Predicting motor rehabilitation outcomes in children with low vision: a nomogram-based study

Abstract

Somatosensory technology offers an innovative approach for motor function rehabilitation in children with low vision. However, limited studies have explored the key factors influencing its effectiveness, and predictive models for clinical guidance are lacking. This study included 183 children with low vision who underwent somatosensory-based motor rehabilitation training from January 2021 to December 2023. Participants were randomly assigned to a training set (n = 128) and a validation set (n = 55). Multivariate logistic regression was used to identify independent factors associated with rehabilitation outcomes. A nomogram prediction model was developed based on significant variables. Model performance was assessed using ROC curves, calibration plots, the Hosmer-Lemeshow test, and decision curve analysis (DCA). Rehabilitation was ineffective in 38 children (29.69%) in the training set and 15 children (27.27%) in the validation set. Independent risk factors for poor outcomes included long disease duration, low family support, presence of complications, low training intensity, poor baseline vision, and limited adaptability to somatosensory technology (P < 0.05). The nomogram showed excellent predictive performance, with C-index values of 0.935 and 0.841, and AUCs of 0.941 and 0.843 in the training and validation sets, respectively. Calibration curves indicated good agreement between predicted and observed outcomes. DCA demonstrated clinical utility across a wide range of threshold probabilities. The nomogram effectively predicts rehabilitation outcomes in children with low vision receiving somatosensory-based training. This tool can assist clinicians in developing individualized rehabilitation strategies and improving intervention effectiveness.

Introduction

Low vision is a common visual disorder in children. It not only seriously affects the visual perception of children, but also is often accompanied by the delay and abnormality of motor function development¹. Motor function is crucial for children’s daily life, learning and social interaction. The impaired motor function of children with low vision further limits their physical and mental development and brings heavy burden to family and society².

With the advancement of medical technology, the application of somatosensory technology in rehabilitation has brought new hope for the motor function rehabilitation of children with low vision. Somatosensory technology can capture human movements in real time and convert them into interactions in a virtual environment, creating an immersed and interesting rehabilitation training mode, which is expected to improve the motor function of children³. Somatosensory and VR-based interventions have demonstrated improvements in balance, postural control, upper-limb skills, and motor learning in pediatric populations, particularly in cerebral palsy, with growing evidence in developmental coordination disorder and acquired brain injury^4,5. However, although somatosensory technology is increasingly widely used in the rehabilitation training of children with low vision, its rehabilitation effect is still affected by a variety of factors, and there is a lack of effective tools to predict the rehabilitation effect⁶.

Other rehabilitation approaches for children with low vision include conventional physiotherapy, visual-motor integration programs, and assistive device training. Unlike these traditional methods, our somatosensory-based approach incorporates real-time motion feedback, multi-sensory stimulation, and interactive game-based tasks, which may enhance engagement and motor learning. This study aims to identify key factors influencing motor rehabilitation outcomes in children with low vision and to develop a validated nomogram for individualized prediction. This will help clinicians more accurately assess rehabilitation potential, design personalized programs, enhance the efficiency and quality of training, and ultimately improve both the quality of life and overall development of children with low vision.

Methods

Subjects

A total of 183 children with low vision who received motor function rehabilitation training based on somatosensory technology in our hospital from January 2021 to December 2023 were selected as the research subjects. Inclusion criteria: (1) Low vision confirmed by ophthalmic examination; (2) Age 3–12 years old; (3) No other nervous system diseases or physical diseases that seriously affect motor function; (4) Being able to cooperate to complete at least one course of somatosensory technical motor function rehabilitation training; (5) Children’s parents sign informed consent form. Exclusion criteria: (1) Suffering from systemic diseases such as congenital heart disease and severe pulmonary disease that may affect rehabilitation training; (2) Recent history of eye surgery or active inflammation of the eye; (3) Combined mental disorder or mental retardation, unable to understand and cooperate with training. This study was approved by the Ethics Committee of our hospital. Both training and validation sets received identical rehabilitation protocols; the distinction was purely methodological for model development and validation.

