Integrating KANET and Doppler indices to predict neurodevelopmental delays in high-risk pregnancies

Introduction

Neurodevelopmental disorders (NDDs) affect approximately 10–15 % of children under five years of age, encompassing conditions such as cerebral palsy, autism spectrum disorders, and cognitive delays [1], [2], [3]. These disorders impose substantial long-term health, societal, and economic burdens. Timely identification of neurodevelopmental risk during the prenatal period has the potential to transform outcomes through early intervention [1], [2], [3], [4], [5], [6], [7], [8], [9].

Recent advances in prenatal imaging have expanded clinicians’ ability to assess fetal well-being beyond anatomical surveillance. The Kurjak Antenatal Neurodevelopmental Test (KANET), based on real-time 4D ultrasonography, enables structured behavioral assessments of the fetus through parameters such as isolated movements, facial expressions, and hand-to-face gestures [1], [2], [3], [4], [5], [6], [7], [8], [9], [10]. It has proven effective in high-risk pregnancies, particularly those complicated by intrauterine growth restriction (IUGR), gestational diabetes, or preeclampsia [4], 10], 11] However, KANET is limited to behavioral observation and does not capture placental or fetal hemodynamics.

To address this gap, Doppler ultrasonography offers vital complementary data. It evaluates key fetal-placental vascular parameters, including resistance indices (RI) of the middle cerebral artery (MCA) and umbilical artery (UA), cerebroplacental index (CPI), and ductus venosus (DV) flow [12]. Abnormal findings – such as CPI ratios below 1.0 or reversed DV A-waves – are associated with hypoxia, placental insufficiency, and adverse perinatal outcomes, including preterm birth and low birth weight [12]. Despite their respective strengths, the combined application of KANET and Doppler indices remains underutilized in prenatal neurodevelopmental screening.

High-risk pregnancies – such as those involving hypertensive disorders, multiple gestations, or maternal infections – demand an integrated diagnostic approach that encompasses both neurobehavioral and hemodynamic parameters [2], [13], [14], [15], [16], [17], [18], [19], [20]. Furthermore, literature suggests that fetal behavior and circulatory patterns vary by sex, gestational age, and clinical condition, underscoring the need for a comprehensive model that considers both dimensions [21], [22], [23], [24], [25].

This study aims to evaluate whether integrating KANET with Doppler indices enhances early prediction of neurodevelopmental delays in high-risk pregnancies. The goal is to provide a multidimensional prenatal screening model that improves sensitivity, specificity, and clinical utility.

We hypothesize that this combined approach will outperform either modality alone in predicting adverse neurodevelopmental outcomes. To validate this hypothesis, the study incorporates postnatal assessments using the Denver Developmental Screening Test II (DDST II) and applies advanced statistical modeling.

Figures and tables presented in this study illustrate the foundational evidence and analytic framework behind this hypothesis. Figure 1 highlights the relative predictive strength of KANET and Doppler features, including CPI, as key antenatal variables. Figure 2 presents receiver operating characteristic (ROC) analysis demonstrating the integrated model’s diagnostic performance. Figure 3 outlines the study protocol from initial assessments to follow-up evaluations, while Figure 4 and Table 3 visualize the relationships between Doppler indices, KANET scores, and neurodevelopmental outcomes. These data collectively support the need for integrated screening in high-risk pregnancies and form the basis for our research approach.

Figure 1:

The diagram highlights the feature importance scores of various factors in predicting developmental delays, emphasizing the contribution of specific prenatal metrics. KANET scores from the first visit are the most significant predictor, followed closely by the Doppler cerebroplacental index (CPI) ratio, reflecting their strong relationship with neurodevelopmental outcomes. Other factors such as maternal BMI, resistance index of the middle cerebral artery (RI MCA), and gestational age at delivery contribute less prominently, underscoring the critical role of integrated neurobehavioral and hemodynamic assessments in early risk stratification.

