Identification and characterization of children with asthma-associated comorbidities through computational and immune phenotyping

点击翻译按钮获取中文摘要

PROJECT SUMMARY Asthma, the most common chronic disease among children, is one of the five most burdensome diseases in the US. While current care and research efforts focus on symptom control and exacerbation risk, children with asthma may suffer from infectious and inflammatory diseases, ie, asthma-associated infectious and inflammatory disease comorbidities (AIICs) (eg, pneumococcal disease, herpes zoster, appendicitis, and celiac disease). Although AIICs pose serious threats to children with asthma, they are largely under-recognized, as evidenced by diabetes mellitus being widely recognized but a less common chronic illness with a magnitude similar to that for asthma. Presently, the mechanisms underlying AIICs are unknown. We postulate immunosenescence might be related, as AIICs coincide with cardinal immunosenescence features. AIICs are not clinically defined and a suitable tool has not been developed to identify children with AIICs. Thus, no strategies mitigating AIICs risks and outcomes exist. Addressing these knowledge gaps depends upon two key questions: (1) “How can asthmatic children subgroups with increased AIICs risk be identified at a population level using electronic medical records?” and (2) “What immune parameters characterize such children?” Answering these is this proposal's primary goal. To this end, our current R01 study successfully developed, validated, and implemented natural language processing (NLP)- empowered computational phenotyping algorithms for two existing criteria for childhood asthma (Predetermined Asthma Criteria, PAC and Asthma Predictive Index, API). NLP-empowered algorithm application to the 1997-2007 Olmsted County Birth Cohort (OCBC) enabled us to profile a subgroup of children with asthma at increased AIICs risk, disproportionately represented by children who met both NLP-PAC and NLP-API. Our new pilot data suggest asthma potentially accelerates immunosenescence leading to AIICs in a subgroup of asthmatic children. In this renewal proposal, we target this subgroup who meet both NLP-PAC and NLP-API. We will develop and apply NLP-empowered computational phenotyping algorithms for AIICs to identify such children at a population level, then characterize their immune parameters measuring immunosenescence. In Aim 1, we will develop new NLP-empowered computational phenotyping algorithms for recognized and unrecognized AIICs (NLP-AIIC) for children enrolled in Mayo Clinic `s 1997-2016 OCBC, then assess portability of NLP-AIIC at Sanford Children's Hospital, Sioux Falls, SD . In Aim 2, we will identify and characterize children with AIICs through new NLP-AIIC and NLP algorithms for asthma status, by utilizing clinical and immune parameters to measure immunosenescence. In Aim 3, we will assess changes (eg, waning adaptive immunity) over time in immune parameters measuring immunosenescence for 300 children in our R01 study, re-enrolling them for further characterization. This proposed study is indispensable to understanding why some children with asthma develop AIICs, while others do not. This knowledge will allow us to identify, manage, mitigate, and improve outcomes for AIICs among children.