of the joints.

[0009] The method may further include, after the step of obtaining the position coordinate data of the joint, the step of obtaining angle data of the joint and angular velocity data of the joint based on the position coordinate data of the joint.

[0010] The step of quantifying the smoothness of the movement may include the step of quantifying the smoothness of the voluntary movement of the infant based on the angle data of the joint and the angular velocity data of the joint.

[0011] The step of quantifying the smoothness of the movement may include the step of measuring log dimensionless jerk (LDJ) and spectral arc length (SPARC) for the angle data of the joint and the angular velocity data of the joint.

[0012] The position coordinate data of the above joint may include position coordinate data of at least one of a left shoulder joint, a right shoulder joint, a left elbow joint, a right elbow joint, a left wrist joint, a right wrist joint, a left hip joint, a right hip joint, a left knee joint, a right knee joint, a left ankle joint, and a right ankle joint.

[0013] According to another embodiment, an automatic evaluation system for the smoothness of spontaneous movements of an infant comprises: a memory; and at least one processor connected to the memory, wherein the at least one processor is configured to: receive a two-dimensional video of the infant moving spontaneously; generate position coordinate data of a joint of the infant for each frame of the video using a deep learning-based pose prediction model; and quantify the smoothness of the spontaneous movements of the infant based on the position coordinate data of the joint.

[0014] The at least one processor may be further configured to generate angle data of the joint and angular velocity data of the joint based on the position coordinate data of the joint after generating the position coordinate data.

[0015] The at least one processor may be configured to quantify the smoothness of voluntary movements of the infant based on the angle data of the joint and the angular velocity data of the joint.

[0016] The at least one processor may be configured to measure log dimensionless jerk (LDJ) and spectral arc length (SPARC) for the angle data of the joint and the angular velocity data of the joint.

[0017] The position coordinate data of the above joint may include position coordinate data of at least one of a left shoulder joint, a right shoulder joint, a left elbow joint, a right elbow joint, a left wrist joint, a right wrist joint, a left hip joint, a right hip joint, a left knee joint, a right knee joint, a left ankle joint, and a right ankle joint.

[0018] According to another embodiment, a computer-readable storage medium can record a program for performing a method according to the above embodiments.

[0019] The automatic evaluation method and system for the smoothness of spontaneous movements of infants of the present invention can evaluate the risk of cerebral palsy or developmental delay in infants simply by recording a video of the infant moving spontaneously.

[0020] The method and system for automatically evaluating the smoothness of spontaneous movements of infants of the present invention have discovered a new parameter, smoothness, associated with the risk of cerebral palsy or developmental delay in infants, and by quantifying the smoothness of the voluntary movements of infants in an automated manner based on machine learning, the risk of cerebral palsy or developmental delay in infants can be quantitatively evaluated regardless of the proficiency of the evaluator.

[0021] According to the present invention, an effective automatic evaluation method and system for the smoothness of voluntary movements of infants can be provided, which can predict cerebral palsy or developmental delay in infants at an early stage using a quantitative method based on machine learning.

[0022] The effects of the present invention are not limited to the effects described above, and effects not mentioned can be clearly understood by a person having ordinary skill in the technical field to which the present invention belongs from this specification and the attached drawings.

[0023] Figure 1 is a diagram showing a series of processes for obtaining angle data and angular velocity data of joints of an infant.

[0024] Figure 2 is a diagram showing spectral arc length measurements based on angle data and angular velocity data of the shoulder joint and knee joint of infants in the normal group and the motor development delay group.

[0025] Figure 3 is a diagram illustrating an automatic method for evaluating the smoothness of an infant's spontaneous movements.

[0026] Figure 4 is a diagram showing an automatic evaluation system for the smoothness of infants' spontaneous movements.

[0027] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings. The advantages and features of the present invention, and methods for achieving them, will become clear with reference to the embodiments described in detail below together with the attached drawings. However, the present invention is not limited to the embodiments disclosed below, but can be implemented in various different forms. These embodiments are provided only to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the present invention of the scope of the invention, and the present invention is defined only by the scope of the claims. Like reference numerals refer to like elements throughout the specification.

[0028] Unless otherwise defined, all terms (including technical and scientific terms) used herein may be used in their common sense by those of ordinary skill in the art to which the present invention pertains. Furthermore, terms defined in commonly used dictionaries are not to be interpreted ideally or excessively unless explicitly and specifically defined otherwise.

