Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Apr 1.
Published in final edited form as: Autism. 2019 Oct 24;24(3):658–669. doi: 10.1177/1362361319878578

Identifying Prognostic Markers in Autism Spectrum Disorder Using Eye Tracking

Elizabeth C Bacon 1,2, Adrienne Moore 2, Quimby Lee 2, Cynthia Carter Barnes 1,2, Eric Courchesne 1,2, Karen Pierce 1,2
PMCID: PMC7166165  NIHMSID: NIHMS1539258  PMID: 31647314

Abstract

While many children with ASD are now detected at young ages given the rise in screening and general awareness, little is known regarding the prognosis of early-detected children. The brain is shaped by experience dependent mechanisms; thus, what a child pays attention to plays a pivotal role in shaping brain development. Eye tracking can provide an index of a child’s visual attention, and as such, holds promise as a technology for revealing prognostic markers. In the current study, 49 1–3 year olds with ASD participated in an eye tracking test, the GeoPref Test, that revealed preference for social vs non-social images. Next, children participated in a comprehensive test battery 5–9 years following the initial GeoPref Test. Statistical tests examined whether early-age eye tracking predicted later school-age outcomes in symptom severity, social functioning, adaptive behavior, joint attention, and IQ. Results indicated that toddlers with higher preference for geometric images demonstrated greater symptom severity and fewer gaze shifts at school age. This relationship was not found in relation to IQ or adaptive behavior. Overall, the GeoPref Test holds promise as a symptom severity prognostic tool; further development of eye tracking paradigms may enhance prognostic power and prove valuable in validating treatment progress.

In recent years the field of autism research has progressed significantly, with an incredible push for early identification, clinical and behavioral characterization, and access to early intervention (Bacon et al., 2017; Bacon et al., 2014; Chlebowski, Robins, Barton, & Fein, 2013; Dawson, 2008; Pierce, Carter, et al., 2011; Pierce et al., 2016; Robins et al., 2014; Wetherby, Brosnan-Maddox, Peace, & Newton, 2008; Zwaigenbaum et al., 2015). However, information regarding how early-age ASD symptoms and cognitive skills predict long-term outcomes is severely lacking (Bradshaw, Steiner, Gengoux, & Koegel, 2015). One longitudinal study found that lower IQ scores at age 2 predicted intellectual disability at age 19 years (Anderson, Liang, & Lord, 2014); however, average to above average IQ cases proved much harder to determine, especially differentiating high functioning cases of autism from those who went on to no longer meet criteria for ASD. Many studies have focused on short term outcomes (1–2 years following identification); however, results are mixed, and the ability to predict response to treatment remains limited (Warren et al., 2011). Furthermore, there is conflicting evidence about a wide range of hypothesized early behavioral predictors. For example, some literature has found early cognitive, language, adaptive behavior, age at treatment start, or lower levels of autism-related symptoms predict better short term outcomes, but there is opposing literature for each domain that fails to find the same relationship (see Magiati, Tay, & Howlin, 2012 for review). One reason research in this area has proven difficult is that ASD encompasses a heterogeneous mix of children who have a variety of symptom presentations, responses to intervention, and outcomes (Bacon et al., 2014; Howlin, Magiati, & Charman, 2009). Some children will go on to be acutely impacted by their ASD (i.e. may remain non-verbal, experience difficulty participating in mainstream public schooling, and living independently), while others may participate in mainstream classrooms and progress to independence in adulthood, although they may continue to struggle with nuanced social interactions. In sum, prognostic indicators at very early ages are needed not only to help inform parents of the expected trajectory of their child, but also to inform future research on how to individualize treatment.

Although early clinical and psychometric profiles have not been identified that clearly and consistently predict long-term outcomes, some bio-behavioral markers provide a promising avenue to objectively identify subtypes of ASD. Eye tracking technology is one such promising marker for ASD as it is rapid, non-invasive, inexpensive, easy to administer and interpret, and can be used with a child of any age and any communication level (Loth et al., 2017). Eye tracking tasks examining social attention have identified differences in social attention in children with ASD and have been shown to be associated with clinical tests of social and communication skills demonstrating clinical applicability (Klin, Jones, Schultz, Volkmar, & Cohen, 2002; Moore et al., 2018; Murias et al., 2018; Pierce, Conant, Hazin, Stoner, & Desmond, 2011; Pierce et al., 2016), and may possibly be under genetic influence (Constantino et al., 2017). Eye tracking has also demonstrated utility in predicting short-term outcomes in toddlers with ASD. For example, better visual search accuracy during an eye tracking task performed in infancy has been associated with increased autism symptoms in baby siblings at 15 and 24 months of age (Gliga, Bedford, Charman, Johnson, & Team, 2015). Another study has demonstrated that infants who went on to receive a diagnosis of ASD at age 3 years were less likely to attend to synchronized audiovisual stimuli than typically developing children (Falck-Ytter, Nystrom, Gredeback, Gliga, & Bolte, 2018).

A core early-age symptom of ASD is dysregulation of social orienting and attention. Clinically, some infants with ASD seem to prefer non-social stimuli to social (Pierce, Conant, et al., 2011; Pierce et al., 2016). In two of the largest eye tracking studies in infants, an eye tracking paradigm called the GeoPref Test has shown robust capabilities in identifying visual preferences unique to toddlers with ASD (Pierce, Conant, et al., 2011; Pierce et al., 2016). A key feature of this computer screen-based test, is that toddlers are shown competing colorful images of dynamic non-social geometric patterns and dynamic social episodes of children dancing. Twenty to forty percent of toddlers with ASD show a pronounced preference for viewing the geometric non-social images (>69% of the time viewing geometric images) while the remaining toddlers with ASD show a preference for the social images similar to typical toddlers. Furthermore, this visual preference for non-social, geometric images in a subset of children with ASD is highly specific to children with the disorder (specificity = 98%) and is virtually absent in typically developing children or children with other developmental delays. Additionally, toddlers with ASD who prefer geometric images (>69% of the time viewing geometric images) are more likely to have greater ASD symptom severity, a lower IQ, and lower adaptive functioning skills than toddlers with ASD who prefer the social images (>69% of the time viewing social images). This suggests some eye tracking paradigms may have the power to identify clinically important subgroups of ASD when used at very early ages (Pierce et al., 2016).