Figure 2:

The ROC curve evaluates the diagnostic performance of the integrated KANET-Doppler model for predicting neurodevelopmental delays. The area under the curve (AUC) is 1.0, indicating perfect sensitivity and specificity, which underscores the model’s exceptional accuracy in distinguishing between normal and delayed outcomes. This result highlights the effectiveness of combining KANET scores and Doppler indices for precise prenatal risk stratification and early detection of neurodevelopmental risks.

Figure 3:

KANET study and Doppler examination protocol. The flowchart outlines the study’s methodology, detailing the progression from participant identification to the development of a predictive model for neurodevelopmental delays. Participants were stratified into low-risk and high-risk groups, with antenatal KANET scoring and Doppler parameters assessed between 28 and 36 weeks of gestation. High-risk cases with borderline or abnormal KANET scores underwent follow-up assessments, and postnatal outcomes were evaluated at 3 and 6 months using DDST II, leading to advanced data analyses including statistical tests and machine learning techniques like SVM and random forest to create a predictive framework.

Figure 4:

Comparative analysis of pregnancy risk and its impact on fetal and neonatal outcomes using Doppler and KANET metrics. The diagrams comprehensively illustrate the relationship between pregnancy risk (low vs. high) and various fetal and neonatal parameters. Scatter plots with regression lines show that high-risk pregnancies are associated with lower RI MCA, CPI ratio, and KANET scores, indicating compromised fetal circulation and neurodevelopmental status. Conversely, RI UA and DV (S/D) ratio are elevated in high-risk cases, reflecting vascular resistance and placental insufficiency. The box plot highlights significant differences in newborn condition distributions, with a higher prevalence of adverse outcomes in high-risk pregnancies. Overall, these visualizations underscore the predictive value of integrating Doppler indices and KANET scoring in prenatal risk assessment and postnatal outcome monitoring.

Materials and methods

Doppler ultrasonography

Doppler ultrasonography was performed alongside KANET scoring to assess placental and fetal hemodynamics. The following parameters were measured:

1. Resistance indices (RI): Middle cerebral artery (MCA) and umbilical artery (UA)

2. Cerebroplacental index (CPI): Ratio of MCA RI to UA RI

3. Ductus venosus (DV): Systolic and diastolic velocities (S/D ratio) and A-wave patterns

Abnormal Doppler findings were categorized based on clinical thresholds, such as CPI <1.0 or reversed A-wave in ductus venosus (Figure 5). These findings were correlated with neurodevelopmental outcomes to evaluate their predictive significance.

Figure 5:

This illustration presents the Doppler ultrasonography evaluation of critical fetal vessels, including the umbilical artery (UA), middle cerebral artery (MCA), and ductus venosus (DV). The waveforms for UA and MCA demonstrate resistance indices (RI), which are integral for calculating the cerebroplacental index (CPI), a marker for placental and fetal circulation balance. The ductus venosus waveform highlights systolic, diastolic velocities, and the A-wave, crucial for assessing fetal cardiac function and hemodynamics. Together, these Doppler measurements provide essential insights when combined with KANET scoring for predicting neurodevelopmental outcomes in high-risk pregnancies.

Results

A total of 111 pregnant women were enrolled, comprising 71 low-risk and 40 high-risk pregnancies. The mean gestational age at the first visit was 28.9 weeks (range: 28.0–36.0). High-risk pregnancies resulted in earlier deliveries, with a mean gestational age of 36.4 weeks compared to 38.4 weeks in the low-risk group (p<0.001). Maternal characteristics are detailed in Table 1.

Obesity was more prevalent in the high-risk group (45.0 vs. 28.0 %; p<0.01), while multiparity was more common among low-risk participants (65.0 vs. 40.0 %). High-risk pregnancies were also associated with lower maternal weight gain (11.4 vs. 13.7 kg; p<0.001), which showed a weak positive correlation with neonatal birth weight (r=0.19).

