Randomized Controlled Trial of a Six-Month Mindfulness-Based Intervention for Type 2 Diabetes

摘要 新发糖尿病和严重的糖尿病酮症酸中毒(DKA)在感染的患者中经常被报告 使用新冠肺炎，即使没有已知的糖尿病病史。然而，新冠肺炎在很大程度上 病毒感染会触发或加速高血糖、糖尿病和DKA的发展，目前尚不清楚。 因此，迫切需要对糖尿病发病率和严重程度进行流行病学研究。 以及它与不同患者群体中新冠肺炎和SARS-CoV-2的潜在关联。整体而言 本研究的目的是确定新发糖尿病和DKA的发生率、严重程度和危险因素 在新冠肺炎感染患者中。我们计划使用基于人工智能(AI)的技术，包括 用于新冠肺炎和糖尿病危险因素研究的新型可解释机器学习预测模型 大型、多样化和多分辨率数据集，包括TriNetX Research中未识别的患者数据 多个医疗保健组织(HCO)的网络，以及新冠肺炎的全球CoviDIAB注册表 相关的糖尿病。我们将调查糖尿病危险因素的预测价值，如频率和 新冠肺炎感染的严重程度、年龄、性别、种族、体重指数、既往健康状况、家庭 糖尿病病史、糖皮质激素的使用和健康的社会决定因素(SDOH)。具体目标1： 确定(A)新发糖尿病的发病率，(B)DKA的发病率，(C)高血糖的严重程度和 (D)糖尿病发病时间，以及(E)新发糖尿病的危险因素。 新冠肺炎，与两个对照组进行比较：(1)非新冠肺炎确诊为流感患者，和(2) 非新冠肺炎非流感患者。具体目标2：应用基于人工智能的新技术 由患者数据库的可解释机器学习模型组成，TriNetX研究网络和 新冠肺炎相关糖尿病全球CoviDIAB注册表，以预测(A)新发糖尿病的发展， (B)DKA和(C)新冠肺炎感染后的严重DKA，与两个对照组相比，定义见 目标1.具体目标3：利用拟议研究的结果制定赠款提案，为 关于生活方式和/或药理学的前瞻性多中心随机对照试验(RCT)的设计 对新冠肺炎相关新发糖尿病高危患者的干预，通过EHR从 多种医疗保健系统，服务于不同的患者群体。我们的研究结果和预测性ML模型 将提供证据，说明哪些网站可能是捕获全国患者代表的最佳地点，变量 不同年龄组、地区和既往疾病的兴趣、临床结果和样本量。我们 设想这项研究将导致一项研究新冠肺炎新发糖尿病的多中心临床试验，即 它的设计非常成功，因为它将基于来自广泛和多样化的美国人口样本的数据。

ABSTRACT New-onset diabetes and severe diabetic ketoacidosis (DKA) have frequently been reported in patients infected with COVID-19, even in the absence of a known history of diabetes. However, the extent to which COVID-19 viral infection triggers or accelerates the development of hyperglycemia, diabetes and DKA remains unclear. Therefore, there is an urgent public health need for epidemiologic studies of diabetes incidence and severity and its potential association with COVID-19 and SARS-CoV-2 in diverse patient populations. The overall objective of this study is to determine the incidence, severity, and risk factors for new-onset diabetes and DKA in patients with COVID-19 infection. We plan to use Artificial Intelligence (AI)-based technology consisting of novel interpretable machine learning predictive models to study COVID-19 and risk factors for diabetes in large, diverse and multi-resolution datasets, including de-identified patient data from the TriNetX Research Network of multiple health care organizations (HCOs), and from the global CoviDIAB Registry of COVID-19 related diabetes. We will investigate the predictive value of diabetes risk factors such as frequency and severity of COVID-19 infection, age, sex, race, body mass index (BMI), pre-existing health conditions, family history of diabetes, use of glucocorticoids, and social determinants of health (SDOH). Specific Aim 1: Determine (a) the incidence of new onset diabetes, (b) incidence of DKA, (c) severity of hyperglycemia and DKA at onset, (d) timing of diabetes onset, and (e) risk factors for new onset diabetes among patients with COVID-19, as compared to two control groups: (1) Non-COVID-19 patients diagnosed with influenza, and (2) Non-COVID-19 patients without influenza. Specific Aim 2: Apply novel Artificial Intelligence-based technology consisting of interpretable machine learning models to patient databases, TriNetX Research Network and the global CoviDIAB Registry of COVID-19 related diabetes, to predict the development of (a) new onset diabetes, (b) DKA, and (c) severe DKA following COVID-19 infection, compared to the two control groups, as defined in Aim 1. Specific Aim 3: Develop a grant proposal by using the results from the proposed study to inform the design of a future prospective multicenter randomized controlled trial (RCT) of a lifestyle and/or pharmacologic intervention for patients at high risk for new onset diabetes related to COVID-19, identified via EHRs from multiple healthcare systems serving diverse patient populations. Our study findings and predictive ML models will provide evidence on what may be the best sites to capture national patient representation, variables of interest, clinical outcomes, and sample size for different age groups, regions, and pre-existing conditions. We envision this study will lead to a multicenter clinical trial to study new onset diabetes in COVID-19, that will be highly successful as its design will be based on data from a broad and diverse sample of the U.S. population.