A general lack of attention to the social environment could have detrimental effects on the acquisition of fundamental social skills, such as joint social attention, that revolve around orienting to social interactions. Development of joint attention skills in children with ASD is of particular interest given the positive association with later language acquisition and social skills (Kasari, Gulsrud, Freeman, Paparella, & Hellemann, 2012; Kasari, Paparella, Freeman, & Jahromi, 2008; Paul, Campbell, Gilbert, & Tsiouri, 2013). In typical infant development, early-age joint social attention is theorized to be a foundational skill to understanding and sharing communication and social experiences with others. Joint social attention also allows for preverbal communication by providing a social partner information about your interests (Paparella & Freeman, 2015). In fact, infants at high risk for developing ASD (i.e. infants with an older sibling diagnosed with ASD) show a reduced number of gaze alterations in comparison to children at a low risk for ASD (i.e. no older siblings with an ASD diagnosis). Therefore, we believe that eye tracking paradigms that measure non-social and social orienting and attention may have huge potential in ASD clinical prognostic assessment. Such paradigms may tap into basic social attention deficits that are observable at very early ages, and may point to a cascading effect in difficultly acquiring high level social skills. The ability of such eye tracking tasks to measure social and communication difficulties taps into skills central to success in life, including creating relationships with others and landing and maintaining a job.

Currently, there is a lack of evidence regarding the potential utility of eye tracking as a prognostic tool for long-term outcomes in children with ASD. One of the greatest challenges in the ASD field is to address the heterogeneous responses to treatment and move towards providing effective individualized treatments to optimize outcomes for all children. Elucidating predictors of long-term response to treatment would give researchers a starting point for development of novel treatments targeting children least likely to respond to the currently available interventions. The GeoPref Test has already demonstrated potential as a robust marker for identifying and subgrouping clinical phenotypes of toddlers with ASD (Pierce et al., 2016). The current study provides preliminary evidence regarding the utility of the GeoPref Test for predicting long-term outcomes in school age children. Given the differences established in toddlers who participated in the GeoPref Test, we expected later-age clinical outcome differences between toddlers with ASD with vs without a pronounced non-social preference. Experience dependent mechanisms shape early development (Benasich, Choudhury, Realpe-Bonilla, & Roesler, 2014; Butz, Worgotter, & van Ooyen, 2009; Guerreiro, Putzar, & Roder, 2016) and those toddlers who do not focus on the appropriate social aspects of their environment will likely continue to have a difficult time during interactions which may, in turn, affect their learning and social skills. In the current study, children were sub-grouped based on performance on the GeoPref Test as toddlers and, then their clinical and cognitive outcomes at school age were examined. Performance on the Autism Diagnostic Observation Schedule (ADOS) was used as a measure of social functioning and core ASD symptoms at older ages, and an eye tracking joint attention task was used to measure social stimuli-based eye gaze shifting. The Social Responsiveness Scale was used to measure general social and behavioral functioning and outcomes in cognitive and adaptive behavior domains were also measured.

METHODS

Participants

Participants from previous early identification studies were contacted for an additional evaluation to assess long-term outcomes. For the current study, children previously identified with ASD at our Center who also participated in the GeoPref Test were recruited. Toddlers aged 1–3 years from the Pierce et al. 2016 sample were recruited at ages 6–12 years for the present study. All potential participants were notified of the study by mail and were called at random to offer participation in the current study. In previous studies of the GeoPref Test, subgroups of ASD have been identified using the 69% cut-off (Pierce, Conant, et al., 2011; Pierce et al., 2016), with toddlers who show a preference for geometric images for 69% of viewing time or greater going on to have a diagnosis of ASD ~98% of the time. It has also been demonstrated that children who show this extreme preference for geometric images (>69% viewing time for geometric images) show more impairment in language, adaptive, and autism related symptoms in comparison to children who show an extreme preference for social images (>69% viewing time for social images). The current sample consisted of n=22 children with extreme social preference (>69% total looking time towards social images), n=14 with extreme geometric preference (>69% time looking at geometric images), and n=13 with GeoPref Scores falling in the middle (see Table 1). All procedures were approved through the University’s Human Research Protections Program.

Table 1. Participant Demographics at Intake.

Demographics and clinical profile on standardized assessments at intake. Participants are sub-grouped based on performance on the GeoPref Test at intake. Extreme geo preference is defined as viewing geometric images for ≥69% of the time, middle preference is defined as viewing geometric images 32–68% of the time, and extreme social preference is defined as viewing social images ≥69% of the time. Means and standard deviations listed.

ASD Extreme Geo Pref (n=14) ASD Middle Pref (n=13) ASD Extreme Social Pref (n=22)
Age in months 26.2 (8.1) 24.1 (10.0) 23.2 (8.6)
Sex
 Male 12 10 17
 Female 2 3 5
Ethnicity
 Hispanic or Latino 6 4 8
 Not Hispanic or Latino 7 8 13
 Not Reported 1 1 1
Race
 American Indian/Alaska Native 0 0 0
 Asian 0 1 2
 Black/African American 0 0 0
 Caucasian 8 9 14
 Pacific Islander/Native Hawaiian 0 0 0
 Multiple Races Reported 3 1 3
 Not Reported 3 2 3
Mullen T Scores
 Visual Reception 35.7 (12.2) 43.5 (10.1) 42.0 (11.4)
 Fine Motor 34.5 (15.1) 42.1 (13.0) 41.3 (11.5)
 Receptive Language 26.1 (8.6) 31.8 (11.7) 32.5 (11.36)
 Expressive Language 26.6 (11.7) 31.5 (11.4) 32.3 (13.1)
 Early Learning Composite 65.9 (16.1) 79.5 (20.2) 77.4 (17.3)
Vineland Standard Scores
 Communication 80.5 (13.2) 75.8 (15.0) 78.7 (15.1)
 Daily Living Skills 86.2 (12.5) 84.2 (10.9) 85.6 (8.4)
 Socialization 83.4 (13.4) 81.8 (10.3) 84.6 (7.9)
 Motor Skills 94.3 (13.2) 89.3 (8.6) 92.6 (9.7)
 Overall Composite 83.8 (12.6) 80.0 (10.7) 82.6 (8.7)
ADOS Algorithm Scores
 Social Communication 15.2 (3.2) 13.5 (3.4) 12.3 (5.2)
 RRB 4.2 (2.0) 3.4 (2.7) 3.7 (2.5)
Open in a new tab

