Eye-Tracking as a Screening Tool in the Early Diagnosis of Autism Spectrum Disorder: A Systematic Review and Meta-Analysis

(Full-text capture 2026-09-21; web artifacts lightly stripped; truncated.)

1RoNeuro Institute for Neurological Research and Diagnostic, 400364 Cluj-Napoca, Romania; cristina.pantelemon@umfcluj.ro (C.T.); livia.popa@umfcluj.ro (L.L.-P.); emanuel.stefanescu@brainscience.ro (E.S.); nicu.draghici@umfcluj.ro (N.C.D.); dafin.muresanu@umfcluj.ro (D.F.M.)

1,†, Lacramioara Eliza Chiperi

Lacramioara Eliza Chiperi

2Monza Ares Hospital, 400347 Cluj-Napoca, Romania

2,*,†, Bianca-Elena Iftimie

Bianca-Elena Iftimie

3Department of Neurosciences, Psychiatry and Pediatric Psychiatry, Iuliu Hațieganu University of Medicine and Pharmacy, 400347 Cluj-Napoca, Romania; biancaiftimie4@gmail.com

3, Livia Livint-Popa

Livia Livint-Popa

4Department of Neurosciences, Neurology and Pediatric Neurology, Iuliu Hațieganu University of Medicine and Pharmacy, 400083 Cluj-Napoca, Romania

1,4, Emanuel Stefanescu

Emanuel Stefanescu

1, Sur Maria Lucia

Sur Maria Lucia

51st Department of Pediatrics, Iuliu Hațieganu University of Medicine and Pharmacy, 400083 Cluj-Napoca, Romania; sur.maria@umfcluj.ro

5, Nicu Catalin Draghici

Nicu Catalin Draghici

4Department of Neurosciences, Neurology and Pediatric Neurology, Iuliu Hațieganu University of Medicine and Pharmacy, 400083 Cluj-Napoca, Romania

1,4, Dafin Fior Muresanu

Dafin Fior Muresanu

4Department of Neurosciences, Neurology and Pediatric Neurology, Iuliu Hațieganu University of Medicine and Pharmacy, 400083 Cluj-Napoca, Romania

1,4

Editor: Irina Esterlis

Author information
Article notes
Copyright and License information

2Monza Ares Hospital, 400347 Cluj-Napoca, Romania

3Department of Neurosciences, Psychiatry and Pediatric Psychiatry, Iuliu Hațieganu University of Medicine and Pharmacy, 400347 Cluj-Napoca, Romania; biancaiftimie4@gmail.com

4Department of Neurosciences, Neurology and Pediatric Neurology, Iuliu Hațieganu University of Medicine and Pharmacy, 400083 Cluj-Napoca, Romania

51st Department of Pediatrics, Iuliu Hațieganu University of Medicine and Pharmacy, 400083 Cluj-Napoca, Romania; sur.maria@umfcluj.ro

Correspondence: lacramioara-eliza.pop@umfst.ro

†

These authors contributed equally to this work.

Roles

Irina Esterlis: Academic Editor

Received 2025 Nov 5; Revised 2025 Dec 8; Accepted 2025 Dec 10; Collection date 2025 Dec.

Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license ( https://creativecommons.org/licenses/by/4.0/).

PMC Copyright notice

PMCID: PMC12733574 PMID: 41464704

Abstract

Background: Early detection of autism spectrum disorder (ASD) is essential, as the first two years of life represent a critical window of neuroplasticity during which timely interventions can improve developmental outcomes. Traditional diagnostic methods, such as ADOS and ADI-R, rely on caregiver reports and structured observations, limiting ecological validity and accessibility. Eye-tracking (ET) offers a non-invasive, scalable approach to assess early atypical gaze patterns. Objectives: This systematic review and meta-analysis synthesized evidence on the diagnostic accuracy of ET for early ASD detection and its potential as an adjunctive screening tool. Methods: A comprehensive search of PubMed, Scopus, Web of Science, Medline, and the Cochrane Library identified studies published between January 2015 and July 2025. Eligible studies evaluated ET in infants and toddlers (≤36 months) for early ASD identification, following PRISMA guidelines. Results: Out of 513 records, 57 studies were included. Most studies reported reduced fixation on social stimuli, atypical gaze following, and preference for geometric over social images in infants later diagnosed with ASD. Pooled effect sizes indicated a moderate-to-large difference between ASD and typically developing groups in social fixation time (Hedges’ g ≈ 0.65, 95% CI: 0.48–0.82, I2 = 58%). Studies integrating machine learning algorithms ( n = 14) achieved improved sensitivity (up to 89%) and specificity (up to 86%) compared with conventional gaze metrics. Conclusions: Overall, ET shows strong potential as an early adjunctive screening method for ASD. Nonetheless, methodological heterogeneity and lack of standardized protocols currently limit clinical translation, underscoring the need for multi-center validation and task standardization.

