Early Identification of ASD: Translating Eye Tracking into Practice

描述(由申请人提供)：目前，父母报告筛查工具(如M-Chat、CSBS)是儿科医生可用于早期发现自闭症的唯一标准化筛查选择。正如我们以前的工作(1年良好婴儿体检方法，Piells等人，2011年)所表明的那样，这种筛查具有宝贵的优势，但也有弱点，包括父母的特征对他们如何评估孩子的行为和高假阳性率的影响。自闭症是一种疾病，其根源是早期大脑发育异常，但早期识别在很大程度上取决于父母报告的衡量标准，而不是更客观、可量化的行为领域。要超越父母报告作为唯一的早期筛查选择，必须研究新的筛查程序。在我们上一个资助周期中，我们开发了一种新的眼球跟踪测试，即针对自闭症的几何偏好测试(GeoPref测试)。这个简单的一分钟测试在计算机显示器的一半上显示了动态的、五颜六色的几何运动图案，在另一半上显示了五颜六色的活跃的人。使用标准的眼动跟踪技术对440名幼儿进行测试，结果显示，几乎每个高度专注于几何图案的幼儿都患有自闭症，而不是典型的自闭症或发育迟缓。因此，我们的实验室实验证明，GeoPref检测ASD的特异度高达99%。在筛查工具中，低的假阳性率是必不可少的，因为假阳性使系统不堪重负，并造成焦虑。在这种异质性障碍中，同样重要的是，1分钟的测试发现了37%的自闭症幼儿。这是一个比任何其他早期遗传学、蛋白质组、神经成像或神经行为筛查更大的百分比，而且它快速、容易，并且高度ASD特异性。目标1将确定将ASD特异性测试整合到临床实践中的方法，作为未通过常规第一层铅笔和纸张筛查的婴儿的第二层筛查(即， CSBS)。结果将大大提高早期发现的准确性和转介婴儿进行诊断和治疗服务的速度。使用便携式眼球追踪器，39名儿科医生将在他们的办公室为未通过CSBS的幼儿进行GeoPref测试，这些儿童将被转介 我们的盲诊中心。这将是第一次将ASD的眼球跟踪发现转化为现实世界的临床实践。虽然GeoPref测试识别出37%的ASD学步儿童，但同样需要快速和准确的测试来识别其余的ASD学步儿童。Aim 3建议开发新的眼球跟踪测试，特别是新的社交定向和母亲范型，以检测这些蹒跚学步的孩子。随机森林分类算法将识别最能识别ASD幼儿亚群的眼睛凝视特征。所有蹒跚学步的孩子都将在30-36个月时接受最终的盲诊。

DESCRIPTION (provided by applicant): Currently, parent report screening tools (e.g., M-CHAT, CSBS) are the only standardized screening options available to pediatricians for the early detection of autism. As demonstrated in our previous work (the 1-Year Well-Baby Check-Up Approach, Pierce et al., 2011), such screens have valuable strengths, but also weaknesses, including the impact of parent characteristics on how they rate their child's behavior and high false positive rates. Autism is a disorder that has its roots in abnormal early brain development, yet early identification rests largely in the hands of parent report measures, rather than in the domain of more objective, quantifiable behavior. To move beyond parent report as the only early screening choice, novel procedures for screening must be researched. During our last grant cycle we developed a novel eye-tracking test, the Geometric Preference Test for Autism (GeoPref Test). This simple one-minute test shows dynamic, colorful geometric moving patterns on half of a computer monitor and colorful active people on the other half. Of >440 toddlers tested using standard eye tracking technology, almost every toddler who fixated at high rates on the geometric patterns was ASD, and not typical or developmentally delayed. Thus, our laboratory experiments demonstrated exceptionally high, 99%, specificity of the GeoPref Test for detecting ASD. A low false positive rate is essential in a screening tool because false positives overload the system and create anxiety. Equally important in this heterogeneous disorder, the 1-min test identified 37% of all ASD toddlers. This is a larger percentage than any other early genetic, proteomic, neuroimaging or neurobehavioral screen, and it is fast, easy, and highly ASD specific. Aim 1 will identify methods to integrate ASD specific test into clinical practice as a 2nd tier screen for babies who fail a routine 1st tier pencil and paper screen (i.e., the CSBS). Results will greatly improve the accuracy of early detection and speed referral of babies for diagnostic and treatment services. Using a portable eye tracker, 39 pediatricians will give the GeoPref Test in their office to toddlers who fail the CSBS who will in turn be referred to our Center for blinded diagnosis. This will be the first translation of an eye tracking finding on ASD into real-world clinical practice. In an effort to fully understand the clinical phenotype of ASD toddlers that "fail" the GeoPref Test, Aim 2 will use novel experimental tests, such as a test of exploration, as well as standardized tests such as the Mullen, to identify clinical profiles tha distinguish these toddlers from others with an ASD. Although the GeoPref Test identifies a large subgroup of 37% of all ASD toddlers, equally fast, and accurate tests are needed to identify the remaining ASD toddlers. Aim 3 proposes to develop new eye tracking tests, specifically novel social orienting and motherese paradigms, to detect these toddlers. Random forest classification algorithms will identify eye gaze signatures that best identify subgroups of ASD toddlers. All toddlers will receive a final blinded diagnosis at 30-36 months.