Children were originally recruited from the general population by using the One-Year Well-Baby Check-Up Approach (Pierce, Carter, et al., 2011) and by community referral at ages 1–3 years. Using this method we partnered with pediatricians to perform standardized developmental screening as standard of care by having parents complete the Communication and Symbolic Behavior Scales Developmental Profile Infant-Toddler Checklist (Wetherby et al., 2008) at well-baby visits starting at age 12 months. Pediatricians then had the option to refer any child who was showing delays to our center for a full developmental evaluation. At our center, toddlers participated in a battery of assessments to assess for delays including autism through a series of measures including a developmental assessment - Mullen Scales of Early Learning (Mullen, 1995), a diagnostic assessment for behaviors associated with autism spectrum disorder - Autism Diagnostic Observation Schedule (Lord et al., 2012; Lord, Rutter, DiLavore, & Risi, 2002), and a standardized parent interview about adaptive skills at home - Vineland Adaptive Behavior Scales (Sparrow, Cicchetti, & Balla, 2005). In addition to the standardized assessments, toddlers also participated in an experimental eye-tracking task to assess social attention, the GeoPref Test, a brief 1-minute assessment of visual preference. Based on overall diagnostic and psychometric performance during the evaluation, children were referred for appropriate services, if necessary. Children were also followed longitudinally every 9–12 months until age three to track development and to confirm diagnoses at age three.

Behavioral Measures

A broad battery of assessments was completed to assess current functioning across domains of cognitive skills, social communication, adaptive skills, and eye tracking (see descriptions below).

Autism Diagnostic Observation Schedule, Second Edition.

The Autism Diagnostic Observation Schedule, Second Edition (ADOS-2; Lord et al., 2012) is a semi-structured assessment used to measure behavioral features of ASD. The appropriate module of the ADOS (i.e. Toddler, 1 or 2) was used as a tool to help inform the clinician’s overall diagnostic judgment. All psychologists were research reliable on administering the ADOS.

Wechsler Intelligence Scales for Children, Fifth Edition.

The Wechsler Intelligence Scale for Children (WISC-V; Wechsler, 2014) assesses cognitive functioning of children, including verbal comprehension, visual spatial skills, working memory, fluid reasoning, and processing speed, to produce an overall composite score.

Vineland Adaptive Behavior Scales, Second Edition.

The Vineland Adaptive Behavior Scales, Second Edition (VABS-II; Sparrow et al., 2005) is a measure of adaptive skills through caregiver report and provides standardized scores for communication, daily living skills, socialization, and motor skills.

Social Responsiveness Scale, Second Edition.

The Social Responsiveness Scale, Second Edition (SRS-2; Constantino, 2012) is a parent rating scale of the severity of social impairments across five domains, receptive, cognitive, expressive, motivation, and preoccupations.

Treatment Survey.

Parents completed a survey regarding past and current treatment received including autism focused services commonly referred to as interventions based on Applied Behavior Analysis (ABA), as well as other developmental services such as speech therapy, occupational therapy, and complementary therapies.

Eye Tracking Tasks

During the eye tracking battery children were seated in front of a computer screen to view short videos. A Tobii 120 eye tracker set at 60Hz was used with children seated at a 60 cm distance from a 17″ thin-film transistor monitor. Five-point calibration was performed with Tobii Studio software using an animated image with sound presented at known X-Y coordinates. Eye tracking data were collected only if the calibration result fell within the parameters reported by the manufacturer to yield an accuracy of 0.5°.

GeoPref Eye Tracking Test.

The Geometric Preference Test is an eye tracking paradigm that analyzes visual stimuli preferences (Pierce, Conant, et al., 2011; Pierce et al., 2016). During the eye-tracking task two dynamic images were presented side-by-side for a total of 60 seconds. One side featured a social stimulus, with scenes of children engaging in aerobics and dancing whereas the other side featured a non-social stimulus, with a series of short sequences of moving geometric shapes. See Figure 1 for sample images. The percentage of time spent looking at social or nonsocial stimuli at initial evaluations was compared across diagnostic groups. Previous research has demonstrated that the Geometric Preference Test has high specificity for ASD and identifies a potentially unique phenotype characterized by a strong preference to visually examine nonsocial, moving geometric stimuli rather than social stimuli (Pierce, Conant, et al., 2011; Pierce et al., 2016).

Figure 1.

Figure 1.

Open in a new tab

Example of the GeoPref Test stimuli. A subset of toddlers with ASD prefer the geometric images and show more impairment in developmental, adaptive, and social skills at young ages as compared to toddlers with ASD who prefer the social images (Pierce, Conant, et al., 2011; Pierce et al., 2016). Image adapted with permission from Pierce, et al. 2011.

Joint Attention Eye Tracking Task.

The Joint Attention Test analyzes gaze shifts during a short two minute video of an actress directly conversing with the child and showing them different toys and objects. Throughout the video the actress overtly directs the child’s attention to an object by shifting her gaze, pointing to it and directing the child to look towards it. Four main segments of the video that contained pointing gestures were analyzed for gaze shifts (see Figure 2). Areas of interests (AOIs) were created for the actress’s face and each object through the video to examine shifts in gaze between the face and the object to capture joint attention. Fixations were classified based on gaze data averaged from both eyes using a velocity threshold Tobii Fixation Filter set to 35 pixels/window, which interpolates to fill in data loss of less than 100ms. Duration of each fixation within each AOI was calculated for every subject and then data was examined to count shifts in gaze between the face AOI and object AOI, providing a total number of gaze shifts for the entire video. Fixations on the face or object were only included if they lasted for 83.3ms or longer to reduce potential for inaccuracy. Subjects with excessive missing data (i.e., less than 60s of data) were excluded, in order to avoid potentially inaccurate lower measurement of gaze shifts.