Neonatal outcomes demonstrated significant differences between groups (Table 2). Infants born to high-risk mothers had lower average birth weights (2,830 g vs. 3,140 g; p<0.001) and were more likely to be admitted to the NICU (15.0 vs. 3.0 %). Developmental delays, assessed using the Denver Developmental Screening Test II (DDST II), were more frequent among the high-risk group at both 3 months (18.0 vs. 3.0 %; p<0.01) and 6 months (25.0 vs. 4.0 %; p<0.001).

Doppler ultrasonography revealed distinct hemodynamic differences (Table 3, Figures 4 and 6). High-risk pregnancies had significantly lower middle cerebral artery resistance indices (MCA RI: 0.67 vs. 0.72; p<0.001) and higher umbilical artery resistance indices (UA RI: 0.62 vs. 0.57; p<0.01). The cerebroplacental index (CPI) was reduced (1.20 vs. 1.50; p<0.001), with CPI <1.0 correlating with both reduced neonatal weight (r=0.42) and increased developmental delays (r=−0.53). Abnormal ductus venosus A-wave patterns were present in 18.0 % of high-risk cases compared to 2.0 % in the low-risk group.

Figure 6:

Comparative trends in Doppler parameters across gestational ages for low- and high-risk pregnancies. This Figure illustrates the trends of six Doppler parameters – RI MCA, RI UA, CPI ratio, DV S-velocity, DV D-Velocity, and velocity ratio DV (S/D) – across gestational ages at the first visit for low- and high-risk pregnancies. Linear regression lines are included to highlight the distinct trends for each risk group. High-risk pregnancies generally exhibit higher RI UA, lower CPI ratio, DV D-velocity, and flatter trends compared to low-risk pregnancies. These findings emphasize the differing Doppler profiles between the risk groups, which could be integral in understanding fetal conditions and tailoring clinical interventions.

KANET assessments further distinguished neurobehavioral profiles (Figures 1 and 7). At the first visit, mean KANET scores were significantly lower in the high-risk group (12.8 ± 2.4 vs. 17.2 ± 1.1; p<0.001). Among high-risk pregnancies, 25.6 % had borderline scores (6–13), while 23.1 % showed abnormal scores (≤5). In follow-up evaluations, KANET scores improved by an average of 1.3 points but remained significantly below those of low-risk cases (p<0.01). Positive correlations were found between KANET scores and both CPI (r=0.52) and neonatal weight (r=0.47), while DV S/D ratios were inversely correlated (r=−0.49). Abnormal KANET scores on the first visit strongly predicted developmental delays at 6 months (r=−0.63).

Figure 7:

The diagram illustrates the correlation between KANET scores and the percentage of normal outcomes based on DDST evaluations at 3 and 6 months postnatally. Higher KANET scores are associated with an increased percentage of normal developmental outcomes, with a steeper improvement observed at 6 months. This trend highlights the predictive value of KANET assessments in identifying neurodevelopmental risks and underscores the importance of early prenatal evaluations for long-term child development.

An integrated predictive model combining KANET and Doppler parameters demonstrated strong diagnostic performance. Receiver operating characteristic (ROC) analysis yielded an area under the curve (AUC) of 0.89 (95 % CI: 0.86–0.92; p<0.001), indicating high sensitivity and specificity (Figure 2). Feature importance analysis highlighted CPI (24.0 %) and KANET scores (23.0 %) as the most influential variables (Figure 1).

Figure 4 and 6 further illustrate Doppler trends and outcome disparities across risk groups. High-risk pregnancies consistently showed elevated UA RI, reduced CPI, and altered ductus venosus flow patterns, which aligned with lower KANET scores and increased incidence of developmental delays. Overall, these findings confirm the utility of integrating KANET and Doppler indices in early prenatal screening. This approach offers a comprehensive and non-invasive framework to identify neurodevelopmental risks in high-risk pregnancies, enabling targeted monitoring and intervention. To complement the scoring framework, Figure 9 provides representative 4D ultrasound images illustrating the key fetal behaviors observed during KANET assessments, such as facial expressions and limb movements.

