Analysis of gait kinematic parameters of Chinese children based on human pose estimation algorithm

doi:10.21203/rs.3.rs-4305625/v1

Background

The change and development of gait in children is of great importance to clinicians; however, reference data for the quantitative analysis of gait in Chinese children is lacking. This study aimed to describe the gait kinematics of Chinese children aged 3–12 years. In addition, we wanted to explore whether there are differences in the gait kinematics parameters in Chinese children of different age groups and establish a database of gait kinematics parameters of Chinese children with average development.

Methods

A total of 198 children with average development were included in the study and divided into three age groups: group A (3–5 years), group B (6–8 years), and group C (9–12 years). Two smartphone camera were used to record the sagittal and coronal planes of the participants while walking. At least three complete walking videos were recorded for each participant. The video was imported into a computer terminal. Gait evaluation software based on the human posture estimation algorithm was used to analyse the video, and the gait kinematic parameter data of the sagittal and coronal planes were extracted. Analysis of variance was used to determine whether there were differences in the data among the three groups.

Results

The comparison of walking posture among the three age groups showed that there were significant differences in the maximum flexion angle of hip joint, the maximum extension angle, the maximum flexion angle of knee joint and the minimum flexion value of knee joint in sagittal plane (P < 0.05). There were significant differences in knee joint swing amplitude and ankle joint swing amplitude in coronal direction (p < 0.05). the post hoc comparisons revealed that it was found that there were significant differences in the above-mentioned kinematic gait parameters between group A and group C (p < 0.05). In addition, the objective reference data of various kinematic parameters varying with age are established, and the confidence band and prediction band of each age are drawn.

Conclusions

Age is an important factor that affects kinematic gait parameters in children. With increasing age, the kinematic parameters of walking posture exhibited a certain trend of change. Therefore, establishing a standard gait database that varies with age is necessary. Through this method, children of each age have corresponding objective reference data, providing scientific quantitative data and clinical significance for clinicians and parents.

Health sciences/Diseases

Health sciences/Health care

Health sciences/Medical research

Health sciences/Signs and symptoms

children

clinical gait analysis

human pose estimation algorithm

lower limb kinematics

markerless gait analysis

Functional gait is an indispensable component of everyday life. Walking is required almost every day[1, 2]. Severe abnormal gait is often inconvenient to patients, greatly reducing their quality of life, requiring family care, and sometimes bringing psychological problems to the patients[3]. Therefore, gait changes have always been an important aspect of clinical research[4]. Some studies have observed and analysed the stumbling gait of patients with Parkinson’s disease[5] and abnormal gait of patients with cerebral palsy[6] to provide treatment and rehabilitation advice after treatment, which is clinically significant[7, 8].

In the clinical practice of paediatric orthopaedics, some parents have doubts about the gait of their children and believe their walking is abnormal and requires treatment. However, clinicians know that an abnormal gait in children is a nonspecific symptom[9], and there are many reasons for this phenomenon[10, 11]. Most children have a physiologically abnormal gait caused by normal gait changes and habitual movements during growth, a normal phenomenon that does not require excessive treatment. How can we quantitatively judge whether these abnormal gaits are normal or are caused by diseases related to the walking posture in a clinical setting? This is an urgent problem that needs to be addressed and discussed.

To determine whether the gait of a child is pathological, we should first establish a standard gait database of children with average development to provide objective reference data[12, 13]. Three-dimensional gait analysis is the gold standard for gait detection and analysis[14]. Many developed countries have established standard gait databases for three-dimensional gait analysis[15–18]. However, because of ethnic, age, and equipment differences in three-dimensional gait analysis rooms[13, 19], there are some differences in the databases provided by different countries. Therefore, establishing a national standard gait database suitable for our conditions is necessary. At present, there are few standard gait databases for Chinese children, primarily because of the high cost of a three-dimensional gait analysis laboratory[20] and the difficulty for it to be widely used in developing countries, especially in some remote areas. Therefore, low-cost gait analysis software may be more suitable for developing countries.

