跳到正文
原文
Nature 健康护理主题·· 8 小时前精选AI 评分70

系统性红斑狼疮患者心血管风险因素及预防策略:一项综述

Social determinants of health, lifestyle factors, and artificial intelligence in cardiovascular risk stratification for systemic lupus erythematosus: a narrative review

AI 导读

文章综述了系统性红斑狼疮患者的心血管风险因素,包括社会经济因素、生活方式和人工智能在风险预测中的应用。研究指出,这些因素共同作用增加了心血管疾病的风险,强调了综合预防策略的重要性。

推荐理由

文章综合分析了系统性红斑狼疮患者的心血管风险因素及预防策略,为临床实践提供了重要参考。

正文 · AI 翻译

译文尚不完整,完整内容请切换到原文。

简介

系统性红斑狼疮(SLE)是一种慢性自身免疫性疾病,以多系统受累和疾病发作与缓解交替为特征。虽然SLE影响多个器官系统,包括皮肤、关节、肾脏和神经系统,但心血管疾病(CVD)已成为这一患者群体中发病率和死亡率的主要原因1,2,3,4。SLE患者患CVD的风险比同龄对照组高2到3倍5, 6。这种升高的心血管风险源于SLE固有的慢性系统性炎症,以及传统心血管风险因素的组合,包括高血压、血脂异常和糖尿病7,8,9,10。狼疮性肾炎(LN),由免疫介导的炎症和肾脏损伤引起的SLE的一种严重肾脏并发症,随着时间的推移可影响多达60%的SLE患者,通过与高血压、蛋白尿和代谢异常的关联进一步增加了心血管风险5。生活方式因素是另一个影响SLE患者心血管风险的关键领域。缺乏体力活动、吸烟、不良饮食习惯和过量饮酒可加剧系统性炎症,并可能促进动脉粥样硬化斑块的加速形成11,12,13,14。

新兴的人工智能(AI)和机器学习(ML)方法可以将基因组、临床和社会环境数据整合到针对不同SLE人群的风险模型中,为公平导向的心血管疾病预防提供了有希望的途径。本综述综合了当前关于SLE患者心血管疾病社会经济和生活方式决定因素的证据,探讨了影响脆弱人群的健康不平等现象,并探讨了AI在推进SLE患者精准心血管护理方面的新兴作用。据我们所知,先前的综述分别探讨了各个领域,包括SLE中的心血管风险因素3, 8, 9、健康的社会决定因素(SDOH)和SLE中的健康不平等现象15, 16以及AI/ML在SLE中的应用,包括新兴的心血管风险预测方法17,但这些综述没有在统一的、公平导向的框架内整合这些领域。通过将这些领域结合起来,本综述填补了文献中的一个重要空白,并将这些证据转化为临床可操作的、公平导向的指导方针,供管理SLE患者的临床医生参考。

方法

文献检索策略

我们对四电子数据库进行了全面检索:PubMed/MEDLINE、CINAHL、Scopus和Web of Science。我们将检索范围限制为2012年1月至2026年3月的出版物,以捕捉SLE患者中CVD风险的当代证据,并包括里程碑研究。当它们提供了基础证据时,我们包括了更早的开创性论文。我们限制搜索范围为英文出版物。

检索词

我们使用了医学主题词(MeSH)和自由文本术语开发了四个主题领域的搜索查询。领域 1(系统性红斑狼疮和心血管疾病):“系统性红斑狼疮” OR “SLE” AND “心血管疾病” OR “CVD” OR “动脉粥样硬化”。领域 2(社会决定因素和健康差异):“社会决定因素的健康” OR “社会经济地位” OR “健康差异” OR “结构性种族主义”。领域 3(生活方式):“体力活动” OR “锻炼” OR “烟草” OR “饮食模式” OR “肥胖”。领域 4(人工智能/机器学习):“人工智能” OR “机器学习” OR “预测模型” OR “风险分层”。我们根据需要使用布尔运算符组合领域术语。

纳入和排除标准

纳入研究需满足以下标准:(1) 入组≥18岁系统性红斑狼疮成人患者;(2) 报告心血管结局、CVD风险因素或CVD风险预测;(3) 至少涉及以下主题领域之一——社会决定因素、生活方式因素或基于人工智能/机器学习的心血管疾病风险预测;(4) 发表为原创研究文章、系统综述、荟萃分析或同行评审期刊中的临床实践指南。排除仅专注于儿科系统性红斑狼疮的研究、未报告心血管相关结局的研究以及缺乏原始数据的会议摘要、编辑评论或信件。数据库搜索共产生了约1,840条记录。去除重复项(约310条)并筛选标题和摘要后,我们审阅了120篇全文。最终有39项当代研究符合所有纳入标准并被纳入综合分析。此外,我们还通过参考追踪纳入了两篇2012年前的开创性出版物,当它们提供了基础证据时。