Rehabilitation training methods

Motor function rehabilitation was delivered using the Microsoft Xbox Kinect V2.0 somatosensory system, with all participants—both in the training and validation sets—receiving the same standardized intervention supervised by certified therapists. Each session lasted 45 min, conducted three times per week over a 12-week program. The training followed a structured, task-oriented paradigm tailored to each child’s age and functional level, incorporating three core modules: balance training, coordination tasks, and postural control exercises. Although the intervention focused on motor function, it was embedded within a somatosensory framework. Using Kinect’s real-time motion capture, children interacted with virtual avatars, receiving visual feedback and auditory prompts that guided movement accuracy and timing. To stimulate somatosensory processing, the training incorporated sensory perturbations—such as delayed or distorted feedback, reduced visual contrast, and unpredictable target shifts—to challenge visual–proprioceptive integration. Task difficulty was dynamically adjusted by therapists based on child performance, fatigue levels, and physiological responses (heart rate, respiratory rate), ensuring progression from low to high intensity while maintaining safety and engagement. This multi-sensory, adaptive protocol aimed to enhance sensorimotor integration and functional recovery through enriched perceptual–motor experiences.

Assessment of motor function

Before and after the training, the same professional rehabilitation physician was used to assess the motor function of the children using the Peabody Movement Development Scale (PDMS-2), which included gross motor function (such as standing, walking, running, and jumping) and fine motor function (such as grasping, writing, and using chopsticks). According to the evaluation results, the rehabilitation effects were divided into effective (improvement of motor function score ≥ 10 points) and ineffective (improvement of motor function score < 10 points) based on previous report⁷.

Vision assessment

Visual acuity was assessed using the LogMAR visual acuity chart. Under standard lighting conditions, the children were separated from the visual acuity chart by a certain distance (generally 5 m or adjusted according to the actual situation). One eye was covered, and the visual indicators on the visual acuity chart were identified with the other eye in sequence. The examiner records the visual acuity value corresponding to the smallest visual marker that the child can correctly identify. A smaller visual acuity value indicates better visual acuity; A larger visual acuity value indicates worse visual acuity. The extent of visual impairment was classified according to prespecified criteria as mild (LogMAR visual acuity ranged from 0.1 to 0.3), moderate (0.3 to 0.6) and severe (> 0.6). To ensure the accuracy of the assessment, measurements should be performed individually for each eye and repeated for multiple times to obtain the average as the final visual acuity assessment.

Training intensity evaluation

Training intensity evaluation integrates many factors. For the difficulty aspect of training content, various tasks were graded according to their complexity, and corresponding scores were assigned to calculate the weighted total training score. The training load was determined by monitoring the physiological indicators such as heart rate, respiratory rate, and exercise energy consumption with wearable devices in combination with the subjective fatigue score (VAS). Low intensity: The training content was mainly low-difficulty tasks, and the weighted total score was less than 30 points. During the training, the heart rate was lower than 60% of the maximum heart rate, the changes in respiration and energy consumption were small, the VAS fatigue score was 0–3, and the training was easy, thus it was suitable for the initial adaptation stage. Moderate intensity: contains appropriate amount of each difficulty task, and the weighted total score is 30–70 points. With heart rate of 60%–80%, reasonable breathing pattern and energy consumption, and VAS score of 4–6 points, the machine can be effectively improved without excessive fatigue. High intensity: High-difficulty tasks were dominant, and the weighted total score was greater than 70 points. Heart rate over 80%, tachypnea, high energy consumption, VAS score of 7 to 10 points, heavy load on the body, we should pay attention to the tolerance and recovery. Criteria for increasing intensity included: heart rate < 75% of max HR during peak task phases, VAS fatigue score < 4, and task completion accuracy > 80%. If fatigue exceeded VAS ≥ 7 or if movement quality deteriorated (e.g., accuracy < 50%), intensity was reduced to avoid overtraining.