In recent years, human pose estimation algorithms have developed rapidly and have been used to monitor gait detection and analysis of falls in older adults and Parkinson’s disease stumbling gait[21, 22]. This algorithm can provide a clinically significant quantitative index and accurate quantitative diagnostic data[23–25], which can be applied to an appropriate environment and easily used by personnel. Sometimes, it can be difficult for patients to follow the commands of the three-dimensional gait analysis staff, and they need to wear tools such as marker points and electromyography devices for three-dimensional gait analysis[26]. Patients often have a low degree of fit in the test, the test time is long, and multiple tests are often required. Wearing additional tools affects the joint activity of the lower limbs to a certain extent, which is not consistent with the usual walking posture of children. Conversely, gait evaluation software using the human pose estimation algorithm can more accurately reflect the walking posture. This method does not require marking, and the degree of fit of children is high. Therefore, this method may be more suitable for assessing gait changes in children. However, to our knowledge, no human pose estimation algorithm exists for extracting gaits and establishing a standard database for children with average development. The main reason for this deficiency is that the gait evaluation software for the human pose estimation algorithm is used more in adults and less in children. Gait assessment software is suitable for detecting the gait in children and is urgently required.

In addition, based on data from the standard gait database of average developing children in different countries, statistically significant differences in the joint kinematics at different ages were shown[15, 16]. In contrast, studies in other countries suggest that the differences in joint kinematics of children at different ages are not statistically significant[15]. Most studies show that children aged approximately 7 years have gait kinematics close to the level of adults[11, 27, 28]. Studies have found that the gait of children aged 11–12 differs from that of children aged approximately 7 years[29]. This study also explored whether the differences in joint kinematics of Chinese children at different ages were statistically significant. All children were divided into three age groups for analysis: Group A (3–5 years), group B (6–8 years), and group C (9–12 years).

In previous studies, the two-dimensional gait evaluation of the human pose estimation algorithm focused more on the analysis of joint activity in the sagittal plane[30], these features are insufficient to describe the gait stability of children. Clinically, clinicians can evaluate gait stability by observing the coronal plane of walking and the swinging amplitudes of the shoulder, hip, knee, and ankle joints on both sides, which have been ignored in previous studies. No study has proposed indicators to describe the stability of gait in children. Therefore, this study adds coronal joint swing amplitude changes to describe gait stability for a more comprehensive quantitative diagnosis of abnormal gait.

The purpose of this study was to (1) explore whether there are differences in gait kinematic parameters of average developing Chinese children in different age groups, (2) establish a standard gait database based on average developing children in each age group, (3) provide objective reference data for clinical use.

Participants.

Between March 2022 and June 2023, 209 children from northwest China were invited to participate in the study. The inclusion criteria were as follows: All subjects were required to have no muscular, skeletal, or nervous system-related diseases or a previous history of limb pathology or surgery. The exclusion criteria were as follows: Children with long-term lower limb pain or lower limb movement limitations, children who did not cooperate with the staff during the test, and children who could not complete the gait analysis. Finally, 198 children with average development were included in the study. Patients were divided into three groups: Group A with 97 young children (3.0–5.9 years, 4.49 ± 0.87 years, group B with 59 middle-aged children (6.0–8.9 years, 7.35 ± 0.89 years), and group C with 42 older children (9.0–12.9 years, 10.94 ± 1.21 years). Before the test, all the participants or their parents provided written informed consent. This study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Review Committee of our institution.

Data collection

Smartphones of the same model recorded straight-line walking videos at 30 frames per second. Two smartphones were fixed to two tripods with a height of 75 cm. They recorded the sagittal and coronal planes of the child while walking, with the sagittal plane recorded 250 cm from the walking runway and the coronal plane recorded 150 cm from the starting point of the walking runway. The sagittal plane recordings were synchronised with the coronal plane recordings. In all experiments, participants walked barefoot in shorts and short-sleeved shirts, which could better expose some identification points and facilitate recognition and analysis of the algorithm software. The participants walked along a 10-meter-long track at their usual walking speed. Before the formal test, the evaluator communicated with the participants and asked them to explain the purpose, basic process, and matters needing attention so that they could fully understand and cooperate during the test. The participants were allowed to walk freely on the runway to adapt to the venue before formal recording. An effective trial was defined as an alternating gait sequence with at least three complete continuous strides. At least three valid experiments were required. The points for attention in walking are as follows: (1) do not run or jump while walking; (2) walk in a straight line and with uniform speed, and do not accelerate or decelerate suddenly; and (3) swing your arms naturally during walking. No hand movements with big motions; (4) look ahead while walking. The video data were imported from the mobile phone to a computer terminal through a wireless network, and gait evaluation software was used to process and analyse the data.