证据综合与质量评估

本综述按照《SANRA(叙述性综述文章评估量表)》框架进行并报告18。表11中的证据质量评级(高、中、低)反映了支持证据的数量、一致性和方法学严谨性:高=系统性红斑狼疮特定的随机对照试验或荟萃分析;中=观察性/队列数据;低=专家共识或推断。我们通过批判性评估研究设计、人群、结局指标以及研究结果的强度和一致性来进行综合分析。特别寻求了来自不同地理区域和多族群队列的研究,包括GLADEL和LUMINA,以确保公平地代表最常受系统性红斑狼疮心血管疾病差异影响的人群19, 20。

表1 系统性红斑狼疮的心血管疾病风险因素及基于证据的医疗保健提供者干预措施

全尺寸表格

社会决定因素的健康和系统性红斑狼疮的心血管疾病风险

社会经济地位和医疗保健可及性

社会经济因素与SLE患者的CVD风险相关。较低的收入、受教育程度有限以及食物不安全都会成为预防和疾病管理的障碍15, 27, 31。医疗费用负担仍然是一个挑战,可能降低患者对专科护理的依从性和可及性。来自较低社会经济背景的SLE患者比高社会经济背景的患者具有更高的心血管风险因素和不良结局发生率24。在Maynard等人分析的Hopkins队列中,多个与收入相关的关联因种族不同而有所不同,并不总是独立于疾病活动或治疗24。

医疗保健的可及性不仅包括保险覆盖,还包括专科护理、语言一致的提供者、文化敏感的团队和灵活的预约时间32。沟通障碍不仅与患者教育或健康素养有关,还反映了护理团队的责任,即使用通俗的语言,避免不必要的行话,并用文化响应的方式解释诊断、治疗和长期管理25。这需要医生、护士、教育者、翻译人员和患者导航员等多方面的协调努力,他们可以使用通俗的语言、视觉辅助工具和回顾性教学方法来支持患者的理解,并使SLE患者能够积极参与管理其心血管健康25。

种族和族裔差异

来自历史上种族和族裔少数群体的个体,特别是黑人(按AMA/NIH风格规范)和西班牙裔人群,承受着不成比例的SLE及其相关心血管并发症的负担。非洲裔美国人患SLE的患病率大约是白人的三倍。非洲裔美国人患SLE的风险比白人高14%的CVD风险3, 23。西班牙裔SLE患者同样面临更高的疾病严重程度和心血管并发症。进一步的研究表明,非洲裔和亚洲裔美国人可能比白人患者表现出更严重的疾病表现16, 23。

研究还显示了不同的治疗模式;非洲裔和西班牙裔患者在与欧洲裔患者相似的疾病严重程度下,接受的疾病修饰疗法不那么积极,专科转诊较少,心血管风险因素管理也不那么积极16, 20, 32, 33

全球范围内,系统性红斑狼疮(SLE)相关的心血管疾病(CVD)在拉丁美洲和全球南方地区的人群中更为普遍,这些地区存在经济和社会劣势、医疗保健基础设施有限以及种族多样性。GLADEL(拉丁美洲系统性红斑狼疮研究组)跨国队列研究了来自拉丁美洲国家的超过1,200名SLE患者的资料,记录了种族和社会经济差异在疾病表现、器官损伤和结果方面的异质性,其中混血(混有美洲原住民和欧洲血统的个体)和非裔拉丁美洲患者比白人患者经历了更为严重的疾病轨迹19。LUMINA(少数族裔系统性红斑狼疮:天性与后天)多族群美国队列研究显示,贫困、低教育水平和社会劣势独立地预测了西班牙裔、非裔和白人SLE患者疾病活动度随时间增加20。因此,在LUMINA中,社会经济劣势最好被视为导致SLE病情加重的因素,这可能后来会增加心血管风险,而不是直接证据表明心血管事件的发生。其他队列研究也提供了将社会经济劣势与SLE中的心血管事件和风险联系起来的证据3,15,23。这些研究共同支持了预防策略,这些策略旨在解决不同人群背景下的结构性和社会决定因素。