[0029] The terminology used herein is for the purpose of describing embodiments and is not intended to limit the present invention. In this specification, singular forms also include plural forms, unless specifically stated otherwise. As used herein, the terms "comprises" and / or "comprising" do not exclude the presence or addition of one or more other components in addition to the components mentioned.

[0030] Unless stated otherwise herein, “connected” or “coupled” may include one element / feature being directly connected or coupled to another element / feature or indirectly connected or coupled through another element / feature, and does not necessarily mean directly mechanically connected or coupled. Accordingly, while the various schematic diagrams depicted in the drawings illustrate exemplary arrangements of elements and components, additional intervening elements, devices, features, or components may be present in actual embodiments (assuming the functionality of the depicted elements is not adversely affected).

[0031] Additionally, in this specification, “transmitting” or “receiving” may include not only transmitting or receiving information directly between a sender and a receiver, but also transmitting or receiving information through another entity, unless otherwise stated.

[0032] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[0033] Figure 1 is a diagram showing a series of processes for obtaining angle data and angular velocity data of joints of an infant.

[0034] First, Figure 1A represents a two-dimensional video of an infant moving spontaneously. The two-dimensional video of an infant moving spontaneously can be captured by a smartphone or a camera. It may be desirable for the two-dimensional video of an infant moving spontaneously to include the infant's head, arms, and legs all within the video frame. In the two-dimensional video of an infant moving spontaneously, it may be desirable for the infant to be lying upright, wearing little clothing, and fully awake. Furthermore, it may be desirable for the infant to be in a comfortable, quiet, undisturbed environment at a normal temperature, and to be able to move its arms, legs, and torso freely. The two-dimensional video of an infant moving spontaneously may be at least three minutes long, and preferably three to five minutes long.

[0035] Next, Fig. 1B illustrates obtaining position coordinate data of an infant's joints for each frame of a video using a deep learning-based pose prediction model. The present invention can use AlphaPose, a deep learning-based pose estimation algorithm, as the pose prediction model. AlPhaPose is a pre-trained pose prediction model based on a CNN (Convolutional Neural Network) architecture. When a 2D video of an infant moving spontaneously is input into AlPhaPose, AlPhaPose can estimate the infant's posture in real time and automatically obtain position coordinate data of the infant's joints for each frame of the video.

[0036] Finally, C of FIG. 1 illustrates acquiring angle data and angular velocity data of the joints of an infant based on position coordinate data of the joints of the infant acquired for each frame of the video. The present invention can automatically acquire angle data and angular velocity data of the joints of the shoulders (S), elbows (E), wrists (W), hips (H), knees (K), and ankles (A) of both sides of an infant based on position coordinate data of the joints of the shoulders (S), elbows (E), hips (H), and ankles (A) of both sides.

[0037] The present invention can automatically obtain angle data and angular velocity data of joints of an infant by using angle-related functions of R. The angle-related functions of R of the present invention are functions provided in the R programming language and can perform angle-related measurements.

[0038] The joint angle of the present invention may refer to the angle formed by the positional coordinates of two adjacent joints centered around a specific joint. For example, the left shoulder joint angle may refer to the angle formed by the right shoulder joint and the left elbow joint centered around the left shoulder joint. The right shoulder joint angle may refer to the angle formed by the left shoulder joint and the right elbow joint centered around the right shoulder joint. The left elbow joint angle may refer to the angle formed by the left shoulder joint and the left wrist joint centered around the left elbow joint. The right elbow joint angle may refer to the angle formed by the right shoulder joint and the right wrist joint centered around the right elbow joint. The left hip joint angle may refer to the angle formed by the right hip joint and the left knee joint centered around the left hip joint. The left knee joint angle may refer to the angle formed by the left hip joint and the left ankle joint centered around the left knee joint. The right knee joint angle may refer to the angle formed by the right hip joint and the right ankle joint centered around the right knee joint.