Gait evaluation software

Gait evaluation software is mainly based on human pose estimation algorithms. Among these, the heat map method is one of the most commonly used human pose estimation algorithms[31, 32]. However, some shortcomings should be noted[33]: (1) slow operation speed: large heat maps need a lot of calculations to predict key points, reducing the speed of reasoning; (2) insufficient accuracy: there is a quantisation error, and the resolution of the heat map limits the prediction of key points. When the two key points are close, it is easy to be regarded as the same key point; (3) at present, the attitude estimation method based on a heat map performs poorly for multi-person scenes. Given the above shortcomings of the heat map method, the gait evaluation software in this study was mainly based on the single-stage, two-dimensional, multi-person key point and attitude detection method proposed by KAPAO[34]. KAPAO uses a top-down attitude estimation method to detect the boundary box of the human body and 17 bone feature points based on the detection model. The method mainly includes three steps: (1) layering the video image according to the backbone network to detect the character image in the video, (2) decoding the character image in the video into the human boundary box and its key points, and (3) using a matching algorithm to fuse the boundary box and key points and output the results. In addition, considering future application scenarios, it is necessary to extract the posture features of the target character from a multi-person scene. The software also uses a Deepsort tracking algorithm. Deepsort uses a convolutional neural network to extract features from the human body and uses the Kalman filter to predict its trajectory (Fig. 1). The method was tested and proven to have good accuracy and reliability.

The gait evaluation software mainly includes three modules: (1) a video acquisition module, mainly used to import the detected video and is not limited to shooting scenes and equipment; (2) a human pose detection module, using the preset KAPAO algorithm to detect the human body detection frame of each human object in the video and the human body key points in the frame; (3) a human body tracking module, used to match the human body detection box of each human body object in the video frame by frame and generating the continuous gait feature of the human body object according to the human body key points in the matched human body detection frame. The angle of each frame was extracted, and the exponential smoothing method was used to smoothen the original kinematic data. The calculation of these angles is shown in Fig. 2.

Data analysis

Gait evaluation software was used to extract and calculate the joint angle of each subject, and three complete gait cycles were intercepted based on the walking video of each subject. Traditionally, the process of one side of the foot landing on that side of the heel while walking is called the walking cycle[35]. In the sagittal plane, by obtaining the flexion and extension angles of the hip and knee joints at each frame of the gait cycle, we obtained the maximum flexion and extension angles and the range of motion of the hip and knee joints in the gait cycle as our data index. In the coronal plane, we can calculate the swing amplitude of each joint in the gait cycle by obtaining the angle between the horizontal line and the connection line of the shoulder, hip, knee, and ankle joints at each frame of the gait cycle. The relevant data indices of the three gait cycles were calculated, and the average value was calculated as the gait data of the participant.

Statistical analysis

All data were analysed using SPSS (version 26.0; IBM, Armonk, NY, USA). The three age groups were group A: 3–5 years, group B: 6–8 years, and group C: 9–12 years. Descriptive statistics (mean and standard deviation) were used to describe the demographic characteristics of the participants, and median and range or mean and standard deviation were used to describe the results. The Shapiro–Wilk test was performed to determine the normal distribution of each variable. One-way analysis of variance or the Kruskal–Wallis test was used to analyse differences in gait kinematics among the three groups. Fisher’s least significant difference was used to compare and analyse the three data groups.

In addition, grouping was performed according to age groups: 3, 4, 5, 6, 7, 8, 9, 10, 11, and 12 years. Descriptive statistics (means and standard deviations) were used to describe the demographic characteristics of the participants. Origin2022 software was used for linear fitting, and the confidence and prediction bands of the gait kinematics data changing with age were drawn to provide objective reference data for clinicians.

Demographic characteristics

Between March 2022 and June 2023, 209 children participated in the study, of whom 207 met the inclusion criteria. After excluding nine children who met the exclusion criteria, 198 children between 3 and 12 years of age were included. The average age, weight, and height were 6.7 years (±2.7), 25.8 kg (±11.1), and 122.07 cm (±17.24), respectively. Table 1 shows the characteristics of the participants after grouping by age. Age and weight significantly differed between the groups (p < 0.001). In terms of sex, there was no significant difference between the groups. Table 2 shows the characteristics of the participants by age group.