Figure 2.

Figure 2.

Open in a new tab

Examples of gaze shifting trials during the ‘Joint Attention’ eye tracking task. Panels A-D depict various scenes where the actress directs the child’s attention to an object within the scene. Shifts in gaze between the actress’ face and object are recorded.

Data Analysis

Correlations between performance on the GeoPref Test at intake (mean age 23 months) and clinical scores on the ADOS-2, WISC-V, VABS-2 and SRS-2 at school age (mean age 8 years) were performed to explore relationships between GeoPref performance and long term outcomes. Follow-up linear regressions were performed for significant correlations to evaluate the potential prognostic utility of the GeoPref Scores at young ages. Age at intake and amount of autism-focused services received (e.g. average number of hours of autism focused therapy or ABA-based therapy) were included in the model. These factors were included given they are routinely studied in the literature as potential factors that may affect outcomes (Magiati, Tay, & Howlin, 2012; Makrygianni & Reed, 2010; Reichow & Wolery, 2009). Correlations between performance on the GeoPref Test at young ages and gaze shifts during the Joint Attention Eye Tracking Task at school age were also performed to explore relationships between eye tracking performance at different ages.

RESULTS

Diagnostic Status

At school age, four children no longer met criteria for ASD, although 2 of the 4 continued to demonstrate features associated with ASD, but did not demonstrate all criteria needed for a diagnosis. Clinical scores at school age for children with an early extreme preference for geometric images, extreme preference for social images, and those in the middle are presented in Table 2. Scores are also presented for children who continued to have an ASD diagnosis (categorized at ‘ASD Persistent’) and those who no longer fully met criteria (categorized as ‘Previous ASD’).

Table 2. Participant Demographics at Follow-up.

Demographics and clinical profile on standardized assessment at school age. Participants are sub-grouped based on performance on the GeoPref Test at intake. Extreme geo preference is defined as viewing geometric images for ≥69% of the time, middle preference is defined as viewing geometric images 32–68% of the time, and extreme social preference is defined as viewing social images ≥69% of the time. Clinical information is also presented subgrouping children on whether they continued to meet criteria for ASD (ASD Persistent) or whether they no longer met criteria for ASD at school age (Previous ASD). Means and standard deviations listed.

ASD Extreme Geo Pref (n=14) ASD Middle Pref (n=13) ASD Extreme Social Pref (n=22) ASD Persistent (n=45) Previous ASD (n=4)
Age in Years 7.5 (1.4) 8.4 (1.9) 8.3 (1.3) 8.1 (1.6) 8.0 (0.8)
Sex
 Male 12 10 17 36 3
 Female 2 3 5 9 1
Ethnicity
 Hispanic or Latino 6 4 8 17 1
 Not Hispanic or Latino 7 8 13 25 3
 Not Reported 1 1 1 3 0
Race
 American Indian/Alaska Native 0 0 0 0 0
 Asian 0 1 2 3 0
 Black/African American 0 0 0 0 0
 Caucasian 8 9 14 28 3
 Pacific Islander/Native Hawaiian 0 0 0 0 0
 Multiple Races Reported 3 1 3 7 0
 Not Reported 3 2 3 7 1
WISC Composite Scores
 Verbal Comprehension 81.2 (24.9) 91.8 (23.6) 89.6 (27.4) 87.4 (25.1) 101.0 (11.8)
 Visual Spatial 90.1 (22.4) 92.1 (29.4) 92.1 (25.0) 92.4 (24.0) 109.0 (14.6)
 Fluid Reasoning 84.6 (24.2) 93.8 (23.7) 85.3 (25.9) 87.0 (25.0) 106.5 (18.4)
 Working Memory 81.1 (21.8) 89.5 (23.6) 87.9 (26.6) 86.2 (24.1) 108.6 (16.6)
 Processing Speed 70.1 (19.7) 81.6 (29.4) 79.5 (21.4) 77.3 (23.5) 93.5 (3.8)
 Full Scale 77.8 (22.3) 88.4 (26.3) 85.0 (26.8) 83.8 (24.7) 101.0 (15.0)
Vineland Standard Scores
 Communication 75.0 (22.7) 77.1 (26.2) 81.0 (15.6) 77.8 (21.4) 91.5 (17.7)
 Daily Living Skills 74.1 (23.9) 77.6 (15.2) 75.7 (13.1) 76.1 (17.8) 82.3 (9.6)
 Socialization 75.4 (18.7) 74.2 (11.9) 75.2 (16.0) 75.0 (15.0) 83.5 (20.9)
 Motor Skills 81.3 (19.8) 82.8 (18.6) 81.0 (14.5) 81.9 (16.0) 91.0 (27.4)
 Overall Composite 74.9 (20.3) 76.0 (13.4) 75.8 (13.1) 75.7 (15.5) 84.0 (14.0)
ADOS Algorithm Scores
 Social Communication 11.8 (5.7) 10.7 (3.7) 9.5 (4.7) 11.3 (4.1) 3.0 (2.6)
 Restricted and Repetitive Behavior 4.1 (2.3) 3.9 (2.6) 3.2 (2.3) 3.7 (2.2) 0.5 (0.6)
Social Responsiveness Scale
 Social Communication & Interaction 65.7 (18.0) 72.4 (9.5) 70.5 (17.6) 71.7 (14.8) 47.0 (8.2)
 Restricted and Repetitive Behavior 65.0 (18.8) 74.8 (12.4) 71.52(14.8) 72.0 (15.1) 53.0 (10.5)
Treatment History
 Avg weekly hours of ABA-based therapy 9.0 (5.9) 7.7 (4.3) 11.1 (4.3) 9.4 (5.4) 11.4 (9.6)
Open in a new tab

Treatment Participation

Parents reported a wide variety of services received at varying levels of intensity. A simplistic value of average number of ABA based therapy received per week based on retrospective parent report was calculated. While this value does not capture all the nuances of treatment received, it provided a basic quantifiable means to estimate the intensity of treatment received by each child. On average, children received 9.6 hours of ABA based therapy a week ranging from 0–26 hours a week. There were no significant differences in amount of treatment received for children that demonstrated a preference for geometric images versus social images.