Keywords: autism spectrum disorder, eye-tracking, diagnostic methods, screening method, neuroplasticity, systematic review, metanalysis

1. Introduction

Autism Spectrum Disorder (ASD) is a neurodevelopmental condition characterized by social communication difficulties, restricted interests, and repetitive behaviors [ 1]. Early signs of ASD can emerge as early as 4–6 months of age [ 2, 3], yet the median age of formal diagnosis remains approximately 4.5 years [ 4]. Delayed diagnosis represents a missed opportunity, as the first two years of life constitute a critical period of neuroplasticity during which early intervention can improve cognitive, language, and social outcomes [ 5].

While clinically relevant, current diagnostic instruments, such as the Autism Diagnostic Observation Schedule (ADOS) and the Autism Diagnostic Interview-Revised (ADI-R) rely heavily on caregiver reporting and structured behavioral observation—conditions that do not fully capture the complexity of real-world social interactions and further delay diagnosis [ 3]. Moreover, they require extensive training and are time-intensive, which diminishes access in disadvantaged communities.

One of the earliest behavioral markers of children with ASD, observed as early as 2–6 months, are differences in gaze behavior [ 2]. Eye tracking (ET) is a procedure that can assess gaze behaviors in real time in response to various types of stimuli—social or non-social, static or dynamic [ 6]. The flexibility of these stimuli provides a closer approximation to real-world social encounters. Gaze behavior is tracked via infrared-based eye-tracking systems and the procedures typically last less than 10 min [ 3]. Thus, the method is feasible for infants or severely impaired children who cannot participate in more demanding procedures [ 3, 4].

One commonly observed finding of ET studies is that individuals with ASD, compared to typically developing (TD) controls, present reduced fixation on face and eye regions [ 3, 7, 8], often associated with a less systematic analysis of facial features [ 8]. Several studies identified that children with ASD prefer non-social stimuli, focusing more on objects or patterns, such as repetitive or geometric ones [ 9, 10]. Furthermore, children with ASD have difficulties in redirecting attention based on the gaze or gesture of another [ 6], a necessary function for joint attention [ 11].

However, other studies have failed to replicate these findings [ 3, 10]. Variables such as age, cognitive ability, and comorbidities can modulate these outcomes [ 3, 6]. Moreover, differences in experimental stimuli and methodology, along with small sample sizes continue to limit cross-study comparability [ 10, 11].

Despite the growing body of research, it remains unclear which gaze-based measures show the greatest robustness across contexts and to what extent between study variability reflects true developmental differences versus methodological heterogeneity. Quantifying the magnitude and consistency of reported effects is therefore necessary before ET can be used reliably as a diagnostic method. Also, standardized stimuli and methodologies must be established to improve cross-study consistency [ 3, 10]. Until then, ET fits the criteria for becoming an adjoining diagnostic and screening step [ 3, 4].

Several attempts were made in order to increase the specificity and sensitivity of ET. One method entails the analysis of both social and non-social stimuli dwell times, resulting in a composite score—ARI, or Autism Risk Index—which has been strongly correlated with ADOS-2 results [ 3]. Recent studies also integrate artificial intelligence (AI) and machine learning (ML) to extract predictive patterns from gaze behavior [ 12, 13, 14], some noting the increase in specificity and sensitivity of diagnostic scores [ 13, 15].

At the point of care, the technical infrastructure required for ET could be largely automated [ 3, 12, 16]. Although initial costs for acquiring the hardware and software components could be substantial [ 3], the implementation at the point of delivery can be managed by trained technicians [ 4]. This approach entails smaller costs for training the personnel, broader availability, making ET a sustainable option for community-based screening and diagnosis [ 3, 16].

Additionally, its repeatability makes it suitable for longitudinal monitoring of developmental trajectories [ 7, 10, 16]. The integration of software and machine learning algorithms allows for objective measures of gaze behavior [ 12, 13, 15]. However, replicability remains constrained by methodological heterogeneity [ 3, 10, 17].