Discussion

This study offers a comprehensive analysis of the integration of Kurjak Antenatal Neurodevelopmental Test (KANET) and Doppler indices to predict neurodevelopmental delays in high-risk pregnancies [26]. By correlating maternal, fetal, and neonatal parameters, the findings reinforce the clinical utility of combining neurobehavioral and hemodynamic assessments to enhance prenatal diagnostics [1], 2], 4], 5], 12], 13], 27], 28].

Gestational age at delivery was a significant determinant of neonatal outcomes. Earlier deliveries, more common in the high-risk group, were associated with lower neonatal weight and increased developmental delays, consistent with literature indicating that preterm birth disrupts critical phases of brain maturation [6], [7], [8], [9, 13]. This underscores the importance of prolonging gestation and implementing preventive strategies in high-risk pregnancies.

KANET scores, particularly at the first visit, demonstrated strong associations with adverse neurodevelopmental outcomes. Approximately 30 % of high-risk pregnancies exhibited borderline or abnormal KANET scores, which improved modestly over time. This trend may reflect fetal neuroplasticity and maturational changes occurring between 28 and 36 weeks, as well as the compensatory capacity of the developing central nervous system. These observations are in line with prior studies linking abnormal fetal behavior with impaired synaptogenesis and delayed myelination in utero [3], 10], 15], 16].

Doppler indices provided essential insight into placental function. The cerebroplacental index (CPI) was notably reduced in high-risk pregnancies, and values <1.0 correlated strongly with both reduced neonatal weight and delayed developmental outcomes [12], 13], 27]. This reinforces CPI’s value as a non-invasive biomarker of placental insufficiency. Additionally, ductus venosus abnormalities, including elevated S/D ratios and A-wave alterations, were significantly associated with low KANET scores, indicating their complementary role in fetal well-being assessment [2], 20].

The predictive power of combining KANET and Doppler indices was demonstrated through integrated modeling. The receiver operating characteristic (ROC) analysis yielded an area under the curve (AUC) of 0.89, confirming high sensitivity and specificity for detecting developmental delays. Feature importance analysis identified KANET scores and CPI as the most influential predictors (Figure 1), highlighting the synergy between neurobehavioral and vascular assessments [18], 23].

Neurodevelopmental follow-up using the Denver Developmental Screening Test II (DDST II) confirmed persistent delays in 25 % of high-risk neonates at 6 months, compared to only 4 % in the low-risk group [29]. These findings are supported by neuroimaging research demonstrating reduced cortical thickness and connectivity in infants exposed to adverse intrauterine environments [19], 21]. However, the limited sensitivity of DDST II for subtle or long-term impairments suggests the need for additional tools, such as the Prechtl General Movements Assessment (GMA) and the Bayley Scales of Infant and Toddler Development, which offer broader neurocognitive and motor domain evaluations [29], [30], [31], [32].

This study also confirmed the impact of maternal factors on neonatal outcomes. High maternal BMI and inadequate gestational weight gain were linked to lower birth weight and poorer developmental scores, consistent with prior evidence of maternal metabolic influence on fetal programming [22], 24]. Additionally, multiparity was associated with improved neonatal outcomes in low-risk pregnancies, suggesting a possible protective role.

Mediation analysis further revealed that compromised Doppler parameters partially mediated the relationship between pregnancy risk and neurodevelopmental delay, offering insights into the pathophysiological pathways of placental insufficiency and fetal adaptation [13], 25], 27].

Compared to more complex or expensive modalities such as fetal MRI or genetic profiling, the KANET-Doppler integration offers a practical, non-invasive, and scalable screening tool, particularly suited for use in low-resource settings. Advances in artificial intelligence (AI) have further enhanced the reproducibility of KANET scoring and Doppler interpretation. AI-driven algorithms have shown promise in automating 4D ultrasound assessments, improving consistency across operators and reducing subjectivity [13], 25], 27], 28], [33], [34], [35], [36], [37], [38]. Recent innovations have also explored AI recognition of fetal facial expressions to refine neurobehavioral analysis [33], [34], [35], [36], [37], though concerns regarding image quality, operator dependence, and fetal positioning remain [39]. Standardization and validation of AI-assisted protocols are essential for widespread clinical adoption.