Training compliance assessment

Training compliance was judged from attendance, task completion ratio, and action accuracy. Poor: attendance rate < 60%, task completion rate < 60%, and action accuracy average score < 1 point; Medium: The attendance rate was 60%-80%, the task completion rate was 60%-80%, and the action accuracy score was 1–2 points. Good: Attendance rate ≥ 80%, task completion rate > 80%, and action accuracy score of 2–3 points. For attendance, low compliance was manifested as frequent lateness and early departure; Low completion of tasks may be due to non-cooperation or insufficient capacity; The poor accuracy of movements reflected the poor quality of movement completion.

Assessment of somatosensory technology adaptability

During the somatosensory technology training, the high-precision sensor provided by the somatosensory device or the external professional motion capture system was used. Poor adaptability of somatosensory technology: The average difference of joint angle deviation is greater than 5 degrees, and the average deviation distance of motion trajectory is greater than 10 cm or the variance is greater than 25. For example, when grabbing objects by hand, the angle deviation of joints such as shoulder and elbow is large, the virtual walking track deviates far from the preset track, and the equipment is difficult to accurately capture the conversion motion. In the adaptability of somatosensory technology, the average difference value of joint angle deviation is 3–5 degrees, and the average deviation distance or variance of motion trajectory is 5–10 cm or 10–25. Actions and equipment requirements have fit, but accuracy, stability to be improved. The somatosensory technique has good adaptability: the average difference of the joint angle deviation is less than 3 degrees, the average deviation distance of the motion trajectory is less than 5 cm, and the variance is less than 10. Motion and equipment highly adaptive, efficient interaction.

Family support

Low: Parents seldom accompanied the children in training, did not understand the training plan, did not participate in family-assisted training, lacked confidence in the rehabilitation of children, and had little communication with the medical staff. Medium: Accompanied for training occasionally, basically understood the training plan, was able to conduct part of home-based auxiliary training, had certain expectations for rehabilitation, and communicated with the medical staff generally. Gao: She often accompanied the training, was familiar with the training plan, and actively carried out family assisted training. She was full of confidence in rehabilitation, and took the initiative to communicate with and cooperate with the medical care.

Data collection

Clinical data of the children were collected, including age, gender, course of disease, etiology (congenital eye diseases, acquired eye diseases, etc.), degree of vision impairment (mild, moderate, and severe), and whether other eye abnormalities (strabismus and nystagmus) were combined. Rehabilitation training of the children was also recorded, including training frequency, training intensity, training compliance, and somatosensory technology adaptability. After the rehabilitation training, the standardized evaluation tools were used to evaluate the rehabilitation effects of the children, including vision improvement and motor function recovery.

Statistical methods

Statistical software SPSS 26.0 was used for data analysis. The counting data were compared by χ test or Fisher’s exact probability method. When the measurement data conformed to the normal distribution, it was indicated by S; when the two groups did not conform to the normal distribution, it was indicated by M (Q1, Q3); when the two groups did not conform to the normal distribution, it was indicated by Mann-Whitney U test. Multivariate Logistic regression analysis screened the influencing factors of rehabilitation effect, and the difference was statistically significant (P < 0.05). The nomogram model was established by using R software “rms” software package, and the predicted value of the model was analyzed by using “pROC” software package. Bootstrap method was used to verify the model internally, and the calibration curve between the predicted results and the actual results was drawn. Cases with missing values for predictor variables were excluded from the regression analyses (complete-case analysis). Sensitivity analysis was performed to confirm that exclusion did not significantly alter the results.

Results

Comparison of rehabilitation effects and clinical features between the training set and the verification set

Among the 128 children in the training set, 38 cases (29.69%) had ineffective rehabilitation, and 15 cases (27.27%) of 55 children in the verification set had ineffective rehabilitation. There was no significant difference in rehabilitation effects and clinical features between the training set and the verification set (P > 0.05), as shown in Table 1.

