Analysis of the application of computerized optokinetic nystagmus analyzer in vision examination

Abstract

Objective

To investigate the practicability, accuracy, and consistency of the computerized optokinetic nystagmus (OKN) analyzer for visual acuity (VA) measurement.

Method

Cross-sectional consistency evaluation study. A total of 219 subjects (13.48 ± 12.89 years, 115 males, 104 females) underwent bilateral VA testing using a computerized OKN analyzer and a traditional OKN drum. Additionally, 104 subjects (12.58 ± 12.05 years, 54 males, 50 females) were tested bilaterally with the computerized OKN analyzer and a standard logarithmic E chart (random order), including 80 myopic eyes (spherical equivalent [SE] ≤ -0.5 D) and 63 hyperopic eyes (SE ≥ + 0.5 D); VA was recorded in logMAR. Concordance rates between the OKN analyzer and OKN drum were calculated. Intraclass correlation coefficients (ICC) and Bland-Altman plots were used to assess the consistency between the OKN analyzer and a standard logarithmic E chart. Subjects completing the lightbox test were stratified by age (3–4 years, 5 years, 6 years, 7–18 years, 18–58 years), refractive error, and VA range to compare inter-method correlations.

Results

Compared with the OKN drum, the computerized OKN analyzer achieved a concordance rate of 74.00% (right eye) and 77.08% (left eye), indicating good agreement. For VA measurements, mean VA (logMAR) of the right eye was 0.24 ± 0.29 (OKN analyzer) vs. 0.24 ± 0.28 (lightbox); left eye was 0.25 ± 0.28 vs. 0.24 ± 0.29. ICC values were 0.92 (right eye) and 0.88 (left eye, both P < 0.001), confirming significant consistency. Bland-Altman analysis showed 95% limits of agreement of -0.228 to 0.233 (right eye) and − 0.266 to 0.293 (left eye). Inter-method consistency improved with age (highest in ≥ 18 years) and was stronger in myopic vs. hyperopic eyes. Notably, correlation weakened as VA improved, with the highest concordance in the 0.1–0.5 decimal VA.

Conclusion

The computerized OKN analyzer exhibits favorable reliability and practicability for VA testing in patients aged ≥ 6 years across different refractive errors, with accuracy that appears sufficient for screening and general clinical estimation of VA. However, in very young children (3–4 years) and in eyes with normal VA, agreement with chart‑based VA is limited, and further validation is required.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12886-026-05165-1.

Introduction

The 0–6 years age window is critical for pediatric visual development: abnormal visual experiences during this period can disrupt visual system maturation and impair visual function [1–3]. Thus, early refractive screening and comprehensive vision assessment in infants and young children are essential for timely intervention in developmental visual disorders and optimal clinical outcomes.

Current pediatric visual function assessment includes qualitative (behavioral observation, no quantitative VA data) and quantitative methods. Quantitative VA tests, based on the Preferential Gaze principle (e.g., forced preferential looking, enhanced preferential looking), rely on psychophysical paradigms but require repeated trials, prolonging examination and limiting clinical utility [4]. Visual Evoked Potential, an electrophysiological method, objectively evaluates retinocortical pathway function but cannot reflect subjective visual perception or assess high-level visual processing centers [5, 6].

Young children’s immature cognitive, motor, and language abilities hinder objective, accurate VA measurement. While quantitative methods provide numerical data, their correlation with lightbox VA chart is weak, partly due to the lack of a standardized framework for infants and young children vision assessment [7]. For children at risk of amblyopia or visual impairment, reliable quantitative VA tools are urgently needed—especially for those unable to complete traditional VA tests.

The computerized OKN analyzer was initially developed for VA testing in mouse models and later adapted for humans [8]. It enables quantitative VA assessment and is particularly suitable for children and patients with language disorders. Previous studies have validated a computer-based OKN device with infrared eye-tracking for subjective VA correlation, but small sample sizes limited its clinical applicability [9, 10]. Further validation of its accuracy against standard VA charts is required.

This study compared the computerized OKN analyzer with the traditional OKN drum and a standard logarithmic E chart. Objectives included: (1) evaluating the performance of the computerized OKN analyzer vs. traditional tools; (2) establishing correlations between OKN responses and standardized VA measurements; (3) exploring the preliminary clinical feasibility of the computerized OKN analyzer for VA assessment. The computerized OKN analyzer is a potential objective VA assessment tool for children and patients with language disorders, and this study aims to validate its performance in a cooperative pediatric and adult population as a first step.

Methods

Study population

A total of 219 subjects (438 eyes) were enrolled from Beijing Tongren Eye Center between October 31 and December 31, 2023. Subgroup analysis: Refractive data were converted to SE, subjects were categorized as myopic (SE ≤ -0.5 D) or hyperopic (SE ≥ + 0.5 D). Agreement analyses were conducted by age: 3–4 years (n = 14), 5 years (n = 13), 6 years (n = 17), 7–18 years (n = 37), and > 18 years (n = 23), as well as by decimal visual acuity range: 0.1–0.5, 0.5–1.0, and 1.0–1.5.

Inclusion criteria: Ability to complete standard logarithmic E chart testing during routine outpatient visits (for children ≤ 6 years, cooperation was defined as able to fixate on the target for ≥ 10 s and respond to optotype identification via verbal/gestural means); (2) No concurrent ocular disease (including conjunctivitis, keratitis, cataract, glaucoma, retinopathy); (3) No systemic neurological diseases (e.g., cerebral palsy, autism) that affect eye movement or cooperation; (4) Complete refractive and VA data for both eyes.

Exclusion criteria: (1) Congenital/secondary nystagmus (any type); (2) Recent ocular infection/active lesions (within 2 weeks before testing); (3) Head shaking during testing (≥ 3 times per test cycle) that affects eye movement recording; (4) Inability to complete the test due to severe anxiety/crying (unable to cooperate after 3 attempts with distraction reduction); (5) Refractive data missing or unreadable; (6) Vertigo or discomfort during OKN testing that leads to test interruption.

Instruments and procedures.

Optotype vision screening

A 21-inch display (1.5 m from the subject) presented scrolling black-and-white striped gratings (resolution: 2560 × 1440; contrast: 100%). Stimulus parameters: stripe width 0.5–8.7 mm, inter-grating distance 12–26 mm, rolling speed 5°/s (Fig. 1). The optokinetic “wheel effect” was mitigated by optimizing inter-stripe spacing. Each test cycle included 3.5 s of left-to-right scrolling and 3.5 s of right-to-left scrolling. An infrared camera recorded nystagmus in real time, transmitting data to a computer for waveform generation (Fig. 2).

Fig. 1.

(A) Computerized OKN analyzer; (B) Scrolling screen bars

Fig. 2.

This shows the nystagmus waveform captured by computer, where A is the left eye and B is the right eye. The vertical axis shows the eye movement, and the horizontal axis shows the time duration

Stimulus sequence for computerized OKN testing

Visual stimuli were presented in a binary sequence. The initial stimulus corresponded to 0.5 decimal VA. If ≥ 2 typical nystagmus waveforms were recorded, the next stimulus was adjusted to 0.3 decimal VA; if not, the next stimulus was 0.7 decimal VA. This iterative process continued until the smallest VA eliciting nystagmus was identified as the OKN analyzer-measured VA. A single 7 s presentation sequence consisting of 3.5 s of left to right grating motion followed immediately by 3.5 s of right to left motion at a constant speed of 5°/s. The smallest spatial frequency at which a positive OKN response was recorded, converted to a VA value using the device’s calibration.

Computerized OKN analyzer test

Subjects rested their chin on a chin rest and forehead against the device. One eye was occluded; the subject viewed screen animations to ensure ocular relaxation. The displayed stripe widths and inter-grating distances were physically validated by measuring the screen display with a digital caliper, with a measurement error of < 1% (within clinical acceptable limits). Testing was performed in a dimly lit examination room (ambient luminance < 10 cd/m²) to eliminate glare and ensure the OKN stimulus was the primary visual target. The infrared system recorded nystagmus, and the computer filtered waveforms to determine valid responses, adjusting stimuli until the minimum responsive VA was identified. A “positive response” was defined as ≥ 2 distinct nystagmus waveforms per cycle; fewer waveforms indicated failure to perceive the target.

OKN drum test

The drum was rotated slowly in front of the subject to observe OKN presence. Initial stimulus corresponds to 0.2 decimal VA. If ≥ 2 nystagmus if ≥ 2 typical OKN beats were observed, VA is adjusted to 0.1 decimal VA. If < 2 waveforms, VA is adjusted to 0.4 decimal. Minimum of three reversals required to determine threshold. Each VA level is confirmed by two consecutive identifications before recording. The test is concluded when three consecutive VA levels show consistent nystagmus responses, indicating the minimum threshold. Concordance was defined as qualitative agreement in the presence/absence of a detectable OKN response for both the computerized OKN analyzer and the traditional OKN drum.

Standard logarithmic E chart test

A standard logarithmic E chart (Chinese national standard: GB 11533 − 2011) was used, with a fixed testing distance of 2.5 m (calibrated with a laser distance meter before each test session). The lightbox was calibrated monthly to ensure uniform luminance (80–320 lx) and clear optotype printing, in accordance with international VA testing standards (ISO 8598:2017). The right eye was tested first, followed by the left, with strict monocular occlusion (opaque eye patch with no light leakage). Testing proceeded downward from the top line; if > 50% of optotypes in a line were identified correctly (verbal response for adults/older children, gestural pointing for children ≤ 6 years), the next line was tested. The last successfully identified line with > 50% correct responses were recorded as the uncorrected VA in logMAR. The standard logarithmic E chart was calibrated monthly to ensure compliance with illumination and optotype size standards. For children aged 3–4 years, the same standardized logarithmic E chart (GB 11533 − 2011) was used. This chart is widely adopted in pediatric ophthalmology clinics in China and has age-appropriate optotypes that can be identified via verbal naming or directional pointing, provided adequate cooperation.

All participants were tested with their habitual refractive correction (glasses/contact lenses) for both the computerized OKN analyzer and the lightbox VA chart. For participants without habitual correction (e.g., mild refractive error), testing was performed uncorrected, and refractive error was recorded as spherical equivalent (SE) for subgroup analysis. Accommodative effort was not artificially suppressed during testing, in accordance with clinical routine VA assessment protocols.

Statistical analysis

Intraclass correlation coefficients (ICC) were calculated using SPSS 21 to quantify the association and consistency between measurements obtained by the two methods. A two-way mixed-effects model for absolute agreement (single measures) was applied, and ICC values were interpreted as follows: <0.5 = poor, 0.5–0.75 = moderate, 0.75–0.9 = good, and > 0.9 = excellent. For all ICC estimates, 95% confidence intervals (95% CI) were calculated to quantify the precision of the agreement estimates. Bland–Altman plots were constructed to assess absolute agreement between methods by plotting the mean difference against the mean of the two measurements; 95% limits of agreement (LOA) were calculated as mean difference ± 1.96 SD. Paired t‑tests were used to detect systematic differences between methods, with P > 0.05 indicating no statistically significant difference. VA ranges were defined using decimal VA (Snellen decimal), while all reported mean values and statistical analyses used logMAR units.

Results

Consistency between computerized OKN analyzer and OKN drum

Of 219 subjects (age distribution: Table 1), 167 (right eye) and 171 (left eye) showed consistent results between the two methods. Concordance rates were 74.00% (right eye) and 77.08% (left eye), indicating good agreement.

Table 1.

Age distribution of subjects

Age (years)	Case	Rate
3~4	23	10.5%
5	29	13.2%
6	33	15.1%
7~18	81	37.0%
> 18	53	24.2%

Consistency between computerized OKN analyzer and standardized logarithmic e chart

A total of 104 subjects were included (age distribution: Table 2). Mean VA (logMAR) was comparable between methods: Right eye: 0.24 ± 0.29 (OKN analyzer) vs. 0.24 ± 0.28 (lightbox); Left eye: 0.25 ± 0.28 (OKN analyzer) vs. 0.24 ± 0.29 (lightbox).

Table 2.

Age distribution of subjects

Age (years)	Case	Rate
3~4	14	13.5%
5	13	12.5%
6	17	16.3%
7~18	37	35.6%
> 18	23	22.1%

ICC values were 0.92 (right eye) and 0.88 (left eye, both P<0.001), confirming good consistency. Bland-Altman analysis showed mean differences of 0.003±0.12 (right eye) and 0.01±0.14 (left eye), with 95% limits of agreement of -0.228 to 0.233 (right eye) and -0.266 to 0.293 (left eye) (Fig 3). Paired t-tests showed no significant differences (all P>0.05).

Fig. 3.

(A) Bland-Altman analysis of right eyes (R); (B) Bland-Altman analysis of left eyes (L); (C) Scatter plots and related lines of right eyes (R); (D) Scatter plots and related lines of left eyes (L)

Subgroup analyses (age, refractive error, VA range)

Subgroup analyses were performed within the 104-subject cohort that completed both the computerized OKN analyzer and standard logarithmic E chart testing.

Age groups

Inter-method consistency improved with age (Table 3). The ≥ 18 years group showed the highest ICC, while the 3–4 years group had the lowest. No significant differences were observed via paired t-tests (all P > 0.05).

Table 3.

Inter-method consistency by age group (logMAR, ICC, P-Values)

Age(years)	Case	Computerized OKN (logMAR)	Standard Logarithmic E Chart (logMAR)	ICC	ICC (p-value)	t-test (p-value)
3~4	14
Right		0.17 ± 0.12	0.22 ± 0.1	0.41	p < 0.05	0.11
Left		0.18 ± 0.16	0.15 ± 0.14	0.55	p < 0.05	0.48
5	13
Right		0.14 ± 0.10	0.15 ± 0.15	0.63	p < 0.01	0.86
Left		0.22 ± 0.23	0.15 ± 0.18	0.65	p < 0.01	0.17
6	17
Right		0.18 ± 0.30	0.15 ± 0.25	0.91	p < 0.001	0.30
Left		0.21 ± 0.29	0.20 ± 0.27	0.88	p < 0.001	0.74
7~18	37
Right		0.20 ± 0.27	0.20 ± 0.29	0.92	p < 0.001	0.82
Left		0.22 ± 0.27	0.22 ± 0.28	0.89	p < 0.001	0.80
> 18	23
Right		0.46 ± 0.36	0.43 ± 0.35	0.94	p < 0.001	0.22
Left		0.38 ± 0.36	0.39 ± 0.37	0.92	p < 0.001	0.75

Note: 95% CIs are provided in the text for key subgroups

Refractive error groups

Both myopic and hyperopic groups showed good inter-method consistency, but the myopic group had higher ICC values than the hyperopic group (Table 4). No significant differences were found via paired t-tests (all P > 0.05).

Table 4.

Inter-Method Consistency by Refractive Error (logMAR, ICC, P-Values)

Group	Case (eye)	Computerized OKN (log MAR)	Standard Logarithmic E Chart (log MAR)	ICC	ICC (p-value)	t-test (p-value)
Myopia
Right	39	0.41 ± 0.35	0.41 ± 0.34	0.94	p < 0.001	0.96
Left	41	0.36 ± 0.33	0.36 ± 0.33	0.93	p < 0.001	0.79
Hyperopia
Right	33	0.13 ± 0.15	0.14 ± 0.16	0.77	p < 0.001	0.76
Left	30	0.16 ± 0.19	0.11 ± 0.17	0.70	p < 0.001	0.08

VA Ranges

Inter-method correlation weakened as VA improved (Table 5). VA-range subgroups were defined using decimal VA. The highest ICC was observed in the 0.1–0.5 decimal VA subgroup, whereas the 1.0–1.5 decimal VA subgroup showed the lowest ICC. Paired t-tests indicated significant inter-method differences in the 1.0–1.5 decimal VA subgroup (all P < 0.0001).

Table 5.

Inter-method consistency by decimal VA range (logMAR, ICC, P-Values)

Decimal VA	Case	Computerized OKN (logMAR)	Standard Logarithmic E Chart (logMAR)	ICC	ICC (p-value)	t-test (p-value)
0.1~0.5
Right	39	0.50 ± 0.30	0.53 ± 0.23	0.84	p < 0.001	0.25
Left	40	0.48 ± 0.30	0.53 ± 0.24	0.80	p < 0.001	0.13
0.5 ~1.0
Right	51	0.12 ± 0.12	0.11 ± 0.10	0.66	p < 0.001	0.65
Left	53	0.13 ± 0.14	0.09 ± 0.09	0.50	p < 0.01	0.02
1.0 ~1.5
Right	15	-0.01 ± 0.05	-0.09 ± 0.03	0.12	p > 0.05	< 0.0001
Left	12	0.002 ± 0.04	-0.10 ± 0.04	0.12	p < 0.05	< 0.0001

Note: VA strata were defined in decimal VA (Snellen decimal); mean VA values are reported in logMAR; 95% CIs are provided in the text for key subgroups

Discussion

Throughout the past few years, a lot of researchers have investigated how modifying the shape and position of stimuli impacts OKN development [11–13]. Research into OKN has confirmed that its involuntary nature enables VA assessment with minimal participant cooperation [14, 15], making it a promising approach for testing young non-verbal, anxious, or language-impaired individuals who struggle with lightbox VA chart testing [16–17]. To establish a connection between subjective and objective VA, the results of optokinetic VA in an OKN apparatus and lightbox VA charts findings have been compared [18]. Camera-based infrared eye tracking and computer-assisted OKN stimulus control further optimize this approach by enabling precise adjustment of stimulus parameters (e.g., stripe width, contrast) and standardized, objective waveform analysis [19–21]. Computerized OKN systems also address key limitations of traditional OKN drums, such as the “wheel effect” (mitigated via optimized grating spacing in this study) and limited stimulus field, while ensuring consistent brightness, uniform movement speed, and smooth visual stimuli-factors that reduce inattention-related false negative responses, especially in children.

This study demonstrated high testability (98.6%; 219/222 subjects) for an OKN-based VA assessment protocol. Test failure was limited to a small number of cases with tension esotropia, test-related crying, or vertigo, and the device’s standardized testing environment (distraction-free, consistent stimulus parameters) contributed to the repeatability of VA measurements. Inter-method consistency between the OKN analyzer and standard logarithmic E chart was found to be age-dependent: ICC values exceeded 0.88 in children aged ≥ 6 years (right eye: 0.91; left eye: 0.88), while consistency was significantly lower in 3–4 years group: right eye ICC = 0.41, 95% CI [-0.07, 0.76], P < 0.05; left eye ICC = 0.55, 95% CI [0.06, 0.83], P < 0.05. This age-related difference is likely attributable to the continued maturation of the subcortical OKN pathway until approximately 6 years of age [6], alongside improved test compliance in older children and adults—both factors that enhance measurement accuracy.

A key finding was the inverse relationship between VA quality and inter-method consistency: correlation weakened as VA improved, with the highest ICC (0.84, 95% CI [0.72, 0.91]; 0.80, 95% CI [0.65, 0.89], P < 0.05) in the 0.1–0.5 decimal VA range and the lowest (0.12, 95% CI [-0.10, 0.44], p > 0.05; 0.12, 95% CI [-0.06, 0.47], P < 0.05) in the 1.0-1.5 decimal VA range, where paired t-tests revealed significant mean VA differences between methods (P < 0.0001). Although the ICC in the 1.0–1.5 decimal VA subgroup reached statistical significance for the left eye (P < 0.05), the absolute ICC value still indicates very poor agreement, consistent with the significant mean differences observed in the paired t‑tests (P < 0.0001). This aligns with prior reports [22] and is likely due to the limited spatial resolution of the study’s OKN stimuli: fine stripes required to assess high-level normal VA (decimal VA ≥ 1.0; logMAR ≤ 0.0) lacked sufficient contrast to induce detectable nystagmus. Future device iterations could address this by increasing maximum stimulus spatial frequency (e.g., stripe width < 0.5 mm) or implementing dynamic contrast adjustment based on initial VA estimates.

In terms of refractive status, the OKN analyzer showed stronger consistency with standard logarithmic E chart in myopic patients (right eye ICC: 0.94, 95% CI [0.88, 0.97], P < 0.05; left eye ICC: 0.93, 95% CI [0.87, 0.96], P < 0.05) than hyperopic patients (right eye ICC: 0.77, 95% CI [0.59, 0.88], P < 0.05; left eye ICC: 0.70, 95% CI [0.46, 0.85], P < 0.05) This difference may be related to accommodative engagement, with a: during standard logarithmic E chart testing, hyperopic individuals can use active accommodation to compensate for refractive error and resolve distant optotypes; in contrast, the high-contrast black-and-white stripes of the OKN stimulus may reduce the need for active accommodation, leading to a slight underestimation of VA in hyperopic patients [5]. In our refractive error subgroup analysis, inter-method consistency was higher in myopic eyes than in hyperopic eyes. One possible explanation is that hyperopic patients, especially children, may exhibit more variable accommodation during testing, which could affect both chart-based VA and OKN responses. However, accommodative function was not directly measured in this study, so this interpretation remains speculative and should be confirmed in future work. This suggests that refractive correction for hyperopic patients may be necessary to improve the OKN analyzer’s measurement accuracy in clinical use.

Taken together, our findings suggest that the current version of the computerized OKN analyzer is reliable for detecting and quantifying reduced VA in cooperative children aged ≥ 6 years and adults, and in eyes with subnormal acuity. Its performance is less accurate in eyes with normal or near-normal VA, and consistency is limited in 3–4-year-old children. Although the largely language-independent and low-cooperation nature of OKN stimuli implies potential future utility in populations who cannot complete chart testing (e.g., some preschool children or individuals with communication difficulties), our study did not include non-verbal or special-needs populations and did not evaluate children under 3 years of age. Moreover, the reduced accuracy in normal VA implies a risk of misclassifying individuals with normal vision as having subnormal VA. Therefore, the present data do not support routine use of this device as a stand-alone tool for universal preschool screening, general population screening, or special-needs populations. These applications should be regarded as potential future directions that require dedicated validation in appropriately designed studies.

Further research should focus on improving stimulus design and detection algorithms to enhance accuracy across the full range of VA, particularly in the normal range, and on evaluating test–retest reliability. Prospective studies in younger children, including those unable to perform chart-based VA tests, using age‑appropriate subjective methods (such as Teller Acuity Cards), will be essential to determine the true utility of OKN‑based VA assessment in the intended pediatric target populations.

Study limitations

This study has four key limitations that should be considered. First, the cross-sectional design precluded longitudinal assessment of the computerized OKN analyzer’s performance; longitudinal data are needed to determine whether OKN-derived VA can track changes in visual function over time (e.g., before and after amblyopia treatment), and this will be the focus of our subsequent prospective cohort study. Second, children under 3 years of age were excluded from the study due to difficulties in obtaining reliable chart VA measurements, leaving the device’s applicability in this high-risk group unvalidated, which limits the generalizability of our findings to the intended target population (e.g., non-verbal children, patients with cognitive or language disorders, low-cooperation preschoolers). Third, the reduced accuracy in individuals with normal VA limits the device’s use in general population screening, as it may misclassify individuals with normal vision as having subnormal VA. Fourth, we did not measure accommodative response, fixation stability, or attention-related factors, which may influence both OKN and chart-based VA, particularly in young children and hyperopic eyes. As a result, some of the mechanistic explanations proposed for the observed subgroup differences (e.g., age-related maturation, accommodative variability) should be regarded as hypotheses rather than definitive conclusions.

Conclusion

The computerized OKN analyzer demonstrates reliable performance for VA testing in cooperative patients aged ≥ 6 years, with higher agreement in myopic than in hyperopic individuals. In this population, it provides preliminary evidence of feasibility as an objective, language-independent VA assessment method. However, its utility in children aged 3–4 years, in individuals with normal VA, and in non-verbal or special-needs populations remains unproven and requires further validation. This cross-sectional study only verifies the preliminary consistency of the OKN analyzer with traditional VA assessment methods; its validity for longitudinal visual function monitoring and for evaluating clinical interventions must be confirmed by subsequent prospective studies. Future research should also focus on optimizing OKN stimulus parameters (e.g., spatial frequency and contrast) for hyperopic patients and individuals with normal VA, and on conducting large-sample, multicenter studies in broader and less cooperative populations to further clarify its clinical applicability.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

VA

Visual acuity

OKN

Optokinetic nystagmus

SE

Spherical equivalent

ICC

Intraclass correlation coefficients

Author contributions

JF and MXJ designed the study. YYS, JH, BWZ, ZJM and NM joined the data collection. MXJ analyzed the data and drafted the manuscript. CC and ZL provided the instruments. YYS and JF contributed to the interpretation of the results and critical revision of the manuscript for important intellectual content and approved the final version of the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the “Yangfan” clinical technology innovation project, Beijing Municipal Administration of Hospitals (YGLX202506); High-Level Public Health Technical Talents Construction Project of Beijing Municipal Health Commission, (2023-1-2025.12).

Data availability

The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki and received ethical approval from the Ethics Committee of Beijing Tongren Hospital, affiliated with Capital Medical University (Ethics Approval Number: TRECKY2021-174). Informed consent was obtained from all participants’ parents or legal guardians.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.