[0039] The joint angular velocity of the present invention may represent the rate of change of joint angle over time. For example, the angular velocity of the left shoulder joint may represent the rate of change of the left shoulder joint over time. The angular velocity of the right shoulder joint may represent the rate of change of the right shoulder joint over time. The angular velocity of the left elbow joint may represent the rate of change of the left elbow joint over time. The angular velocity of the right elbow joint may represent the rate of change of the right elbow joint over time. The angular velocity of the left hip joint may represent the rate of change of the left hip joint over time. The angular velocity of the right hip joint may represent the rate of change of the right hip joint over time. The angular velocity of the left knee joint may represent the rate of change of the left knee joint over time. The angular velocity of the right knee joint may represent the rate of change of the right knee joint over time.

[0040] Previously, there have been few attempts to evaluate spontaneous movements using joint angle data and joint angular velocity data rather than joint position coordinate data of infants. Position coordinates are greatly affected by the height and angle of the camera relative to the object, but angle and angular velocity are relatively less affected. Therefore, in the present invention, log dimensionless jerk (LDJ) and spectral arc length (SPARC) were measured based on the angle data and joint angular velocity data of the infant's joints.

[0041] The LDJ of the present invention can be defined by the following mathematical equations 1 and 2 (see Balasubramanian et al., 2012). First, mathematical equation 1 is as follows.

[0042] [Mathematical Formula 1]

[0043]

[0044] The dimensionless jerk (DLJ) of the present invention can represent the rate of change in acceleration of a movement. Referring to mathematical expression 1, the DLJ of the present invention can be defined as a value obtained by multiplying the square of the rate of change in acceleration of a movement by the cube of the duration of the trial, and then taking the negative of the value obtained by dividing the result by the square of the maximum velocity of the joint angle.

[0045] Mathematical formula 2 is as follows.

[0046] [Equation 2]

[0047]

[0048] Referring to mathematical equation 2, the LDJ of the present invention can be defined as a value obtained by taking the negative logarithm of the DLJ (Gulde and Hermsdφrfer, 2018). For example, in the present invention, Python (version 3.10; Python Software Foundation, Wilmington, DE, USA) can be used to measure the LDJ.

[0049] Because LDJ is more sensitive to small changes in movement than DLJ, it may be preferable to use LDJ rather than DLJ as an indicator for assessing infants' voluntary movements (see: Balasubramanian et al., 2015; Hogan and Sternad, 2009).

[0050] The SPARC of the present invention can be defined by the following mathematical formulas 3 and 4 (see: Balasubramanian et al., 2012). Mathematical formula 3 is as follows.

[0051] [Equation 3]

[0052]

[0053] In mathematical expression 3, refers to the Fourier magnitude spectrum in the frequency domain of the movement, may mean cut-off frequency. can be defined by the following mathematical formula 4.

[0054] [Equation 4]

[0055]

[0056] That is, the SPARC of the present invention and , and SPARC may mean the arc length of the Fourier magnitude spectrum. For example, the present invention may use Python (version 3.10; Python Software Foundation, Wilmington, DE, USA) to measure SPARC.

[0057] Below, we describe a specific experimental example for quantifying the smoothness of an infant's voluntary movements based on measured LDJ and SPARC for the angle data and angular velocity data of the infant's joints, and thereby assessing the risk of cerebral palsy or motor development delay in the infant.

[0058] In this invention, a total of 111 infants were recruited, either prematurely born at less than 32 weeks of age or weighing less than 1500 g. However, infants with genetic syndromes or major congenital malformations were excluded.

[0059] In the present invention, two-dimensional videos of infants spontaneously moving at a specific postnatal age were first recorded. However, for premature infants, the term-equivalent age (TEA) was adjusted based on the expected date of birth rather than the actual postnatal age to ensure a more accurate assessment of the risk of cerebral palsy or motor developmental delay. Next, LDJ and SPARC were measured for the infants' joint angle data and joint velocity data, and the results are shown in Table 1 below.

[0060]

[0061] Table 1 shows the mean values ​​(standard deviations) of LDJ and SPARC measured at the joints of the right shoulder, left shoulder, right elbow, left elbow, right hip, left hip, right knee, and left knee of 100 infants in the normal group without motor developmental delay and 11 infants in the group with motor developmental delay. The p-value is a concept used in statistical hypothesis testing, and if the p-value is less than the significance level (e.g., 0.05), it means that the observed result is unlikely to have occurred by chance.