	Age(years)	sex,n(%)	Height (cm)	Weight (kg)
Group A(n=97)	4.5±0.87		108.10±7.52	18.41±4.28
Group B(n=59)	7.35±0.89		127.41±7.12	26.99±6.38
Group C(n=42)	10.94±1.21		146.86±9.53	41.27±10.83
P value	<0.0001	>0.05	<0.0001	<0.0001

Table1. Demographic characteristics of the participants (N = 198). Group A (3-5 years old), Group B (6-8 years old), Group C (9-12 years old). Differences in age, height, weight, but not sex, between the three groups were analyzed using One-way analysis of variance.

	Age(years)	sex,n(%)	Height(cm)	Weight (kg)
3(n=38)	3.48±0.28		100.98±4.19	15.80±1.89
4(n=28)	4.51±0.31		109.03±5.52	18.08±2.85
5(n=36)	5.42±0.23		113.90±5.68	21.06±5.20
6(n=21)	6.36±0.24		120.75±3.73	22.17±3.53
7(n=19)	7.39±0.33		127.78±4.89	26.87±4.38
8(n=19)	8.39±0.28		134.42±4.52	32.43±6.29
9(n=12)	9.49±0.24		137.29±3.01	32.88±5.30
10(n=10)	10.49±0.34		144.60±7.83	34.81±8.42
11(n=10)	11.51±0.33		148.94±4.50	45.81±4.08
12(n=10)	12.54±0.32		158.51±6.15	53.26±9.47
P value	<0.0001	>0.05	<0.0001	<0.0001

Table2. Demographic characteristics of the participants (N = 198). Divided into 10 groups according to age. Differences in age, height, weight, but not sex, between the ten groups were analyzed using the One-way analysis of variance.

Kinematic parameters of sagittal plane

Table 3 shows a comparative analysis of the maximum flexion and extension angles and the range of motion of the hip and knee joints among the three age groups in the gait cycle. The maximum extension angle of the hip joint increased with the age group increase, and the analysis between the groups was statistically significant (p < 0.001). The post hoc comparisons revealed a statistically significant difference in the maximum hip extension angle between groups A and C. The maximum flexion angle of the hip joint decreased with an increase in age group, and the difference between the groups was statistically significant (p < 0.001). The post hoc comparisons revealed a statistically significant difference in the maximum flexion angle of the hip joint between groups A and B and between groups A and C. No difference in the intragroup analysis of hip joint range of motion was observed. The minimum flexion of the knee joint decreased with increasing age group, and the analysis between the groups was statistically significant (p < 0.001). Post-hoc comparisons revealed statistical significance among the three groups. The maximum flexion of the knee joint decreased with age, and statistical significance existed in the analysis between groups. Post-hoc comparisons revealed a statistically significant difference in the maximum flexion of the knee joint between groups A and C. The range of the knee joint increased with age, and the difference between the groups was statistically significant. Post hoc comparisons revealed a statistically significant difference in the knee joint range of motion between groups A and C.

variables	Group A(n=97)	Group B(n=59)	Group C(n=42)	P value
Maximum hip extension	-11.12(4.33)	-12.06(3.87)	-14.40(4.89)	A and B:>0.05
				A and C:<0.001
				B and C:<0.05
Maximum hip flexion	32.55(3.49)	30.45(3.29)	29.53(2.54)	A and B:<0.001
				A andC:<0.001
				B and C:>0.05
Hip ROM	43.67(6.28)	42.45(5.38)	43.93(5.80)	A and B:>0.05
				A and C:>0.05
				B and C:>0.05
Minimum knee flexion	8.36(3.54)	6.44(3.16)	4.54(2.89)	A and B:<0.001
				A and C:<0.001
				B and C:<0.001
Maximum knee flexion	60.48(3.61)	59.91(3.37)	58.64(4.63)	A and B:>0.05
				A and C:<0.001
				B and C:>0.05
Knee ROM	52.12(4.71)	53.47(4.46)	54.23(4.04)	A and B:>0.05
				A and C:<0.05
				B and C:>0.05

Table3. The minimum and maximum values for the gait kinematics of the hip and knee angle for the three age groups (N = 198). Group A (3–5 years), Group B (6–8 years), Group C (9–12 years). Differences between the three groups were analyzed using the Kruskal–Wallis test or one-way analysis of variance. Variables with significant differences were subsequently compared using multiple comparison analyses with Bonferroni correction. Data are presented as means (standard deviation) or median values (range). P values < 0.05 were considered statistically significant.

Kinematic parameters of coronal plane

Table 4 presents the joint swing amplitudes of the shoulder, hip, knee, and ankle joints during the gait cycle for the three age groups. Among them, the swing amplitude of the shoulder joint decreased with age, and there was no statistical significance in intergroup analysis. The hip joint amplitude increased with age, and intergroup analysis had no statistical significance. The knee joint amplitude decreased with increasing age, and the intergroup analysis was statistically significant. The post hoc comparisons revealed a statistical significance between group A and group C, group B, and group C. The ankle joint amplitude increased with age, and the post hoc comparisons revealed a statistical significance between group A and group B, group A, and group C.

variables	Group A(n=97)	Group B(n=59)	Group C(n=42)	P value
Shoulder swinging angle	4.30(1.21)	4.09(1.14)	3.86(0.79)	A and B:>0.05
				A and C:<0.05
				B and C:>0.05
Hip swinging angle	10.78(3.01)	11.24(2.86)	12.14(3.59)	A and B:>0.05
				A and C:<0.05
				B and C:>0.05
Knee swinging angle	28.34(5.40)	27.33(6.74)	22.84(5.23)	A and B:>0.05
				A and C:<0.001
				B and C:<0.001
Ankle swinging angle	87.01(13.75)	95.42(17.26)	95.30(14.58)	A and B:<0.001
				A and C:<0.005
				B and C:>0.05

Table4. the joint swing amplitudes of shoulder, hip, knee and ankle joints in three age groups during the gait cycle. Group A (3–5 years), Group B (6–8 years), Group C (9–12 years). Differences between the three groups were analyzed using the Kruskal–Wallis test or one-way analysis of variance. Variables with significant differences were subsequently compared using multiple comparison analyses with Bonferroni correction. Data are presented as means (standard deviation) or median values (range). P values < 0.05 were considered statistically significant.

Objective reference data for age groups

Figure 3 shows the changing trend of sagittal and coronal kinematic data with age groups, in which a linear regression curve with age was drawn, and the confidence and prediction bands of the kinematic data were added. This can provide objective reference data for children seeking medical treatment.

The quantitative evaluation of gait index in children has always been the focus of clinical research, especially in the fields of orthopaedics and neurosurgery[36–38], and has been developing in the direction of lightweight and low cost[39–41]. This study aimed to use a new low-cost gait evaluation method to test whether there are differences in parameters among children of different age groups to establish an age-related gait kinematics database to provide objective quantitative reference data for clinical use.

The results showed differences in gait kinematics data among the three age groups. In the sagittal plane, the maximum flexion of the hip and knee joints decreases with age, similar to the literature reported in other countries[12, 42, 43]. In terms of the maximum flexion of the hip joint, there were significant differences between children aged 3–5 years and 6–8 years, 3–5 years and 9–12 years. The maximum flexion of the knee joint showed a significant difference between children aged 3–5 years and those aged 9–12 years. The minimum flexion of the knee joint revealed significant differences among the three groups. The maximum extension angle of the hip joint increased with age, and there was a significant difference between 3–5-year-old and 9–12-year-old children. These differences could be attributed to muscle development with age, which impacts joint flexion and extension activities. Walking has a high degree of control from the nervous system to the musculoskeletal system[44–46]. Some studies have shown that this occurs because of the flexor bias pattern during infancy[47]. In infancy, joint flexion is dominant; therefore, younger children have a larger joint flexion angle when walking than older children. With increasing age, the extensor muscle gradually becomes consistent with the adult level; therefore, the maximum extension angle of the hip joint increases gradually, while the minimum flexion value of the knee joint decreases gradually. From the results, it can be seen that there are significant differences in the gait kinematics data between 3–5-year-old and 9–12-year-old children, whereas there are differences between 6–8-year-old and 9–12-year-old gait kinematics data, although not significant. This is similar to the current international literature report[48], stating that the kinematic parameters of children are stable at approximately 7 years of age [3].

Sutherland was one of the earliest researchers to study gait in children. The research showed[49] that gait is stable at the age of 3–4, a conclusion drawn by comparing the gait parameters of 2 to 7-year-old children. In this study, there were more significant differences between the 3–5-year-old and 9–12-year-old age groups. However, the two age groups were not compared. Gait in children is constantly growing and changing, and mature gait is controlled not only by the central nervous and musculoskeletal systems but also by the accumulation of walking experience[45, 48, 50]. This can explain why gait changes with age; it may not change significantly at 3–4 years, but there may be significant differences with time. Therefore, it is necessary to establish a standard age-related gait database to provide objective reference data for different age groups.