SLE患者的CVD风险的生活方式因素

体力活动和锻炼

体力活动在SLE患者中极为普遍,由于疾病相关的疲劳、关节痛、担心诱发发作以及心理社会因素如抑郁和焦虑11,13。然而,定期的体力活动可以带来显著的心血管益处,包括内皮功能的改善、血脂谱的改善、胰岛素敏感性的提高、血压控制和炎症标志物的改善11,13。

卢等人进行的一项系统综述和荟萃分析表明,SLE中的锻炼干预措施显著改善了有氧能力、肌肉力量、疲劳和健康相关的生活质量,而没有增加疾病活动度或不良事件11。有氧运动和阻力训练在适当开具和监控的情况下对大多数患者都是安全且有益的。国际共识指南现在支持为SLE患者开具个性化的锻炼处方13。医疗保健提供者应在每次临床就诊时评估体力活动水平,并在适当情况下考虑将患者转介至物理治疗进行监督性编程。

烟草使用

吸烟是SLE中最强大的可改变的心血管风险因素之一,但在SLE患者中的吸烟率往往高于一般人群14,21。吸烟与疾病活动度增加、更频繁的发作、加速的动脉粥样硬化、心血管事件风险增加以及免疫抑制药物疗效降低有关14,21。

雷蒙德等人报告称,SLE患者吸烟与B细胞激活因子(BAFF)水平升高和干扰素-γ水平降低有关,这是SLE发病机制中的关键免疫学异常21。基于证据的戒烟方法,包括行为咨询、尼古丁替代疗法和药物治疗(伐尼克兰、安非他酮),应优先作为所有吸烟患者心血管疾病预防的基本组成部分。

饮食模式

饮食模式显著影响SLE的心血管风险,通过影响脂质代谢、血糖稳态、体重、血压和系统性炎症。12, 34 地中海饮食模式,以高摄入水果、蔬菜、全谷物、豆类、坚果、橄榄油和鱼类为特征,已在一般人群中被广泛研究以获得心血管益处,并逐渐成为SLE的一种有前景的饮食方法。Pocovi-Gerardino等报告称,地中海饮食的更高依从性与SLE患者疾病活动度降低、血脂谱改善、炎症标志物减少以及心血管风险评分降低相关。34 饮食咨询应强调抗炎食物选择、足够的ω-3脂肪酸摄入、钠的限制以及维持健康体重的策略。建议转介给具有自身免疫疾病专长的注册营养师。

肥胖和体重

SLE患者中肥胖日益普遍,通过胰岛素抵抗、血脂异常、高血压、炎性脂肪细胞因子产生和机械性心血管压力增加心血管风险22。糖皮质激素治疗促进体重增加和中心性脂肪积累,进一步增加了代谢风险。综合体重管理,结合饮食调整、结构化锻炼和行为策略,已在SLE患者中证明了减重和心血管风险降低的有效性22。应无污名化地提供,承认影响体重的复杂生物学、心理和社会因素。

整合社会和生活方式决定因素:整体框架

SLE患者的心血管疾病不能通过还原论视角来理解。社会生态框架认识到,心血管健康源自生物脆弱性、健康行为、社会环境、卫生保健系统因素和更广泛结构性决定因素的动态相互作用27。教育程度影响健康素养和自我管理能力。社会经济资源决定了获得营养食物、安全的体育活动环境和高质量卫生保健的机会。种族和族裔身份影响了终身暴露于歧视、居住隔离、环境危害以及卫生保健系统中的不同待遇,为最脆弱的人群创造了累积的不利条件。将这些社会和生活方式决定因素纳入心血管风险预测模型中,可能在传统临床变量之外改善风险分层。

Shantz 等人的生物心理社会综合分析同样主张超越孤立的生物医学风险因素27。图. 1 总结了一个整合框架,在该框架中,AI 是潜在的分析层——而不是临床判断或结构性干预的替代品。医务人员可以使用经过验证的社会需求工具,包括响应和评估患者资产、风险和经历的协议(PRAPARE)和可问责健康社区健康相关社会需求筛查工具。筛查应在转诊途径、隐私保护和患者偏好得到解决时进行;单独进行筛查不足以改善SLE-CVD结果的证据是不足的。