Clinical Scores

When examining clinical results of the 45 children who continued to have a diagnosis of ASD, GeoPref scores at young ages were most closely associated with scores on the Autism Diagnostic Observation Schedule (ADOS) at school age, with a higher score on the ADOS indicating more impairment. A Shapiro-Wilks test of normality indicated the GeoPref data at toddler age were not normally distributed (p = .001) and therefore a nonparametric test was used. A Spearman’s correlation revealed a significant positive correlation (rs = .427, p=.003), with children who spent more time looking at geometric images at young ages, showing more impairment on the ADOS at school age. Furthermore, a linear regression was performed to predict ADOS Total Scores at school age (mean age 8 years) based on initial GeoPref Scores (mean age 23 months), age at intake, and average number of hours of autism-focused treatment. A significant regression was found (F(3,41) = 3.81, p = .016), with an R2 of .221. GeoPref scores at intake were the only significant predictor in the model (β = .068, p = .012). The GeoPref test at intake was not correlated with scores on the WISC-V, VABS-II or the SRS-2 at school age.

Eye Tracking

GeoPref Test.

Thirty-six subjects were included in the analysis. Subjects were required to attend to >50% of the video to be included in the analysis, and 11 subjects were excluded due to insufficient data. Two subjects were excluded due to technical difficulties. No other subjects were excluded. Of the 36 subjects, 19 originally demonstrated a geo preference and 17 a social preference. There was not a significant correlation between GeoPref performance as a toddler and performance at school age. Overall, children demonstrated a higher percent fixation duration for geometric images at older ages.

Joint Attention Task.

Thirty- eight subjects were included in the analysis. Three subjects were excluded due to technical difficulties, and eight subjects did not attend to the video long enough to produce useable data. Of the 38 subjects, 18 were originally classified with a geo preference, and 20 were with a social preference. A Spearman’s correlation revealed a significant negative correlation (rs = −.412, p=.010), with children who spent more time looking at geometric images at young ages, showing fewer shifts in gaze during the Joint Attention Task performed at school age (see Figure 4).

Figure 4.

Figure 4.

Open in a new tab

Correlation between GeoPref scores at young ages (mean age 23 months) and number of gaze shifts during a Joint Attention eye tracking task at school age (mean age 8 years). Children with preference for geometric images are depicted in red, children with preference for social images are depicted in blue. Children who spent more time looking at geometric images at young ages, performed fewer gaze shifts at school age.

Relationship between Clinical Scores and GeoPref Scores

In order to broadly examine the relationship between GeoPref Scores and ADOS Total Scores for all children, scores for all 49 participants, including those who no longer met criteria for ASD were examined. ADOS Total scores and GeoPref scores at intake were correlated with ADOS Total Scores at school age. A Spearman’s correlation revealed a significant positive correlation between ADOS Total Scores at intake and at school age. (rs = .367, p=.009). There was also a significant positive correlation between GeoPref scores at intake and ADOS Total Scores at school age (rs = .345, p=.015).

DISCUSSION

In one of the first tests of the ability of early-age eye tracking in ASD to predict long-term clinical outcome 5–9 years later, we find strong evidence that the GeoPref Test can predict later symptom severity and gaze shifting during joint attention. In contrast, eye tracking results during the toddler period were not associated with later IQ or adaptive functioning. In our study, stronger non-social preference scores were correlated with higher, more severe ADOS scores and reduced gaze shifting during the joint attention task at school-age. These relationships complement our previous research on the GeoPref Test showing a relationship between GeoPref scores in infancy and short-term outcomes at age three, with toddlers who demonstrated a preference for non-social geometric images showing more impairment and higher ratings of ASD symptoms than toddlers who preferred social images (Pierce et al., 2016). The fact that scores on the GeoPref Test - a simple 60 second social/non-social preferential looking task – were correlated with symptom severity many years later, highlights the robust ASD symptom features that the GeoPref Test is capturing at very early ages. Furthermore, when comparing beta weights in a regression analysis, GeoPref scores at early-age intake demonstrated a stronger relationship with later-age ADOS symptoms than with age at first diagnosis and treatment start, or how much autism-focused treatment they received. Additionally, the GeoPref test was not correlated with scores on the WISC-V or the VABS-II, suggesting the GeoPref Test captures information most relevant to core features of ASD rather than general cognitive or adaptive functioning level.

However, the lack of differences found on the WISC-V could be driven by small sample size. On average there was a 9 point difference in Full Scale IQ scores between children in the extreme Geo Pref group and extreme Social Pref group. Therefore, within a larger sample, we may find the GeoPref Test provides some prognostic information regarding cognitive abilities as well. The GeoPref Test was also not correlated with the SRS-2 a measure of social functioning. It may not be surprising that a relationship was not found with the SRS-2, it is hypothesized the ADOS-2 is a more representative measure of ASD core symptoms. One limitation of the SRS-2 is the use of a single form for all individuals of a large age range, in comparison to the ADOS-2 that utilizes different modules based on age, and language level. Other researchers have found that SRS-2 may exaggerate severity of symptoms in older children, children with lower language ability, and those with behavior problems. Therefore, the SRS-2 may not be capturing ASD-specific symptomatology and applicability of the measure may be diluted within a heterogeneous population (Hus, Bishop, Gotham, Huerta, & Lord, 2013). Overall, the GeoPref Test seems to specifically measure the ASD specific symptoms that are best captured using the gold-standard diagnostic tool of the ADOS-2.

Four of the 49 children no longer continued to meet criteria for ASD at school age. Two of the four children continued to demonstrate some symptoms associated with ASD but did not meet full criteria for a diagnosis. Within the literature, it has been reported that 3–25% (Helt et al., 2008; Solomon et al., 2018) of children will no longer continue to meet criteria for the disorder at older ages. The exact percent to expect continues to be somewhat unclear as diagnostic criteria have changed, and variation in sample size and ascertainment across studies may affect estimates. Our rate of 8% of children no longer meeting criteria for ASD at older ages falls right in the mid-range of the expected number of cases. Within our sample, the four children who no longer met full criteria were equally as likely to show a preference for geometric or social images at a young age. These children may even represent a different etiology altogether, and the GeoPref Test may not work as well as a prognostic tool for this particular group of children with optimal outcomes.