A new kinematic parameter for the coronal plane was also proposed in this study. Through clinical observation, clinicians could qualitatively judge whether gait was stable based on the swing amplitude of the two shoulders or two hips. Therefore, this study quantitatively evaluated the gait stability of children by calculating the joint swing amplitude of the two shoulders, two hips, two knees, and two ankles during the gait cycle. The results showed a relationship between the swing amplitudes of the four joints and age. In terms of knee joint swing, there was a statistically significant difference between children aged 3–5 years and 6–8 years, as well as the 6–8 years and 9–12 years groups. This is primarily because of the large flexion of the hip joint of young children during walking, which increases the longitudinal coordinates of the knee joint in the coronal plane; thus, the swing will be large. In terms of ankle swing, there were significant differences between children aged 3–5 years and 6–8 years, 3–5 years and 9–12 years. This is because young children have larger stride width[46]. Consequently, the abscissa increases when calculating the swing amplitude; thus, the swing amplitude decreases.

A standard database of gait in children has been created in many developed countries and reported in international literature[12, 17, 18, 51, 52]. Through a comparative analysis of the database reports of these countries, we found some differences in standard gait databases in different countries[15, 16]. This may be caused by cultural differences, living environment, sample size, and laboratory equipment. Therefore, it is necessary to establish a national standard gait database to provide objective reference data for children. Children in every age group from 3 to 12 years were recruited. The gait data of each child with average development were extracted, a data diagram of the variation in gait kinematic parameters with age was drawn, and the confidence and prediction bands were drawn to provide objective reference data for other children with abnormal gait. This finding has clinical and practical significance.

Through the objective reference data provided by these kinematic parameters, clinicians can quickly understand changes in gait. More importantly, it can also help parents understand the changing trend in the gait of their children rather than worry about abnormal gait. Through these data, healthcare providers or parents can also screen and identify diseases that affect the activities of the lower limbs, which has clinical significance. The data can also guide postoperative gait rehabilitation and guide children towards a normal gait.

Limitation

In this study, low-cost gait evaluation software can extract the relevant gait kinematics data; however, the data are still small and lack spatiotemporal and ankle sagittal motion parameters. In addition, the recruits in this study were mainly from northwest China, the sample size was small, and children from different areas may have different gait development patterns. Future work should conduct recruitment experiments in various regions of the country and increase the sample size to establish a more accurate standard gait database of Chinese children.

Age is an important factor that affects kinematic gait parameters in children. With increasing age, the kinematic parameters of the walking posture exhibited a certain trend of change. Therefore, it was necessary to establish a reference database for gait kinematics according to age. Through this method, children of each age have corresponding objective reference data providing scientific quantitative data for clinicians and parents, which have clinical significance.

Acknowledgement

Thanks to all those who have contributed to this work. All authors reviewed the results and approved the final version of the manuscript.

Data availability statement

All the relevant data are presented in the manuscript. All data are available from the authors upon request. Author Jincong Lin can be contacted at [email protected].

Ethics approval This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of Air Force Medical University (Date: March 3, 2023,

Decision No.: KY20233214-1).

Consent for publication All authors agreed with the content and gave explicit consent to submit.

Competing interests The authors declare no competing interests.

Rosengren, K.S., et al., Differences in gait complexity and variability between children with and without developmental coordination disorder. Gait Posture, 2009. 29(2): p. 225–9.
Deconinck, F.J., et al., Differences in gait between children with and without developmental coordination disorder. Motor Control, 2006. 10(2): p. 125–42.
Klupp, S., A. Grob and W. Mohring, Gait Variability Relates to Prosocial, Emotional and Risk-Taking Behavior in Typically Developing Children. Percept Mot Skills, 2023. 130(1): p. 191–207.
Middleton, A., S.L. Fritz and M. Lusardi, Walking speed: the functional vital sign. J Aging Phys Act, 2015. 23(2): p. 314–22.
Ouyang, C.S., et al., Evaluating Therapeutic Effects of ADHD Medication Objectively by Movement Quantification with a Video-Based Skeleton Analysis. Int J Environ Res Public Health, 2021. 18(17).
Gomez-Perez, C., et al., Gait parameters in children with bilateral spastic cerebral palsy: a systematic review of randomized controlled trials. Dev Med Child Neurol, 2019. 61(7): p. 770–782.
Kennedy, R.A., et al., Walking and weakness in children: a narrative review of gait and functional ambulation in paediatric neuromuscular disease. J Foot Ankle Res, 2020. 13(1): p. 10.
Ma, N., et al., Three-Dimensional Gait Analysis in Children Undergoing Gastrocsoleus Lengthening for Equinus Secondary to Cerebral Palsy. Medicina (Kaunas), 2021. 57(2).
Payares-Lizano, M., The Limping Child. Pediatr Clin North Am, 2020. 67(1): p. 119–138.
Whittle, M.W., Generation and attenuation of transient impulsive forces beneath the foot: a review. Gait Posture, 1999. 10(3): p. 264–75.
Lythgo, N., C. Wilson and M. Galea, Basic gait and symmetry measures for primary school-aged children and young adults. II: walking at slow, free and fast speed. Gait Posture, 2011. 33(1): p. 29–35.
Ito, T., et al., Three-dimensional gait analysis of lower extremity gait parameters in Japanese children aged 6 to 12 years. Sci Rep, 2022. 12(1): p. 7822.
Dusing, S.C. and D.E. Thorpe, A normative sample of temporal and spatial gait parameters in children using the GAITRite electronic walkway. Gait Posture, 2007. 25(1): p. 135–9.
Pantzar-Castilla, E., et al., Knee joint sagittal plane movement in cerebral palsy: a comparative study of 2-dimensional markerless video and 3-dimensional gait analysis. Acta Orthop, 2018. 89(6): p. 656–661.
Smith, Y., Q. Louw and Y. Brink, The three-dimensional kinematics and spatiotemporal parameters of gait in 6–10 year old typically developed children in the Cape Metropole of South Africa - a pilot study. BMC Pediatr, 2016. 16(1): p. 200.
Ito, T., et al., Three-dimensional gait analysis of lower extremity gait parameters in Japanese children aged 6 to 12 years. Sci Rep, 2022. 12(1): p. 7822.
Chester, V.L., M. Tingley and E.N. Biden, A comparison of kinetic gait parameters for 3–13 year olds. Clin Biomech (Bristol, Avon), 2006. 21(7): p. 726–32.
Nikolajsen, H., et al., Gait pattern in 9-11-year-old children with generalized joint hypermobility compared with controls; a cross-sectional study. BMC Musculoskelet Disord, 2013. 14: p. 341.
Pinzone, O., et al., The comparison of normative reference data from different gait analysis services. Gait Posture, 2014. 40(2): p. 286–90.
Saner, R.J., E.P. Washabaugh and C. Krishnan, Reliable sagittal plane kinematic gait assessments are feasible using low-cost webcam technology. Gait Posture, 2017. 56: p. 19–23.
Sato, K., et al., Quantifying normal and parkinsonian gait features from home movies: Practical application of a deep learning-based 2D pose estimator. PLoS One, 2019. 14(11): p. e0223549.
Mehdizadeh, S., et al., Concurrent validity of human pose tracking in video for measuring gait parameters in older adults: a preliminary analysis with multiple trackers, viewing angles, and walking directions. J Neuroeng Rehabil, 2021. 18(1): p. 139.
Liang, S., et al., The reliability and validity of gait analysis system using 3D markerless pose estimation algorithms. Front Bioeng Biotechnol, 2022. 10: p. 857975.
van Bloemendaal, M., et al., Concurrent validity and reliability of a low-cost gait analysis system for assessment of spatiotemporal gait parameters. J Rehabil Med, 2019. 51(6): p. 456–463.
Lee, J.B., et al., Identifying symmetry in running gait using a single inertial sensor. J Sci Med Sport, 2010. 13(5): p. 559–63.
Saner, R.J., E.P. Washabaugh and C. Krishnan, Reliable sagittal plane kinematic gait assessments are feasible using low-cost webcam technology. Gait Posture, 2017. 56: p. 19–23.
Vaughan, C.L., N.G. Langerak and M.J. O'Malley, Neuromaturation of human locomotion revealed by non-dimensional scaling. Exp Brain Res, 2003. 153(1): p. 123–7.
Berger, W., E. Altenmueller and V. Dietz, Normal and impaired development of children's gait. Hum Neurobiol, 1984. 3(3): p. 163–70.
Moreno-Hernandez, A., et al., Temporal and spatial gait parameters analysis in non-pathological Mexican children. Gait Posture, 2010. 32(1): p. 78–81.
Pantzar-Castilla, E., et al., Knee joint sagittal plane movement in cerebral palsy: a comparative study of 2-dimensional markerless video and 3-dimensional gait analysis. Acta Orthop, 2018. 89(6): p. 656–661.
G., B., K. N. and L. L. The Center of Attention: Center-Keypoint Grouping via Attention for Multi-Person Pose Estimation. in 2021 IEEE/CVF International Conference on Computer Vision (ICCV). 2021.
B., C., et al. HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation. in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). 2020.
Luo, Z., et al., Rethinking the Heatmap Regression for Bottom-up Human Pose Estimation. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020: p. 13259–13268.
McNally, W., et al. Rethinking Keypoint Representations: Modeling Keypoints and Poses as Objects for Multi-person Human Pose Estimation. 2022. Cham: Springer Nature Switzerland.
Stenum, J., C. Rossi and R.T. Roemmich, Two-dimensional video-based analysis of human gait using pose estimation. PLoS Comput Biol, 2021. 17(4): p. e1008935.
Kainz, H., et al., Joint kinematic calculation based on clinical direct kinematic versus inverse kinematic gait models. J Biomech, 2016. 49(9): p. 1658–1669.
Ounpuu, S., Gait analysis is a viable tool for the assessment of transverse plane motion in children with cerebral palsy. Dev Med Child Neurol, 2013. 55(10): p. 878–9.
Willerslev-Olsen, M., et al., Gait training facilitates central drive to ankle dorsiflexors in children with cerebral palsy. Brain, 2015. 138(Pt 3): p. 589–603.
Castelli, A., et al., A 2D Markerless Gait Analysis Methodology: Validation on Healthy Subjects. Comput Math Methods Med, 2015. 2015: p. 186780.
Werner, C., et al., Validity and reliability of the Apple Health app on iPhone for measuring gait parameters in children, adults, and seniors. Sci Rep, 2023. 13(1): p. 5350.
Spanos, S., et al., Reliability and applicability of a low-cost, camera-based gait evaluation method for clinical use. Expert Rev Med Devices, 2023. 20(1): p. 63–70.
Chester, V.L., M. Tingley and E.N. Biden, A comparison of kinetic gait parameters for 3–13 year olds. Clin Biomech (Bristol, Avon), 2006. 21(7): p. 726–32.
Lee, S.Y., et al., Age-related changes in physical examination and gait parameters in normally developing children and adolescents. J Pediatr Orthop B, 2013. 22(2): p. 153–7.
Xiong, Q., et al., Age-related differences in gait symmetry obtained from kinematic synergies and muscle synergies of lower limbs during childhood. Biomed Eng Online, 2022. 21(1): p. 61.
Rygelova, M., et al., Comparison of spatiotemporal gait parameters and their variability in typically developing children aged 2, 3, and 6 years. PLoS One, 2023. 18(5): p. e0285558.
Miyagishima, S., et al., Developmental changes in straight gait in childhood. PLoS One, 2023. 18(2): p. e0281037.
Forssberg, H., Ontogeny of human locomotor control. I. Infant stepping, supported locomotion and transition to independent locomotion. Exp Brain Res, 1985. 57(3): p. 480–93.
Abbruzzese, L.D., et al., Effects of manual task complexity on gait parameters in school-aged children and adults. Gait Posture, 2014. 40(4): p. 658–63.
Sutherland, D.H., et al., The development of mature gait. J Bone Joint Surg Am, 1980. 62(3): p. 336–53.
Breniere, Y. and B. Bril, Development of postural control of gravity forces in children during the first 5 years of walking. Exp Brain Res, 1998. 121(3): p. 255–62.
Smith, Y., Q. Louw and Y. Brink, The three-dimensional kinematics and spatiotemporal parameters of gait in 6–10 year old typically developed children in the Cape Metropole of South Africa - a pilot study. BMC Pediatr, 2016. 16(1): p. 200.
Guffey, K., et al., Gait parameters associated with balance in healthy 2- to 4-year-old children. Gait Posture, 2016. 43: p. 165–9.

No competing interests reported.

Analysis of gait kinematic parameters of Chinese children based on human pose estimation algorithm

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Introduction

Materials and Methods

Data collection

Gait evaluation software

Data analysis

Statistical analysis

Results

Discussion

Limitation

Conclusion

Declarations

References

Additional Declarations

Status:

Version 1