[0062] Referring to Table 1 above, the LDJ in the joints of the right shoulder, right elbow, left elbow, right hip, and right knee of infants in the group with motor development delay was measured as -22.97, 23.04, 23.05, -22.99, and -23.02, respectively, and the LDJ in the joints of the right shoulder, right elbow, left elbow, right hip, and right knee of infants in the normal group was measured as -22.02, -22.02, -22.04, -22.15, and -22.08, respectively. In other words, the LDJ of infants in the group with motor development delay was measured to be greater than that of infants in the normal group in most joints. This indicates that the LDJ measured in the joints of infants shows a significant positive correlation with the possibility of developing motor development delay in the infants in the future.

[0063] In addition, the SPARC in the joints of the right shoulder, right hip, right knee, and left knee of the infants in the group with motor development delay were measured as -33.91, -33.46 -32.86, and -33.84, respectively, and the SPARC in the right shoulder, right hip, right knee, and left knee of the infants in the normal group were measured as -27.09, -27.63, -26.89, and -25.25, respectively. In other words, the SPARC of the infants in the group with motor development delay was measured to be greater than that of the infants in the normal group in the joints of the right shoulder, right hip, right knee, and left knee. This indicates that the SPARC in the joints of infants also shows a significant positive correlation with the possibility of developing motor development delay in the infants in the future.

[0064] These characteristics can also be seen in Fig. 2. Fig. 2 shows the SPARC in the shoulder and knee joints of infants in the normal group and infants with motor development delay. Referring to Fig. 2, the smoothness of movement in the shoulder and knee joints of infants in the motor development delay group was confirmed to be greater than the smoothness of movement in the shoulder and knee joints of infants in the normal group. In other words, it was found that the smoothness (LDJ and SPARC) in the joints of infants in the motor development delay group was greater than the smoothness (LDJ and SPARC) in the joints of infants in the normal group.

[0065] Furthermore, in the present invention, in order to confirm whether the smoothness of the infant's voluntary movements is an accurate indicator of the risk of developing motor developmental delay in the infant, a prospective longitudinal cohort study was conducted on 111 infants as the subjects of the experiment. Specifically, when the 111 infants were 9 months old at the adjusted age, their motor composite scores were measured using a motor development test kit defined in the Bayley Scales of Infant and Toddler Development, Third Edition (BSID-Ⅲ). In the present invention, it was confirmed that the motor composite scores based on the Bayley Scales of Infant and Toddler Development, Third Edition of the infants in the group with motor developmental delay were 80 points or less when rehabilitation support was provided.

[0066] Table 2 below shows the correlation between the LDJ and SPARC measured at the infants' joints and the infant movement composite score based on the Bayley Scales of Infant and Toddler Development, Third Edition. The correlation coefficients in Table 2 are Pearson correlation coefficients or Spearman correlation coefficients, which can be obtained by analyzing the correlation between the absolute values ​​of LDJ and SPARC measured at the infants' joints and the infant movement composite score based on the Bayley Scales of Infant and Toddler Development, Third Edition. These statistical analyses were performed using SPSS software (version 26; SPSS Inc., Chicago, IL, USA). The p-value is a concept used in statistical hypothesis testing. If the p-value is less than the significance level (e.g., 0.05), it means that the observed result is unlikely to have occurred by chance.

[0067]

[0068] Referring to Table 2 above, the absolute value of LDJ at the infant's right shoulder joint and the infant's movement composite score based on the Bailey Scales of Development, 3rd Edition had a correlation coefficient of 0.208, and the absolute value of SPARC at the infant's right shoulder, right hip, right knee, and left knee joints and the infant's movement composite score based on the Bailey Scales of Development, 3rd Edition had correlation coefficients of 0.318, 0.211, 0.234, and 0.219, respectively. This shows that the larger the absolute value of the smoothness of movement (LDJ and SPARC) at the infant's joints, the higher the infant's movement composite score based on the Bailey Scales of Development, 3rd Edition. In other words, the absolute value of the smoothness of infant's voluntary movements (LDJ and SPARC) has a positive correlation with the infant's movement composite score based on the Bailey Scales of Development, 3rd Edition, at adjusted age of 9 months. In this way, the present invention found that the smoothness of an infant's voluntary movements is an important parameter associated with the risk of delayed motor development in the infant.

[0069] Figure 3 is a drawing showing an automatic method for evaluating the smoothness of spontaneous movements of an infant according to the present invention.