Not surprisingly, performance on the GeoPref Test at toddler age was not correlated with performance on that same test an average of 6-years later at school age. The GeoPref Test was developed as an early biomarker test of autism and was specifically designed to capture the attention and interest of infants and toddlers and may capture attention differently in school-age children. This is consistent with previous findings that report that children tend to show an increased preference for the geometric images as they age (Pierce et al., 2016). Interestingly, performance on the GeoPref Test at toddler age was associated with performance on a Joint Attention eye tracking task at school age. The Joint Attention Task provides an objective, quantitative rating of the child’s joint attention skills during a passive viewing task, which may provide an estimation of the child’s ability to engage in gaze shifting in social settings. The GeoPref and Joint Attention eye tracking tasks appear to be measuring related skills of social attention. Furthermore, performance is related even approximately six years later. Therfeore, the GeoPref test appears to be capturing a relevant and relatively stable characterisitic of dysregulated social attention that can help inform prognosis and later joint attention eye gaze. Additionally, the correlation between ADOS Total Scores at intake and ADOS Total Scores at school age was comapred to the correlation between GeoPref scores at intake and ADOS Total Scores at school age. As expected, there was a correlation between ADOS Total Scores at both timepoints. Scores on the GeoPref Test at intake were also correlated with ADOS Total scores at school age, further highlighting the prognostic utility of an eye tracking test such as the GeoPref Test.

Limitations of the current study include the preliminary analysis from a small sample size. The analyses presented were exploratory in nature and a series of correlations present preliminary information for evaluating the utility of eye tracking as a prognostic tool. A limited number of variables including age at intake and amount of autism-focused services received were also explored as additional factors that may influence outcome. However, there are many other factors that would be beneficial to explore within a larger sample, including a range of demographic varaibles and family medical history. With a larger sample, categorical analyses comparing children with extreme geo preference and extreme social preference would be beneficial to explore potential subgroup differences in clinical outcomes as well. Additionally, although all children were initially recruited from a general population sample, families that opted to return for a follow-up at a later age could represent a specific selection of that population that were more apt to participate in research.

Although preliminary, the present evidence is consistent with our hypothesis that children with a preference for non-social geometric stimuli would demonstrate across ages more impairment related to ASD core symptoms than those with a social preference. We believe this is due to the fact that children with the geometric preference show an unusual preference for nonsocial stimuli and are failing to attend to important social information in their environment, which may negatively impact opportunities to learn a variety of skills. However, preference for non-social stimuli was not related to more impairment across all cognitive and language domains. Heterogeneity in ASD exists at all levels including etiology (Jeste & Geschwind, 2014), symptom profile (Masi, DeMayo, Glozier, & Guastella, 2017), developmental timing (Courchesne et al., 2019),and neural functional profiles (Hahamy, Behrmann, & Malach, 2015; Lombardo et al., 2015). Although this study was an important first step in examining the utility of eye tracking as a prognostic tool, given this heterogeneity, it is not surprising that a single eye tracking test could not predict outcome at every symptom level. Futhermore, it has been suggested that visual attention as measured by eye tracking may be under genetic influence (Constantino et al., 2017). For example, in a large eye-tracking study that examined visual attention in twin and non-twin toddlers, there was significantly higher concordance in preferential visual attention patterns between monozygotic twins than dizygotic twins (Constantino et al., 2017). This finding suggests visual attention may be influenced by genetics. Therefore the differences measured by eye tracking tests may be influenced by specific genetic etiologies of ASD. In a companion paper using resting state fMRI methodology (Lombardo et al., in review), we discovered that ASD toddlers with a pronounced preference for attending to the non-social geometric images in the GeoPref Test, have a pronounced neural functional disconnection between visual and attention networks and social brain networks (eg., the default mode network). This dysregulation of neurofunctional communication between the higher-order social brain and lower-order visual processing systems, has long been theorized as part of the explanation for why visual orienting and attention in some children with ASD can sometimes appear to be strongly dominated and directed by low-level visual percepts at the expense of social (Courchesne & Pierce, 2005). That the present study finds this early-age pronounced preference for non-social visual stimuli predicts later-age persistently higher ASD symptoms, raises the hypothesis that, in this subtype of ASD, the underlying neurofunctional disconnection between the social and visual brain at early-ages either may persist across childhood in driving atypical social attention and behavior or may have biased daily experiences during a critical early period of social development leading towards diminished social learning. Since brain functioning strongly influences what is experienced as well as is influenced by experience, dysregulation of both may conspire to alter what a child pays attention to and how this shapes the child’s social brain development. Therefore, there may be a complex relationship between genetics, brain activity, and experiences that shape heterogeneity in ASD. Future research will likely need to explore the relationships between all of these domains to best identify prognostic markers in ASD.

Figure 3.

Figure 3.

Open in a new tab

Correlation between GeoPref scores at young ages (mean age 23 months) and ADOS Total Scores at school age (mean age 8 years). Children with an extreme preference for geometric images are depicted in red, children with an extreme preference for social images are depicted in blue, and children in the mid-range are depicted in grey. Children who showed preference for geometric images at young ages had higher scores on the ADOS, indicating more impairment, at school age.

Acknowledgements

This work was supported by a Thrasher Early Career Award (E. Bacon), NIH R01-MH080134 (K. Pierce), P50-MH081755 (E. Courchesne), NFAR grant (K. Pierce), and NIMH R01-MH036840 (E. Courchesne). A special thank you to Eric Courchesne for his support of this project. We sincerely thank the children and families who participated in this research.

Footnotes

Conflicts of Interest: The GeoPref Test is licensed by the University of California, San Diego and Karen Pierce receives royalties. The other authors have no conflicts of interest or financial disclosures relevant to this article. The funding agency had no role in the conception or writing of this manuscript.

