Identifying Early Markers of Autism Spectrum Disorder Based on Patterns of Medical Symptoms and Healthcare Service

摘要 自闭症谱系障碍(ASD)是一种复杂的大脑疾病，以言语和非语言障碍为特征 言语社交和重复行为模式。在美国出生的大约400万名婴儿中 美国每年约有6万人被诊断出患有自闭症，根据最近的疾病预防控制中心的数据，大约每59名儿童中就有1人被诊断出患有自闭症 估计一下。早在12个月大的时候就提供的干预措施已经被证明是有效的，而且 诊断是ASD干预成功的关键。然而，在确定身份方面进展甚微 自闭症的高危儿童在很小的时候。虽然普遍筛查可以提高早期发现 ASD，各种障碍阻碍了它的广泛采用。 数十年的深入研究不仅确定了行为ASD标记物，还揭示了 许多其他风险因素，如遗传变异、家族和兄弟姐妹的病史、大脑异常、 低出生体重，以及父亲和母亲分娩时的年龄。此外，没有确凿证据表明 一些医疗条件，如中耳炎、感染、癫痫、胃肠道问题、分娩 并发症以及发育和生理里程碑的延迟可能与ASD有关，但 可能早在ASD的标志性行为症状出现之前就表现出来。尽管事实是单独的 这些医学症状可能不够敏感，不足以作为ASD诊断的可行标记物， 结合在一起，它们有望在任何疾病发生之前就确定儿童患自闭症的风险 现有的ASD筛查工具目前能够。消除延迟诊断将允许 通过早期干预优化结果。据我们所知，几乎没有 研究以收获这些积累的知识。通过利用大型、全国性的、纵向的、私人的 保险医疗索赔数据库(MarketScan®)和医疗补助索赔数据库(Medicaid Analytic Extral 或MAX)，我们将全面调查某些医疗症状和医疗保健的集体作用 服务使用模式作为预测ASD风险的早期标记。 如果成功，这项研究将展示一种改进ASD风险预测的新方法，在此基础上，我们 可以构建一个基于医疗索赔的ASD监测系统。在后台工作，这样一个系统 可以筛选大量的儿童电子医疗索赔记录，寻找模式 指出潜在风险并识别儿童，以便在他们的ASD风险出现时进行进一步面对面评估 跨过了一个关键的门槛。这最终将促进ASD的早期发现，从而最终改善 早期干预疗法的影响。

ABSTRACT Autism Spectrum Disorder (ASD) is a complex brain disorder marked by difficulties in verbal and non- verbal social interactions and patterns of repetitive behaviors. Of the approximately 4 million babies born in the U.S. each year, about 60,000 will be diagnosed with ASD, or about 1 in 59 children based on the recent CDC estimate. Interventions delivered as early as 12 months of age have been shown to be effective, and early diagnosis is critical to the success of ASD interventions. However, there has been little progress on identifying children at high risk for ASD at an early age. Although universal screening could improve the early detection of ASD, various barriers have kept it from being widely adopted. Decades of in-depth research have not only identified behavioral ASD markers but have also shed light on many other risk factors, such as genetic variants, family and siblings’ medical history, brain abnormalities, low birth weight, and paternal and maternal ages at childbirth. In addition, there is inconclusive evidence that some medical conditions, such as otitis media, infections, epilepsy, gastrointestinal problems, birth complications, and delay in developmental and physiological milestones, may be associated with ASD, but may manifest well before the onset of hallmark behavioral symptoms of ASD. Despite the fact that individually these medical symptoms may not be sensitive enough to be used as a viable marker for ASD diagnosis, combined together, they hold the promise of robustly determining children’s risks for ASD well before any existing ASD screening tool is currently capable of. The elimination of delayed diagnosis would allow for optimization of outcomes through early intervention. To the best of our knowledge, there has been little research to harvest this accumulated knowledge. By leveraging a large, national, longitudinal, private insurance medical claims database (MarketScan®) and Medicaid claims database (Medicaid Analytic eXtract or MAX), we will comprehensively investigate the collective role of certain medical symptoms and healthcare service use patterns as early markers for predicting ASD risk. If successful, this study will demonstrate a novel way of improving ASD risk prediction, upon which we can construct a medical claims-based ASD surveillance system. Working in the background, such a system can sift through an extensive volume of children’s electronic medical claims records, looking for patterns indicating potential risk and identifying children for further in-person evaluations when their ASD risk has crossed a critical threshold. This would ultimately advance ASD early detection and thus ultimately improve the impact of early intervention therapies.