[0070] Referring to FIG. 3, a method for automatically evaluating the smoothness of voluntary movements of an infant may include the steps of (S1) obtaining a two-dimensional video of the infant moving voluntarily; (S2) obtaining position coordinate data of the joints of the infant for each frame of the video using a deep learning-based pose prediction model; and (S3) quantifying the smoothness of the voluntary movements of the infant based on the position coordinate data of the joints.

[0071] In step S1, a two-dimensional video of an infant moving spontaneously can be acquired. In step S2, a deep learning-based pose prediction model can be used to obtain position coordinate data for each frame of the two-dimensional video of the infant moving spontaneously. In step S3, the smoothness of the infant's spontaneous movements can be quantified based on the joint position coordinate data.

[0072] Figure 4 is a diagram showing an automatic evaluation system for the smoothness of infants' spontaneous movements.

[0073] Referring to FIG. 4, the automatic evaluation system (100) for the smoothness of spontaneous movements of an infant may include a photographing unit (10), a processor (20), a memory (30), and an output unit (40).

[0074] The camera unit (10) can capture the spontaneous movements of an infant lying straight. The camera unit (10) can include a camera mounted on a tablet, smartphone, etc.

[0075] The processor (20) may correspond to a processor (20) provided in various types of computing devices such as a personal computer (PC), a server device, a mobile device, an embedded device, an Internet of Things (IoT) device, etc. For example, the processor (20) may correspond to a processor (20) such as a central processing unit (CPU), a graphics processing unit (GPU), an application processor (AP), a neural processing unit (NPU), etc., but is not limited thereto.

[0076] The processor (20) may be implemented as an array of a plurality of logic gates, or may be implemented as a combination of a general-purpose microprocessor and a memory (30) storing a program that can be executed on the microprocessor. Furthermore, it will be understood by those skilled in the art to which the present embodiment pertains that the processor (20) may be implemented as other types of hardware.

[0077] The memory (30) can be electrically connected to the processor (20). The memory (30) is hardware that stores various data processed by the processor (20), and for example, can store two-dimensional video data of an infant moving spontaneously.

[0078] The memory (30) may include at least one of volatile memory and nonvolatile memory. Nonvolatile memory includes Read Only Memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable and Programmable ROM (EEPROM), flash memory, Phase-change RAM (PRAM), Magnetic RAM (MRAM), Resistive RAM (RRAM), Ferroelectric RAM (FeRAM), etc. Volatile memory includes Dynamic RAM (DRAM), Static RAM (SRAM), Synchronous DRAM (SDRAM), PRAM, Magnetic RAM (MRAM), Resistive RAM (RRAM), etc. In an embodiment, the memory (30) may be implemented as at least one of a hard disk drive (HDD), a solid state drive (SSD), a compact flash (CF), a secure digital (SD), a micro secure digital (Micro-SD), a mini secure digital (Mini-SD), an extreme digital (xD), or a memory stick.

[0079] The output unit (40) can be electrically connected to the shooting unit (10), the processor (20), and the memory (30), respectively. The output unit (40) can be a visual interface. In this way, the method and system for automatically evaluating the smoothness of spontaneous movements of an infant according to the present invention can provide the effect of evaluating the risk of cerebral palsy or developmental delay in an infant by simply shooting a video of the infant moving spontaneously.

[0080] In particular, the automatic evaluation method and system for the smoothness of voluntary movements of infants of the present invention can provide the effect of quantitatively evaluating the risk of cerebral palsy or developmental delay in infants regardless of the proficiency of the evaluator by quantifying the smoothness of voluntary movements of infants in an automated manner based on machine learning.

[0081] That is, the automatic evaluation method and system for the smoothness of spontaneous movements of infants of the present invention can provide the effect of predicting cerebral palsy or developmental delay in infants at an early stage using a quantitative method based on machine learning.

[0082] Meanwhile, the processing sequence described in the methods, systems, and processes disclosed herein is provided as an example. Therefore, the order of each step may be adjusted within the scope of the present invention, as needed. Furthermore, the devices and systems disclosed herein may include means capable of performing the functions described herein, and may be implemented as independent devices or systems, or may be linked or integrated with other systems, as needed.

[0083] The operations described with reference to FIGS. 1 to 4 may be performed or implemented in a computer-readable storage medium recording a program. In addition, the commands stored in the computer-readable medium may cause a processor in each device or system to perform methods and processes related to the commands.

[0084] By way of non-limiting example, computer-readable media can include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other storage medium that can be used to store desired program code means in the form of instructions or data structures and that can be accessed by a general-purpose or special-purpose computer, or a general-purpose or special-purpose processor. Disk and disc, as used herein, include compact discs (CDs), laser discs, optical discs, digital versatile discs (DVDs), floppy disks, and Blu-ray discs, where disks usually reproduce data magnetically, while discs reproduce data optically with a laser. Combinations of the above should also be included within the scope of computer-readable storage media.

[0085] Although embodiments of the present invention have been described with reference to the attached drawings, those skilled in the art will appreciate that the present invention can be implemented in other specific forms without altering the technical concept or essential features thereof. Therefore, the embodiments described above should be understood to be illustrative in all respects and not restrictive.

Claims

1. A method for automatically evaluating the smoothness of an infant's voluntary movements performed by a processor. The above method, A step of obtaining a two-dimensional video of the infant moving spontaneously; A step of obtaining position coordinate data of the infant's joints for each frame of the video using a deep learning-based pose prediction model; and A step of quantifying the smoothness of the infant's voluntary movements based on the position coordinate data of the joints. A method for automatically assessing the smoothness of voluntary movements of an infant, including:

2. In paragraph 1, The above method, After the step of obtaining the position coordinate data of the above joint, A method for automatically evaluating the smoothness of voluntary movements of an infant, further comprising the step of obtaining angle data of the joint and angular velocity data of the joint based on the position coordinate data of the joint.

3. In paragraph 2, The step of quantifying the smoothness of the above movement is: A method for automatically assessing the smoothness of voluntary movements of an infant, comprising the step of quantifying the smoothness of voluntary movements of the infant based on the angle data of the joint and the angular velocity data of the joint.

4. In paragraph 3, The step of quantifying the smoothness of the above movement is: A method for automatically assessing the smoothness of voluntary movements of an infant, comprising the step of measuring log dimensionless jerk (LDJ) and spectral arc length (SPARC) for the angle data of the joint and the angular velocity data of the joint.

5. In paragraph 1, A method for automatically evaluating the smoothness of voluntary movements of an infant, wherein the position coordinate data of the joint includes position coordinate data of at least one of a left shoulder joint, a right shoulder joint, a left elbow joint, a right elbow joint, a left wrist joint, a right wrist joint, a left hip joint, a right hip joint, a left knee joint, a right knee joint, a left ankle joint, and a right ankle joint.

6. A system that automatically evaluates the smoothness of an infant's voluntary movements. memory; and comprising at least one processor connected to said memory, At least one processor of the above: The above infant receives a two-dimensional video of spontaneous movement; Generate position coordinate data of the infant's joints for each frame of the video using a deep learning-based pose prediction model; and A system for automatically evaluating the smoothness of voluntary movements of an infant, the system configured to quantify the smoothness of voluntary movements of the infant based on the position coordinate data of the joints.

7. In paragraph 6, At least one processor of the above, A system for automatically evaluating the smoothness of voluntary movements of an infant, further configured to generate angle data of the joint and angular velocity data of the joint based on the position coordinate data of the joint.

8. In paragraph 7, At least one processor of the above, A system for automatically evaluating the smoothness of voluntary movements of an infant, the system configured to quantify the smoothness of voluntary movements of the infant based on the angle data of the joint and the angular velocity data of the joint.

9. In paragraph 8, At least one processor of the above, A system for automatically evaluating the smoothness of voluntary movements of an infant, configured to measure log dimensionless jerk (LDJ) and spectral arc length (SPARC) for the angle data of the joint and the angular velocity data of the joint.

10. In paragraph 6, A system for automatically evaluating the smoothness of voluntary movements of an infant, wherein the position coordinate data of the joint includes at least one of the left shoulder joint, the right shoulder joint, the left elbow joint, the right elbow joint, the left wrist joint, the right wrist joint, the left hip joint, the right hip joint, the left knee joint, the right knee joint, the left ankle joint, and the right ankle joint.

11. A computer-readable storage medium recording a program for performing any one of the methods of clauses 1 to 5.

Images