Table 1.

Comparison of clinical characteristics between training set and verification set.

Index		Training set (n = 128)	Validation set (n = 55)	t/χ²	P
Age (years)		6.34 ± 2.15	6.18 ± 2.34	0.449	0.653
Course of disease (month)		16.25 ± 10.56	15.82 ± 11.23	0.247	0.804
Gender	Male	70(54.69)	32(58.18)	0.190	0.662
	female	58(45.31)	23(41.82)
Cause of a disease	Congenital	85(66.41)	38(69.09)	0.125	0.722
	Acquired	43(33.59)	17(30.91)
Degree of visual impairment	mild	38(29.69)	19(34.55)	0.810	0.667
	Moderate	52(40.63)	23(41.82)
	Serious	38(29.69)	13(23.64)
Combined with other ocular abnormalities	Be	48(37.50)	20(36.36)	0.021	0.884
	No	80(62.50)	35(63.64)
Family support	Low	30(23.44)	13(23.64)	0.035	0.982
	Middle	45(35.16)	20(36.36)
	High	53(41.41)	22(40.00)
Training frequency	≤ 3 times per week	58(45.31)	26(47.27)	0.059	0.807
	> 3 times per week	70(54.69)	29(52.73)
Intensity of training	Low	32(25.00)	15(27.27)	0.433	0.805
	Middle	50(39.06)	23(41.82)
	High	46(35.94)	17(30.91)
Training compliance	Poor	28(21.88)	13(23.64)	0.077	0.962
	Moderate	42(32.81)	18(32.72)
	Good	58(45.31)	24(43.64)
Somatosensory technology adaptability	Poor	30(23.44)	14(25.45)	0.457	0.765
	Moderate	47(36.72)	22(40.00)
	Good	51(39.84)	19(34.55)
Basal visual acuity level (LogMAR visual acuity value)		0.82 ± 0.25	0.85 ± 0.23	0.762	0.447
Complications	With	15(11.72)	8(14.55)	0.279	0.596
	Without	113(88.28)	47(85.45)

Analysis of influencing factors of rehabilitation effect in training set

The results of single factor analysis of the training set showed that there were significant differences in the course of disease, family support, complications, training intensity, basic visual acuity level and somatosensory technology adaptability between the effective and ineffective rehabilitation groups (P < 0.05), as shown in Table 2. The rehabilitation effect was taken as the dependent variable (0 = effective, 1 = ineffective), and the factor P < 0.2 in the single factor analysis was taken as the covariate for multi-factor Logistic regression analysis, and the assignment operation was performed according to Table 3. The results showed that the long course of disease, low family support, complications, low training intensity, low basic vision level and poor adaptability of somatosensory technology were the independent factors for the rehabilitation effect of children with low vision (P < 0.05). In the regression model, the tolerance of each variable was > 0.1, VIF was < 10, condition index was < 30, and the proportion of variances of multiple covariates was > 50% without the same feature value. Hence, there was no collinearity of each covariate. See Table 4.

Table 2.

Univariate analysis of rehabilitation effects in training set.