The diagnostic strength of the KANET-Doppler model is further underscored by its ROC performance (Figure 2), supporting its integration into routine prenatal screening. By enabling earlier and more accurate identification of at-risk fetuses, this model provides a foundation for personalized obstetric care and timely neonatal intervention [23], 25], 40].

Strengths

Despite these limitations, this study has several strengths. The integration of KANET scoring and Doppler indices provides a holistic assessment of fetal well-being, addressing both neurobehavioral and hemodynamic dimensions. The longitudinal design, encompassing both antenatal and postnatal periods, ensures robust validation of predictive models. Advanced statistical techniques, including machine learning algorithms and mediation analysis, enhance the reliability and clinical relevance of the findings. Furthermore, the inclusion of both low-risk and high-risk pregnancies allows for comprehensive comparisons, highlighting the nuanced effects of maternal-fetal variables on neurodevelopment. Finally, the study’s adherence to rigorous ethical standards and robust data collection methods underscores its contribution to the evolving field of maternal-fetal medicine (Figures 8 and 9).

Figure 8:

The Figure provides a detailed framework for the KANET fetal neurological assessment, including criteria for scoring neurobehavioral parameters during prenatal evaluations. Each parameter, such as isolated head anteflexion, cranial suture assessment, and hand-to-face movements, is scored from 0 to two based on the quality and frequency of movements, with a total score ranging from 0 to 20. The results are categorized as abnormal (0–5), borderline (6–13), or normal (14–20), enabling clinicians to identify and classify potential neurodevelopmental delays in fetuses.

Figure 9:

This Figure visually demonstrates various fetal behaviors evaluated in the KANET neurological assessment using 4D ultrasound imaging. It includes examples of key neurobehavioral parameters such as isolated head anteflexion, cranial suture assessment, isolated eye blinking, and movements like hand-to-face and leg motions. These illustrations provide a clear understanding of the observable fetal actions that contribute to KANET scoring, aiding in the identification of potential neurodevelopmental delays.

Conclusions

This study highlights the clinical value of integrating the Kurjak Antenatal Neurodevelopmental Test (KANET) with Doppler ultrasonography for early prediction of neurodevelopmental delays in high-risk pregnancies. By combining neurobehavioral and hemodynamic assessments, this approach provides a comprehensive, non-invasive framework for prenatal evaluation, enhancing the accuracy of fetal risk stratification. KANET scores and the cerebroplacental index (CPI) emerged as key predictors of developmental outcomes, supporting their routine inclusion in advanced fetal assessments. Early identification of at-risk fetuses through this integrated model enables timely and targeted interventions, potentially mitigating long-term neurodevelopmental impairments.

Beyond fetal metrics, the study reinforces the importance of maternal health factors such as BMI and gestational weight gain, which significantly influence neonatal outcomes. These findings underscore the need for holistic prenatal care that addresses both maternal and fetal risk profiles. While the study was limited by sample size and the exclusive use of DDST II for postnatal follow-up, it lays the groundwork for further research. Future studies should validate these findings across larger, more diverse populations and explore the use of complementary tools such as neuroimaging and developmental scales like GMA and the Bayley Scales.

Technological advancements, particularly in artificial intelligence, hold promise for automating KANET scoring and Doppler interpretation, thereby improving diagnostic efficiency and consistency – especially in resource-limited settings. The integration of AI with prenatal diagnostics could facilitate broader global adoption, offering scalable solutions for equitable maternal-fetal care. This integrated KANET-Doppler model represents a significant advancement in prenatal screening, with the potential to transform care pathways for high-risk pregnancies and improve neurodevelopmental outcomes worldwide.