References

  1. Anderson DK, Liang JW, & Lord C (2014). Predicting young adult outcome among more and less cognitively able individuals with autism spectrum disorders. J Child Psychol Psychiatry, 55(5), 485–494. doi: 10.1111/jcpp.12178 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bacon EC, Courchesne E, Barnes CC, Cha D, Pence S, Schreibman L, … Pierce K (2017). Rethinking the idea of late autism spectrum disorder onset. Development and Psychopathology, 1–17. doi: 10.1017/S0954579417001067 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bacon EC, Dufek S, Schreibman L, Stahmer AC, Pierce K, & Courchesne E (2014). Measuring outcome in an early intervention program for toddlers with autism spectrum disorder: use of a curriculum-based assessment. Autism Res Treat, 2014, 964704. doi: 10.1155/2014/964704 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Benasich AA, Choudhury NA, Realpe-Bonilla T, & Roesler CP (2014). Plasticity in developing brain: active auditory exposure impacts prelinguistic acoustic mapping. J Neurosci, 34(40), 13349–13363. doi: 10.1523/JNEUROSCI.0972-14.2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bradshaw J, Steiner AM, Gengoux G, & Koegel LK (2015). Feasibility and effectiveness of very early intervention for infants at-risk for autism spectrum disorder: a systematic review. J Autism Dev Disord, 45(3), 778–794. doi: 10.1007/s10803-014-2235-2 [DOI] [PubMed] [Google Scholar]
  6. Butz M, Worgotter F, & van Ooyen A (2009). Activity-dependent structural plasticity. Brain Res Rev, 60(2), 287–305. doi: 10.1016/j.brainresrev.2008.12.023 [DOI] [PubMed] [Google Scholar]
  7. Chlebowski C, Robins DL, Barton ML, & Fein D (2013). Large-scale use of the modified checklist for autism in low-risk toddlers. Pediatrics, 131(4), e1121–1127. doi: 10.1542/peds.2012-1525 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Constantino JN (2012). Social Responsiveness Scale, Second Edition: Western Psychological Services. [Google Scholar]
  9. Constantino JN, Kennon-McGill S, Weichselbaum C, Marrus N, Haider A, Glowinski AL, … Jones W (2017). Infant viewing of social scenes is under genetic control and is atypical in autism. Nature, 547(7663), 340–344. doi: 10.1038/nature22999 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Courchesne E, & Pierce K (2005). Why the frontal cortex in autism might be talking only to itself: local overconnectivity but long-distance disconnection. Curr Opin Neurobiol, 15(2), 225–230. doi: 10.1016/j.conb.2005.03.001 [DOI] [PubMed] [Google Scholar]
  11. Courchesne E, Pramparo T, Gazestani VH, Lombardo MV, Pierce K, & Lewis NE (2019). The ASD Living Biology: from cell proliferation to clinical phenotype. Mol Psychiatry, 24(1), 88–107. doi: 10.1038/s41380-018-0056-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dawson G (2008). Early behavioral intervention, brain plasticity, and the prevention of autism spectrum disorder. Dev Psychopathol, 20(3), 775–803. doi: 10.1017/S0954579408000370 [DOI] [PubMed] [Google Scholar]
  13. Falck-Ytter T, Nystrom P, Gredeback G, Gliga T, & Bolte S (2018). Reduced orienting to audiovisual synchrony in infancy predicts autism diagnosis at 3 years of age. J Child Psychol Psychiatry doi: 10.1111/jcpp.12863 [DOI] [PubMed] [Google Scholar]
  14. Gliga T, Bedford R, Charman T, Johnson MH, & Team B (2015). Enhanced Visual Search in Infancy Predicts Emerging Autism Symptoms. Curr Biol, 25(13), 1727–1730. doi: 10.1016/j.cub.2015.05.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Guerreiro MJ, Putzar L, & Roder B (2016). The Effect of Early Visual Deprivation on the Neural Bases of Auditory Processing. J Neurosci, 36(5), 1620–1630. doi: 10.1523/JNEUROSCI.2559-15.2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hahamy A, Behrmann M, & Malach R (2015). The idiosyncratic brain: distortion of spontaneous connectivity patterns in autism spectrum disorder. Nat Neurosci, 18(2), 302–309. doi: 10.1038/nn.3919 [DOI] [PubMed] [Google Scholar]
  17. Helt M, Kelley E, Kinsbourne M, Pandey J, Boorstein H, Herbert M, & Fein D (2008). Can children with autism recover? If so, how? Neuropsychol Rev, 18(4), 339–366. doi: 10.1007/s11065-008-9075-9 [DOI] [PubMed] [Google Scholar]
  18. Howlin P, Magiati I, & Charman T (2009). Systematic review of early intensive behavioral interventions for children with autism. Am J Intellect Dev Disabil, 114(1), 23–41. doi: 10.1352/2009.114:23;nd41 [DOI] [PubMed] [Google Scholar]
  19. Hus V, Bishop S, Gotham K, Huerta M, & Lord C (2013). Factors influencing scores on the social responsiveness scale. J Child Psychol Psychiatry, 54(2), 216–224. doi: 10.1111/j.1469-7610.2012.02589.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Jeste SS, & Geschwind DH (2014). Disentangling the heterogeneity of autism spectrum disorder through genetic findings. Nat Rev Neurol, 10(2), 74–81. doi: 10.1038/nrneurol.2013.278 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Kasari C, Gulsrud A, Freeman S, Paparella T, & Hellemann G (2012). Longitudinal follow-up of children with autism receiving targeted interventions on joint attention and play. J Am Acad Child Adolesc Psychiatry, 51(5), 487–495. doi: 10.1016/j.jaac.2012.02.019 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kasari C, Paparella T, Freeman S, & Jahromi LB (2008). Language outcome in autism: randomized comparison of joint attention and play interventions. J Consult Clin Psychol, 76(1), 125–137. doi: 10.1037/0022-006X.76.1.125 [DOI] [PubMed] [Google Scholar]
  23. Klin A, Jones W, Schultz R, Volkmar F, & Cohen D (2002). Visual fixation patterns during viewing of naturalistic social situations as predictors of social competence in individuals with autism. Arch Gen Psychiatry, 59(9), 809–816. [DOI] [PubMed] [Google Scholar]