Index		Ineffective group (n = 24)	Effective groups (n = 104)	t/χ²	P
Age (years)		6.85 ± 2.11	6.31 ± 2.03	1.166	0.245
Course of disease (month)		20.51 ± 12.30	15.62 ± 9.87	2.085	0.039
Gender	Male	13(54.17)	57(54.81)	0.003	0.954
	Female	11(45.83)	47(45.19)
cause of a disease	Congenital	16(66.67)	69(66.35)	0.001	0.976
	Acquired	8(33.33)	35(33.65)
Degree of visual impairment	Mild	4(16.67)	34(32.69)	3.122	0.209
	Moderate	10(41.67)	42(40.38)
	Serious	10(41.67)	28(26.92)
Combined with other ocular abnormalities	Be	11(45.83)	38(36.54)	0.713	0.398
	No	13(54.17)	66(63.46)
Family support	Low	12(50.00)	18(17.31)	12.471	0.002
	Middle	6(25.00)	29(27.88)
	High	6(25.00)	57(54.81)
Training frequency	≤ 3 times per week	16(66.67)	48(46.15)	3.282	0.070
	> 3 times per week	8(33.33)	56(53.85)
Intensity of training	Low	12(50.00)	20(19.23)	9.863	0.007
	Middle	6(25.00)	44(42.31)
	High	6(25.00)	40(38.46)
Training compliance	Poor	8(33.33)	26(25.00)	2.990	0.224
	Moderate	10(41.67)	32(30.77)
	Good	6(25.00)	46(44.23)
Somatosensory technology adaptability	Poor	12(50.00)	23(22.12)	9.952	0.006
	Moderate	8(33.33)	32(30.77)
	Good	4(16.67)	49(47.11)
Basal visual acuity level (LogMAR visual acuity value)		0.95 ± 0.32	0.78 ± 0.25	2.841	0.005
Complications	With	10(41.67)	22(21.15)	4.376	0.036
	Without	14(58.33)	82(78.85)

Table 3.

Variable assignment table.

Variable	Meaning	Evaluation
X1	Course of a disease	continuous variable
X2	Family support	0 = high, 1 = medium, 2 = low
X3	Intensity of training	0 = high, 1 = medium, 2 = low
X4	Somatosensory technology adaptability	0 = good, 1 = medium, 2 = poor
X5	Basic visual acuity level	continuous variable
X6	Complications	0 = none, 1 = yes
Y	Rehabilitation effects	Valid = 0, Invalid = 1

Table 4.

Multivariate logistic analysis of influencing factors on rehabilitation effect of training set.

Factor	B	standard error	Wald	P	OR	95% confidence interval
Course of a disease	0.043	0.021	4.110	0.043	1.044	1.001 ~ 1.089
Family support	0.934	0.290	10.413	0.001	2.546	1.443 ~ 4.491
Intensity of training	0.758	0.309	6.030	0.014	2.134	1.165 ~ 3.907
Somatosensory technology Adaptability	0.903	0.300	9.063	0.003	2.466	1.370 ~ 4.439
Basic visual acuity level	2.408	0.896	7.224	0.007	11.116	1.920 ~ 64.369
Complications	0.979	0.479	4.186	0.041	2.662	1.042 ~ 6.802

Establishment of nomogram prediction model for ineffective rehabilitation

Based on the independent influencing factors identified by multivariate Logistic regression analysis, an nomogram prediction model for poor rehabilitation of children with low vision was constructed, and the independent influencing factors in the model were assigned to calculate the total score of poor rehabilitation prediction, which was reflected in the probability of poor rehabilitation prediction. The higher the total score was, the higher the accuracy was in predicting the poor rehabilitation of children with low vision (Fig. 1).

Fig. 1.

Nomogram prediction model for poor rehabilitation effect.

Assessment and validation of nomogram prediction model for rehabilitation effect

In the training set and verification set, the C-index values of the nomogram models were 0.935 and 0.841, respectively, the mean absolute errors of the predicted values and the true values in the calibration curves were 0.062 and 0.148, respectively, and the results of Hosmer-Lemeshow test were χ = 8.244, P = 0.409 and χ = 14.111, P = 0.078, respectively. The ROC curves were displayed in the training set and the verification set, and the AUC of the nomogram model for predicting rehabilitation effect was 0.941 (95% CI: 0.874-1.000) and 0.843 (95% CI: 0.712–0.973), respectively (Fig. 2). The sensitivity and specificity were 0.857, 0.961, and 0.600 and 0.893, respectively. The calibration plot (Fig. 3) indicated good agreement between predicted and observed outcomes.