Given these considerations, a systematic review and meta-analysis is warranted in order to synthesize existing findings, formally assess the magnitude and heterogeneity of gaze-related differences, and examine the influence of the methodological factors across studies.

Accordingly, the present study was guided by the following research questions:

(1) What eye-tracking measures of early social attention differentiate infants and toddlers with ASD or an elevated likelihood thereof from typically developing peers?
(2) What is the magnitude and heterogeneity of gaze-based group differences reported across studies?
(3) To what extent do task paradigms and methodological characteristics contribute to variability in reported findings?
(4) What is the current translational potential of eye-tracking measures, including AI- and ML-based approaches, for early ASD risk stratification?

This systematic review and meta-analysis aims to 1. summarize current evidence on ET-based biomarkers for early ASD detection; 2. quantify the effect size of gaze-based differences between ASD and typically developing infants; and 3. identify methodological and technological challenges for clinical translation of ET.

2. Materials and Methods

2.1. Search Strategy

A systematic literature search was conducted in five electronic databases: PubMed, Scopus, Web of Science, Medline, and the Cochrane Library. The search covered the period from 1 January 2015 to 28 July 2025. The following combination of keywords and MeSH terms was used for PubMed:

(“Eye Movement Measurements” OR “eye tracking” OR “gaze behavior”) AND (“autism spectrum disorder” OR “autism” OR “ASD”) AND (“Early Diagnosis” OR “early detection” OR “early identification” OR “infants” OR “toddlers”).

Database-specific syntax was adapted for each platform, and a full search strategy is provided in Supplementary Table S1. Reference lists of included studies and relevant reviews were screened manually to identify additional records. Search filters were restricted to English-language original research involving human infants or toddlers aged ≤ 36 months.

This review followed the PRISMA 2020 guidelines. See Supplementary Materials for details. No review protocol was registered in PROSPERO prior to conducting this study, due to the exploratory nature of the project. Future updates of this review will be registered to ensure full transparency.

2.2. Eligibility Criteria

Studies were eligible if they met the following criteria: (1) Population: Infants or toddlers aged ≤ 36 months with a confirmed ASD diagnosis, elevated familial likelihood of ASD, or high-risk status based on standardized screening tools. (2) Intervention/Index test: Eye-tracking assessment of gaze behavior, social attention, or related metrics; (3) Comparator: Typically developing or non-ASD control groups, or within-group analyses over time; (4) Outcomes: Quantitative measures relevant to ASD diagnosis, such as fixation time, gaze-following metrics, composite indices (e.g., Autism Risk Index), or diagnostic performance (sensitivity, specificity, area under the curve); (5) Study design: Original research (cross-sectional, longitudinal, or case–control) published in English. Several studies initially appeared to meet the inclusion criteria based on title and abstract but were excluded after full-text review for specific reasons (see Supplementary Table S4).

Exclusion criteria included: reviews, editorials, conference abstracts, and case reports; animal studies; studies lacking relevant quantitative data and non-English publications. These exclusion criteria ensured that only studies directly assessing quantitative eye-tracking metrics for social versus non-social visual attention in infants and toddlers (≤36 months) with confirmed or high-likelihood ASD were retained for synthesis.

2.3. Study Selection Process

To ensure transparency and reproducibility, we predefined explicit decision rules for both title/abstract and full-text screening. Two reviewers independently screened all records. At the full-text stage, each article was independently evaluated according to a structured checklist derived from the eligibility criteria (population, paradigm, outcomes, study design, and data extractability) in a non-blinded manner. Disagreements were resolved through discussion; unresolved cases were adjudicated by a third reviewer. All full-text exclusion decisions were documented with a specific exclusion reason, and exported in detail into Supplementary Table S4. The selection process is summarized in Figure 1 (PRISMA 2020 flow diagram). Coding of study characteristics, classification into diagnostic/predictive/descriptive categories, and extraction of eye-tracking metrics were performed independently by two reviewers, following predefined coding rules. Any discrepancies in extracted data (e.g., outcomes, sample characteristics, effect-size inputs) were resolved through consensus.

Figure 1.

PRISMA flow diagram for new systematic review: selection process of included studies.

After full-text review, 84 studies were excluded for the following specific reasons: (1) Unavailable quantitative data ( n = 27): studies reporting only qualitative descriptions, lacking extractable fixation metrics, group means, or diagnostic accuracy parameters; (2) Inadequate population ( n = 18): samples older than 36 months, non-human studies, or studies without ASD/high-likelihood infants; (3) Wrong intervention/index test or wrong focus ( n = 16): studies using paradigms unrelated to social eye-tracking (e.g., oculomotor randomness, EEG–ET combined tasks, visual search tasks); (4) Wrong outcomes or wrong study type ( n = 11): studies that did not provide primary eye-tracking diagnostic metrics or used designs incompatible with extraction (e.g., simulation, case reports, methodological notes); (5) Publication type ( n = 16): reviews, conference abstracts, short communications, editorials, or early online supplements without full data; (6) Language restrictions ( n = 4): studies published in languages other than English for which no translation was available. All exclusion decisions are documented in Supplementary Table S4, with the exact reason for exclusion listed for each study.

2.4. Classification of Study Types

To reduce conceptual ambiguity, all included studies were classified into three predefined categories based on their primary objective:

Diagnostic accuracy studies: Studies that directly compared ASD vs. typically developing (TD) or non-ASD comparison groups and reported metrics related to group discrimination, diagnostic accuracy, or effect-size differences in social fixation.
Predictive or longitudinal studies: Studies that assessed whether early eye-tracking metrics predicted later ASD outcomes, developmental status, or familial high-likelihood trajectories. These studies typically involved infants with elevated likelihood of ASD (e.g., infant siblings) and reported associations with later clinical endpoints.
Descriptive or exploratory studies: Studies that characterized gaze patterns across groups or paradigms without assessing diagnostic performance or predictive value. This includes feasibility studies, paradigm-development studies, and early-phase research using prototype or low-N paradigms.

This classification was used in both qualitative synthesis and interpretation of heterogeneity. Only diagnostic accuracy studies were included in the quantitative meta-analysis. Predictive and descriptive studies were synthesized narratively because their aims, outcomes, and study designs were conceptually distinct from diagnostic accuracy evaluation.

2.5. Data Extraction

A standardized extraction form was used to collect data on: study characteristics (authors, publication year, country, design); participant demographics (sample size, age, sex, risk status); eye-tracking equipment and tasks; outcomes (fixation duration, gaze shifts, joint attention metrics, composite indices); diagnostic performance (sensitivity, specificity, AUC); statistical methods and effect sizes.

2.6. Quality Assessment

Quality assessment of diagnostic accuracy was performed using the QUADAS-2 tool, evaluating risk of bias and applicability concerns across four domains: Patient Selection, Index Test, Reference Standard, and Flow and Timing. For each included study, domains were rated as ‘low risk,’ ‘some concerns/unclear,’ or ‘high risk.’ Detailed domain-by-domain ratings for all studies are provided in Supplementary Table S3.

2.7. Statistical Analysis

2.7.1. Prespecified Analytic Plan

The analytic plan was developed a priori, prior to data extraction, and was based on prespecified hypotheses aligned with the objectives of the review. Specifically, we hypothesized that:

(1) children with ASD would show reduced fixation to social stimuli compared to typically developing controls;
(2) paradigm type (e.g., Geo/Social, dynamic social scenes, joint-attention tasks) would contribute systematically to heterogeneity;
(3) studies using validated, conventional eye-tracking metrics would yield more consistent effect sizes than studies using prototype or ML-enhanced measures.

Based on these hypotheses, we predefined the following analytic rules before reviewing the extracted data: only diagnostic accuracy studies would contribute to meta-analysis; effect sizes would be computed as Hedges’ g for comparability; when studies reported multiple dependent outcomes, they would be aggregated within-study to prevent artificial inflation of precision; predefined subgroup analyses (paradigm category, age group, type of metric) would be performed regardless of statistical significance; predictive and descriptive studies would be excluded from quantitative synthesis, consistent with the conceptual focus on diagnostic performance. These prespecified elements guided all stages of data synthesis, and no analytic decisions were modified based on the pattern of results.

2.7.2. Diagnostic-Accuracy Context

Because most included studies did not report conventional diagnostic metrics (sensitivity, specificity, AUC), effect sizes (Hedges’ g) were used as standardized measures of between-group separation. To align with diagnostic accuracy conventions, we contextualized these effect sizes relative to their expected impact on discrimination performance. Specifically, effect sizes above ~0.80 were interpreted as compatible with large group separations, while smaller effects (<0.50) imply limited diagnostic utility unless paired with high sensitivity/specificity thresholds. This approach enables comparison across heterogeneous paradigms while acknowledging that standardized effect sizes are not substitutes for diagnostic performance metrics.

2.7.3. Meta-Analysis Process

A random-effects meta-analysis was performed to calculate standardized mean differences for social fixation time between ASD and TD groups. Effect sizes were calculated as standardized mean differences (Hedges’ g) to ensure comparability across studies using different social-attention metrics (e.g., percentage of fixation time to faces, eyes, social scenes; dwell time; social/non-social preference ratios). When studies reported multiple eligible social-fixation outcomes, we applied the following strategy:

Selection of a single primary outcome per paradigm: For studies using the same paradigm (e.g., Geo/Social, dynamic social scenes), the outcome representing overall social fixation (e.g., % time on faces or eye region) was prioritized.
Averaging within-study effects: If a study reported several non-independent social-fixation outcomes within the same task (e.g., eyes, mouth, whole face), these were aggregated into a single composite effect size, following recommended procedures for dependent outcomes.
Multiple task paradigms within the same study: When a study included multiple paradigms, we extracted only the outcome that best aligned with the primary aim of this review (i.e., early social attention or social vs. non-social preference), to avoid double-counting participants.

All effect sizes and variances were calculated using reported means, standard deviations. When studies did not report means and standard deviations required for quantitative synthesis, we derived missing values following established procedures. For studies reporting medians and interquartile ranges, means and SDs were estimated. When only standard errors, confidence intervals, or test statistics (t, F) were available, we converted these to SDs using standard formulas. If outcomes were presented only in graphical form, numerical values were extracted using digital plot-reading software (WebPlotDigitizer version 4.6), and SDs were computed from extracted data points. Imputation was performed only when sufficient information existed to support accurate derivation of summary statistics; no assumptions were made when underlying distributional parameters could not be reliably reconstructed. All imputation rules and conversion formulas were applied consistently across studies.

2.7.4. Subgroup Analyses

A random-effects model (restricted maximum likelihood, REML) was used to account for expected methodological heterogeneity across studies. Heterogeneity was quantified using I2, with values < 25%, 25–50%, and >50% interpreted as low, moderate, and high heterogeneity, respectively.

To explore sources of heterogeneity, the following prespecified subgroup analyses were performed: (1) ET paradigm: Geo/Social preference tasks; Dynamic social scenes; Joint attention or gaze-following tasks; (2) Age group: ≤18 months; 19–36 months; (3) Machine learning-enhanced metrics vs. traditional ET measures. Due to insufficient reporting, meta-regression could not be performed.

2.7.5. Sensitivity Analyses

We conducted multiple sensitivity analyses: (1) Exclusion of studies rated as high risk of bias based on QUADAS-2; (2) Exclusion of effect-size outliers (>2 SD from the pooled mean); (3) Leave-one-out analyses to assess the stability of pooled estimates.

Publication bias was assessed using funnel plot asymmetry and Egger’s regression test. All analyses were performed using R (version 4.3.2) with the metafor package.

2.8. Data Synthesis

To enhance conceptual clarity, results were organized into three complementary subsections. First, ‘main behavioral findings’ synthesize outcomes from traditional eye-tracking and pupillometry paradigms examining group-level differences. Second, studies employing machine learning approaches were reported separately due to their distinct analytic pipelines, classification-oriented objectives, and non-comparability with conventional behavioral outcomes. Finally, meta-analytic results were presented in a dedicated section to describe the quantitative aggregation of effect sizes. This structure reflects the methodological heterogeneity of the literature and allows each domain to be interpreted within its appropriate analytic framework.

3. Results

3.1. Included Studies

The database search identified 513 records. After removing 140 duplicates, 373 records were screened by title and abstract, and 287 full-text articles were assessed for eligibility. A total of 57 studies met the inclusion criteria and were included in the systematic review and meta-analysis ( Figure 1, PRISMA 2020 flow diagram).

3.2. Study Characteristics

A summary of individual study characteristics is provided in Supplementary Table S2.

Across the 57 included studies, 9 were diagnostic accuracy studies [ 18, 19, 20, 21, 22, 23, 24, 25, 26], 19 were predictive or longitudinal [ 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45], and 29 were descriptive or exploratory [ 9, 13, 39, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69]. Only diagnostic studies contributed to pooled effect-size estimation, whereas predictive and descriptive studies were analyzed narratively. This stratification was applied to avoid conceptual ambiguity and ensure that meta-analytic inferences reflect actual diagnostic performance rather than developmental associations or feasibility outcomes.

Sample sizes ranged from 18 to 635 participants per study.

Geographically, the 57 studies were conducted across a diverse range of settings, with the largest number originating from high-income countries such as the United States, Sweden, Italy, Switzerland, the United Kingdom, Israel, China, and Japan. A smaller number of studies were carried out in lower-resource settings, including Peru and Malta, often with a focus on feasibility and algorithm development for scalable screening tools.

The included studies encompassed a total of 5214 participants (mean age range: 9–30 months).

Across the included studies, a wide variety of stimuli were employed: static image presentations ( n = 22); dynamic social videos ( n = 18); gaze-contingent paradigms ( n = 10); live, naturalistic interactions ( n = 7).

Eye-tracking hardware included Tobii ( n = 31), SMI ( n = 12), and other infrared-based systems. Stimuli calibration procedures were standardized in 85% of studies.

Given the methodological diversity across included studies, we report findings in three complementary subsections: (1) behavioral eye-tracking and pupillometry results derived from traditional experimental paradigms, (2) machine learning-based classification approaches, and (3) quantitative meta-analysis. This structure enables each type of evidence to be interpreted within its appropriate methodological context.

3.3. Main Behavioral Findings

A total of 51 studies contributed to the behavioral synthesis [ 9, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 53, 55, 57, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72]. These studies examined group-level differences in gaze behavior, pupillometry, social attention, and attentional dynamics.

Across the included studies, eye-tracking paradigms varied in complexity, but several consistent patterns emerged. Most studies converged on differences in social attention allocation, including reduced gaze to faces and atypical orienting toward social cues in children later diagnosed with ASD. Studies using pupillometry reported altered autonomic arousal and slower pupillary light reflex responses, while dynamic gaze-contingent paradigms highlighted atypical processing of interactive social stimuli. Despite methodological diversity, the findings consistently indicated that early differences in visual attention—particularly those related to social processing—were detectable across age ranges, from early infancy to preschool years.

3.4. Conceptual Themes

Across the included literature, several conceptual themes emerged that characterize how eye-tracking and pupillometry differentiate children with ASD from typically developing peers.

(1) Social attention differences as a core feature. A consistent finding across paradigms was reduced attention to socially relevant cues—faces, mutual gaze, biological motion, and joint attention signals. This pattern appeared robust across age groups, including infants at elevated familial likelihood and children with confirmed ASD diagnoses. Studies varied in design but converged on the conclusion that attenuated social orienting represents a stable marker across developmental stages.
(2) Altered autonomic and sensory responsivity. Pupillometry-based studies highlighted atypical modulation of autonomic arousal, particularly slower or blunted pupillary light reflex responses and elevated baseline pupil size in some cohorts. These findings are showing broader differences in sensory responsivity that accompany social attention atypicalities, reflecting potential dysregulation in underlying neurophysiological systems.
(3) Differences in interactive and dynamic processing. Paradigms involving live or gaze-contingent interaction revealed inconsistencies in responsiveness to contingent social cues. Children with ASD or infants who later developed ASD demonstrated reduced adaptation to shifting gaze cues and atypical modulation of attention during dynamic, socially meaningful exchanges.
(4) Developmental trajectories rather than static differences. Longitudinal studies consistently showed that atypicalities in gaze behavior are detectable early, may widen over time, and cor

(Truncated: full text at source URL.)

Ian's Noèsis Wiki

Explorer

Eye-Tracking as a Screening Tool in the Early Diagnosis of Autism Spectrum Disorder: A Systematic Review and Meta-Analysis

Cristina Tecar

Lacramioara Eliza Chiperi

Bianca-Elena Iftimie

Livia Livint-Popa

Emanuel Stefanescu

Sur Maria Lucia

Nicu Catalin Draghici

Dafin Fior Muresanu

Roles

Abstract

1. Introduction

2. Materials and Methods

2.1. Search Strategy

2.2. Eligibility Criteria

2.3. Study Selection Process

Figure 1.

2.4. Classification of Study Types

2.5. Data Extraction

2.6. Quality Assessment

2.7. Statistical Analysis

2.7.1. Prespecified Analytic Plan

2.7.2. Diagnostic-Accuracy Context

2.7.3. Meta-Analysis Process

2.7.4. Subgroup Analyses

2.7.5. Sensitivity Analyses

2.8. Data Synthesis

3. Results

3.1. Included Studies

3.2. Study Characteristics

3.3. Main Behavioral Findings

3.4. Conceptual Themes

Graph View

Table of Contents

Backlinks