  24. Lombardo MV, Pierce K, Eyler LT, Carter Barnes C, Ahrens-Barbeau C, Solso S, … Courchesne E (2015). Different functional neural substrates for good and poor language outcome in autism. Neuron, 86(2), 567–577. doi: 10.1016/j.neuron.2015.03.023 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lord C, Rutter M, DiLavore P, Risi S, Gotham K, & Bishop S (2012). Autism Diagnostic Observation Schedule–2nd edition (ADOS-2) Los Angeles, CA: Western Psychological Corporation. [Google Scholar]
  26. Lord C, Rutter M, DiLavore PC, & Risi S (2002). Autism Diagnostic Observation Schedule (Western Psychological Services, Los Angeles: ). [Google Scholar]
  27. Loth E, Charman T, Mason L, Tillmann J, Jones EJH, Wooldridge C, … Buitelaar JK (2017). The EUAIMS Longitudinal European Autism Project (LEAP): design and methodologies to identify and validate stratification biomarkers for autism spectrum disorders. Mol Autism, 8, 24. doi: 10.1186/s13229-017-0146-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Magiati I, Tay XW, & Howlin P (2012). Early comprehensive behaviorally based interventions for children with autism spectrum disorders: a summary of findings from recent reviews and meta-analyses. Neuropsychiatry, 2(6), 543–570. [Google Scholar]
  29. Makrygianni MK, & Reed P (2010). A meta-analytic review of the effectiveness of behavioural early intervention programs for children with Autistic Spectrum Disorders. Research in Autism Spectrum Disorders, 4, 577–593. doi: doi: 10.1016/j.rasd.2010.01.014 [DOI] [Google Scholar]
  30. Masi A, DeMayo MM, Glozier N, & Guastella AJ (2017). An Overview of Autism Spectrum Disorder, Heterogeneity and Treatment Options. Neurosci Bull, 33(2), 183–193. doi: 10.1007/s12264-017-0100-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Moore A, Wozniak M, Yousef A, Barnes CC, Cha D, Courchesne E, & Pierce K (2018). The geometric preference subtype in ASD: identifying a consistent, early-emerging phenomenon through eye tracking. Mol Autism, 9, 19. doi: 10.1186/s13229-018-0202-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Mullen EM (1995). Mullen scales of early learning: AGS Circle Pines, MN. [Google Scholar]
  33. Murias M, Major S, Davlantis K, Franz L, Harris A, Rardin B, … Dawson G (2018). Validation of eye-tracking measures of social attention as a potential biomarker for autism clinical trials. Autism Res, 11(1), 166–174. doi: 10.1002/aur.1894 [DOI] [PubMed] [Google Scholar]
  34. Paparella T, & Freeman SFN (2015). Methods to improve joint attention in young children with autism: a review. Pediatric Health Med Ther, 6, 65–78. doi: 10.2147/PHMT.S41921 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Paul R, Campbell D, Gilbert K, & Tsiouri I (2013). Comparing spoken language treatments for minimally verbal preschoolers with autism spectrum disorders. J Autism Dev Disord, 43(2), 418–431. doi: 10.1007/s10803-012-1583-z [DOI] [PubMed] [Google Scholar]
  36. Pierce K, Carter C, Weinfeld M, Desmond J, Hazin R, Bjork R, & Gallagher N (2011). Detecting, studying, and treating autism early: the one-year well-baby check-up approach. Journal of Pediatrics, 159(3), 458–465 e451–456. doi: 10.1016/j.jpeds.2011.02.036 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Pierce K, Conant D, Hazin R, Stoner R, & Desmond J (2011). Preference for geometric patterns early in life as a risk factor for autism. Arch Gen Psychiatry, 68(1), 101–109. doi: 10.1001/archgenpsychiatry.2010.113 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Pierce K, Marinero S, Hazin R, McKenna B, Barnes CC, & Malige A (2016). Eye Tracking Reveals Abnormal Visual Preference for Geometric Images as an Early Biomarker of an Autism Spectrum Disorder Subtype Associated With Increased Symptom Severity. Biol Psychiatry, 79(8), 657–666. doi: 10.1016/j.biopsych.2015.03.032 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Reichow B, & Wolery M (2009). Comprehensive synthesis of early intensive behavioral intervnetion for young children with autism based on the UCLA young autism project model. Jounral of Autism and Developmental Disorders, 39, 23–41. doi: 10.1007/s10803-008-0596-0 [DOI] [PubMed] [Google Scholar]
  40. Robins DL, Casagrande K, Barton M, Chen CM, Dumont-Mathieu T, & Fein D (2014). Validation of the modified checklist for Autism in toddlers, revised with follow-up (M-CHAT-R/F). Pediatrics, 133(1), 37–45. doi: 10.1542/peds.2013-1813 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Solomon M, Iosif AM, Reinhardt VP, Libero LE, Nordahl CW, Ozonoff S, … Amaral DG (2018). What will my child’s future hold? phenotypes of intellectual development in 2–8-year-olds with autism spectrum disorder. Autism Res, 11(1), 121–132. doi: 10.1002/aur.1884 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Sparrow SS, Cicchetti DV, & Balla DA (2005). Vineland Adaptive Behavior Scales, Second Edition. Bloomington, MN: Pearson. [Google Scholar]
  43. Warren Z, McPheeters ML, Sathe N, Foss-Feig JH, Glasser A, & Veenstra-Vanderweele J (2011). A systematic review of early intensive intervention for autism spectrum disorders. Pediatrics, 127(5), e1303–1311. doi: 10.1542/peds.2011-0426 [DOI] [PubMed] [Google Scholar]
  44. Wechsler D (2014). Wechsler Intelligence Scales for Children, Fifth Edition: Pearson. [Google Scholar]
  45. Wetherby AM, Brosnan-Maddox S, Peace V, & Newton L (2008). Validation of the Infant-Toddler Checklist as a broadband screener for autism spectrum disorders from 9 to 24 months of age. Autism, 12(5), 487–511. doi: 10.1177/1362361308094501 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Zwaigenbaum L, Bauman ML, Choueiri R, Fein D, Kasari C, Pierce K, … Wetherby A (2015). Early Identification and Interventions for Autism Spectrum Disorder: Executive Summary. Pediatrics, 136 Suppl 1, S1–9. doi: 10.1542/peds.2014-3667B [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES