Informatics Core

信息学核心（IC）将翻译UCSF收集的慢性下背痛（cLBP）患者数据 REACH进入临床智能的转化模型，改善cLBP的结果。把加州大学旧金山分校的 在这个首屈一指的基础设施上拥有丰富的临床经验，可以根据EHR数据进行人工智能研究， 信息中心有潜力开展创新性的合作研究，显著改善cLBP的结果 对患者IC的首要目标是与其他UCSF REACH研究和临床合作 核心开发用于解释成像和其他电子健康记录（EHR）数据的新型AI方法 显著改善cLBP护理。为了充分利用UCSF REACH收集的数据，IC 将有助于统计和机器学习方法的开发和临床应用， 具体目标。在目的1中，为了基于cLBP表型鉴定患者亚组，我们将使用 让其他核心使用现有的策划EHR和新收集的cLBP患者数据， 了解cLBP疾病的途径。为了分析治疗目标之间的权衡和协同作用，IC 将应用典型相关分析，以发现临床见解之间的潜在协同作用和权衡 开发机器学习的各种治疗结果（疼痛、身体和心理社会残疾指标） 用于预测患者特异性治疗反应的方法。最后，IC将应用最先进的卷积 神经网络技术在预测医学成像的临床相关结果，以优化这一点， 为患者提供临床见解和个性化治疗计划。

The Informatics Core (IC) will translate patient data on chronic lower back pain (cLBP) collected by UCSF REACH into translational models of clinical intelligence that improve cLBP outcomes. By coupling UCSF's extensive clinical experience with this second-to-none infrastructure to perform AI research based on EHR data, the IC has the potential to produce innovative collaborative research that significantly improves cLBP outcomes for patients. The overarching goal of IC is to work in symbiosis with other UCSF REACH Research and Clinical cores to develop novel AI methodologies for interpreting imaging and other Electronic Health Record (EHR) data to significantly improve cLBP care. In order to enable the full utilization of data collected by UCSF REACH, IC will aid in the development and clinical application of statistical and machine learning methods with the following specific aims. In in Aim 1, to identify patient subgroups based on cLBP phenotypes, we will work with the Reall other cores to use the existing curated EHR and newly collected data from cLBP patients to better understand cLBP disease pathways. To analyze trade-offs and synergies between treatment objectives, IC will apply canonical correlation analysis to discover clinical insights potential synergies and trade-offs between various treatment outcomes (pain, physical and psychosocial disability metrics) to develop machine learning methods for predicting patient-specific treatment response. Finally, IC will apply state-of-the art convolutional neural network techniques in predicting clinically relevant outcomes from medical imaging to optimize this clinical insight and personalize treatment plans for patients.