Fig. 2.

ROC curves of training set (A) and verification set (B).

Fig. 3.

Calibration curves of training set (A) and verification set (B).

Analysis of decision-making curve of rehabilitation effect nomogram prediction model

The decision curve showed that when the threshold probability was about 0.1–0.9, the application of the nomogram model constructed in this study to predict the rehabilitation effects of children with low vision had more clinical benefits than the decisions that were all considered valid or invalid before surgery, as shown in Fig. 4.

Fig. 4.

Calibration curves of training set (A) and verification set (B).

Discussion

Low vision is a common visual disorder in children, which seriously affects their visual perception and motor function development and brings many inconveniences to their life and learning. The application of somatosensory technology in rehabilitation provides a new opportunity for the motor function rehabilitation of children with low vision. However, the rehabilitation effect is affected by many factors. Therefore, in-depth study of these influencing factors and building an effective prediction model are of great significance for optimizing the rehabilitation program and improving the rehabilitation effect. In this study, although the somatosensory technique was effective in the rehabilitation training for children with low vision, some children still failed to achieve the ideal rehabilitation effect, suggesting that we need to further explore the factors affecting the rehabilitation effect to improve the overall success rate of rehabilitation.

The results showed that the long course of disease was an independent factor affecting the rehabilitation of children with low vision. For children with a long course of disease, long-term visual impairment may lead to more severe impacts on the development of motor function and more complex adaptive changes in the neuromuscular system, thus increasing the difficulty of rehabilitation⁸. For example, long-term low vision may cause children to form incorrect postures and exercise patterns during exercise, which are difficult to correct in a short time, thus affecting the effectiveness of rehabilitation training. In addition, with the prolongation of the course of disease, children may have psychological problems, such as inferiority complex and anxiety, which affect their enthusiasm and compliance to participate in rehabilitation training, and then have a negative impact on the rehabilitation effect. Training frequency and gender remained significant in both univariate and multivariate analyses. Higher training frequency may enhance motor learning opportunities, while gender differences could reflect variations in developmental trajectories or activity preferences, as observed in previous pediatric rehabilitation studies.

In clinical practice, for children with a long course of disease, more attention should be paid to the development of personalized rehabilitation program, appropriate extension of the rehabilitation training cycle, strengthen psychological support and counseling, to help children overcome difficulties, and gradually improve the motor function⁹. Low family support was identified as one of the important factors affecting the rehabilitation effect. The family is an important support system in the rehabilitation process of children, and the active participation and encouragement of family members are crucial for the rehabilitation of children. The low degree of family support may be reflected in the parents’ insufficient attention to rehabilitation training and failure to actively assist children in training, such as failing to take children to training on time, and not supervising children in auxiliary training at home¹⁰. The lack of family support may also lead to children not receiving enough emotional encouragement in the training process, which is conducive to the generation of negative emotions and reducing training compliance. On the contrary, children with high family support have more confidence and motivation in the training process, and parents can give positive feedback in time to help children to better master training skills and improve the rehabilitation effect. Therefore, clinical medical staff should strengthen communication and education with the parents of children, improve their understanding of rehabilitation training, encourage parents to actively participate in the rehabilitation process of children, and create a good atmosphere for family rehabilitation¹¹.

The rehabilitation effect of children with complications is poorer. The common complications in children with low vision, such as strabismus and nystagmus, will further affect the visual perception and motor coordination ability of children. For example, strabismus may result in binocular visual dysfunction in children, resulting in inability to accurately determine the position and distance of objects, and deviations during motor function training. Nystagmus decreases the visual stability of children and makes it difficult to focus on the target object, thus affecting the accuracy and efficiency of training¹². Complications may also increase the children’s fatigue and discomfort during training, and reduce their tolerance to training. For children with complications, complications should be actively treated while motor function rehabilitation training is carried out to improve the visual function and the overall situation of children, to improve the rehabilitation effect. At the same time, the training intensity and method should be adjusted according to the specific situation of children in the training process to avoid affecting the rehabilitation process due to the aggravation of complications¹³. Low training intensity has a significant effect on rehabilitation. Appropriate training intensity is one of the key factors to promote the recovery of motor function in children. Low-intensity training may not sufficiently stimulate the neuromuscular system of children to achieve effective training thresholds, which limits the improvement of motor function. For example, in balance training, if the training intensity is insufficient, the balance ability of children is difficult to be fully challenged and improved, and they cannot adapt to the more complex sports environment¹⁴. However, excessive training intensity may also lead to excessive fatigue, injury or resistance in children, which is also not conducive to recovery. Therefore, when formulating the rehabilitation training plan, the training intensity should be reasonably adjusted according to the children’s age, physical condition, basic motor ability and other factors, to ensure that the training is challenging and affordable for the children, to achieve the optimal rehabilitation effect¹⁵. At the same time, we should pay close attention to the children’s reaction in the process of training, and timely adjust the intensity of training, to avoid the impact of improper intensity on the rehabilitation process.

Rehabilitation for children with low basic visual acuity is more difficult. Vision is an important basis for children to perceive the external environment and engage in motor activities. Low basic vision will affect children’s perception and judgment of the surrounding environment, and thus affect the development of their motor function and the effect of rehabilitation training¹⁶. For example, during coordination training, children may not be able to accurately observe the movement demonstration or target object due to poor vision, resulting in decreased movement accuracy and coordination. In addition, low vision may also affect the children’s spatial orientation ability and motor response speed, increasing the complexity of rehabilitation training. For children with low basic visual acuity, the visual characteristics should be fully considered in rehabilitation training, and targeted training methods, such as the addition of auditory cues, tactile guidance and other multi-sensory stimulation, should be adopted to help children to better understand and complete the training tasks, and improve the rehabilitation effect¹⁷. The poor adaptability of somatosensory technology is also an important factor affecting the rehabilitation effect. Somatosensory technology requires children to be able to accurately control their body movements by capturing their movements in real time and transforming them into interactions in a virtual environment, so that the devices can be accurately identified. Children with poor adaptability may have problems in the accuracy of motion capture and the efficiency of task completion, resulting in inability to make full use of the advantages of somatosensory technology for effective rehabilitation training¹⁸. For example, children may be unfamiliar with the operation of somatosensory devices or lack of control ability of their own actions, which causes great deviation between the actions recorded by the device and the actual actions, and affects the accuracy and effectiveness of training feedback. During the training process, we should strengthen the training of children’s somatosensory technology adaptability, through gradually guiding, increase the number of exercises, help children familiar with and master the somatosensory technology, improve their interaction ability in the virtual environment, to better play the role of somatosensory technology in rehabilitation training.

In this study, the nomogram prediction model was comprehensively evaluated based on the C-index index, calibration curve, and ROC curve. In the training set, the C-index index was 0.935, indicating that the model had high prediction accuracy and could well distinguish between effective and ineffective children undergoing rehabilitation. The C-index of the validation set was 0.841, which was slightly lower than the training set but still showed the reliability and extrapolation of the model in the new samples. The calibration curve shows that the prediction model can predict the rehabilitation effect well overall. ROC curve analysis showed AUCs of 0.941 and 0.843 for the training and validation sets, respectively, further confirming the diagnostic efficacy of the model.

The nomogram prediction model has good value in clinical application, but in the use process, the prediction results of the model need to be comprehensively analyzed in combination with the clinical actual situation, to improve the accuracy of the prediction of the rehabilitation effect of children with low vision. Clinicians can use the nomogram by locating each predictor’s value on the corresponding axis, summing the points, and translating the total into a predicted probability of ineffective rehabilitation. This can guide early intervention strategies and resource allocation. While a digital calculator was not developed for this study, such a tool could be easily created and integrated into electronic medical records for point-of-care use. Based on the model prediction results, clinicians can formulate personalized rehabilitation plan for the specific situation of children, including adjusting the training cycle, strengthening psychological intervention, collaborative treatment of complications, adjusting the training method according to the level of vision and improving the adaptability of somatosensory technology, to improve the rehabilitation effect¹⁹. Through understanding the influence of training intensity on rehabilitation effect, doctors can reasonably adjust the training intensity to avoid the adverse effect of too low or too high on the rehabilitation effect. At the same time, the physical response and motor function changes of children are closely monitored to ensure the effectiveness and safety of training.

In view of the negative influence of the course of the disease on the rehabilitation effect, the importance of early detection and timely rehabilitation training was emphasized. At the same time, the rehabilitation of children with low vision needs comprehensive treatment, including ophthalmic treatment, rehabilitation training and psychological support²⁰. Family plays an important role in the rehabilitation of children with low vision. The results of this study emphasized the impact of family support on rehabilitation outcomes. Therefore, health education for parents should be strengthened to encourage them to actively participate in the rehabilitation training of children to form an all-round support system. This is a single-center study, and the sample size is relatively small, so there may be a certain selection bias, and the representation of the research results is limited. Future multicenter, large sample studies are needed to verify the reliability and universality of the results. In addition, the evaluation method for some influencing factors in the study may be subjective, and the evaluation method should be further optimized to improve objectivity and accuracy. In addition, the short-term rehabilitation effect was only observed in this study, and the follow-up time should be extended in future studies to evaluate the long-term rehabilitation effect and the long-term quality of life of the children²¹.

There some limitations in this research. (1) The definitions and sub-components of training compliance and training intensity in this study were based on clinical experience and prior small-scale practice rather than standardized, validated scales. This may limit the generalizability of our findings and should be addressed in future research. (2) In addition, this was a single-center study, which may limit generalizability. The single-center nature of the study may limit the external validity of the findings, as patient characteristics, rehabilitation resources, and clinical practices may differ across institutions and regions. (3) Some variables, such as family support, were assessed using subjective ratings, which may introduce measurement bias. Additionally, the follow-up period was limited to the short term; thus, long-term rehabilitation outcomes remain unknown.

Despite the limitations, this study still provides valuable information for the application of somatosensory technology-based motor function rehabilitation training in children with low vision. Future studies may further explore other factors that may affect the rehabilitation effect, build better prediction models, and pay attention to the application effect of new somatosensory technologies in the rehabilitation of children with low vision to provide more effective rehabilitation treatment for children²². Meanwhile, future studies should include multicenter cohorts with larger sample sizes, extend follow-up durations for longitudinal validation, and explore integrating multimodal sensor data (e.g., wearable motion sensors, eye-tracking) to enhance predictive accuracy.

Conclusions

In summary, this study identified key factors influencing the effectiveness of somatosensory-based motor rehabilitation in children with low vision and developed a nomogram with strong predictive performance. Independent risk factors included longer disease duration, low family support, complications, low training intensity, poor baseline vision, and limited adaptability to somatosensory technology. These findings support the development of personalized rehabilitation strategies, early interventions, and enhanced family involvement, ultimately improving outcomes for children with low vision.

Author contributions

Liangfeng Lin (Conceptualization, Methodology, Writing-Original Draft), Wenjian Shi (Data curation, Formal analysis), Qiaoyu Li (Validation, Software), Wenwen Ye (Resources, Supervision). All authors reviewed the manuscript.

Funding

The study was funded by Natural Science Foundation of Fujian Province (2020J01225).

Data availability

The data could be obtained from corresponding author.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and informed consent

This study was approved by the Ethics Committee of the Second Affiliated Hospital of Fujian Medical University. The informed consents were obtained from participants. This study was performed in compliance with the Helsinki Declaration. Parents and/or legal guardians have given informed consent for the minor to participate in this study.