图. 1

概念框架:系统性红斑狼疮(SLE)心血管风险的多级决定因素。上游决定因素 → 生活方式和护理 → 心血管结果 | AI 支持分析和护理;它不是因果决定因素。缩写:SLE,系统性红斑狼疮;AI,人工智能;CVD,心血管疾病。* AI 分析层中的工具是重新部署的研究工具。改进的模型区分度(例如,更高的受试者操作特征曲线面积 [AUC])是必要的,但不足以用于临床应用。改善患者结果需要前瞻性验证;对于审查的 SLE 特异性模型,这种证据尚不可用。注意。箭头表示上游决定因素通过生物途径对心血管结果的贡献的潜在路径,直接或通过生活方式和护理中介。对于观察性证据,方向箭头表示假设的关联而非确立的因果关系。

SLE 特异性因素与社会决定因素相互作用,通过生物途径增加心血管风险。高疾病活动度(SLEDAI)、累积器官损伤(SLICC 损伤指数)、狼疮性肾炎(影响约 40-60% 的不同 SLE 群体患者)、长期皮质类固醇暴露、自身抗体谱(抗 dsDNA、抗磷脂抗体、低补体)各自独立地增加心血管风险5,31。至关重要的是,低社会经济地位的患者可能经历诊断延迟、更不积极的免疫抑制管理以及更高的狼疮性肾炎相关器官损伤率,从而通过生物途径放大疾病严重程度和心血管风险15,24,32。

与这种综合视角一致,Shantz 等人使用范围性回顾的方法学来映射 SLE 中心血管疾病的心理社会决定因素,识别研究缺口,并强调需要从传统的生物医学模型转向明确解决社会决定因素、心理因素和心血管健康公平结构性障碍的方法27。应将使用经过验证的工具(如 PRAPARE 协议或健康相关社会需求筛查工具)进行社会决定因素的常规筛查纳入 SLE 临床护理。识别社会需求并与强大的转诊系统相结合,将患者与社区资源、社会服务、社区健康工作者、患者导航员和社会工作者联系起来。与食品银行、住房援助计划和交通服务的合作可以帮助减少心血管疾病预防和管理的障碍。

临床意义和建议

以下建议根据证据来源和强度进行区分。由 SLE 特异性试验或队列研究支持的建议被评为高/SLE 特异性。从一般心血管指南(ACC/AHA, EULAR)推断出适用于 SLE 的建议被评为中等或低/推断;这些需要在获得 SLE 特异性试验数据之前根据临床判断进行,如表 1 所示。

医疗保健提供者在减轻 SLE 患者心血管疾病风险方面发挥着关键作用,通过全面的风险评估、基于证据的干预措施、患者教育和推动健康公平。表 1 总结了 SLE 中的关键心血管疾病风险因素及其相应的基于证据的医疗保健提供者干预措施。以下临床建议综合了审查文献中的证据:

  • 系统地筛查未满足的社会需求,并将患者与适当的资源联系起来。与社会工作者、社区健康工作者和基于社区的组织合作,解决食品不安全、住房不稳定、交通挑战和经济限制。倡导扩大获得负担得起的医疗保健、健康食品和安全社区的政策27, 33。

  • 积极管理传统风险因素:SLE 患者需要比一般人群推荐的更严格的心血管疾病风险因素目标。适当的靶标包括血压<130/80 mm Hg,高风险个体的低密度脂蛋白胆固醇<70 mg/dL,糖尿病患者的糖化血红蛋白<7%。应根据个体心血管风险评估(≥10 年 ASCVD 风险≥7.5%)使用 ACC/AHA 池合队列方程或 EULAR 建议启动他汀类药物治疗,认识到 SLE 独立地放大了风险,超过了标准计算器的捕捉范围;在活动性疾病、狼疮性肾炎或其他 SLE 特异性风险修饰因素的患者中,他汀类药物治疗可能在计算风险阈值较低时是适当的35。

  • 健康生活方式的促进:结构化的项目,包括体力活动、饮食模式、体重管理和戒烟,应成为SLE治疗的核心组成部分。根据疾病活动度和功能能力制定个性化的运动计划。饮食咨询应强调地中海式饮食模式、钠限制和充足的微量营养素摄入。

  • 药物优化:通过早期使用替代糖皮质激素的免疫抑制剂来最小化累积糖皮质激素暴露。除非禁忌,否则继续使用羟氯喹,因其对脂质谱和血栓形成风险有有利影响。根据个体风险特征考虑使用心脏保护药物(ACE抑制剂或ARBs、他汀类药物、适当候选者的低剂量阿司匹林)28, 30。

  • 提升健康素养:使用简单语言、视觉辅助工具和教授法提供患者教育,以确保理解。文化定制的教育材料增强了相关性和参与度。尊重患者偏好和价值观的共同决策方法加强了治疗联盟并改善了依从性25, 28。

  • 解决社会决定因素:系统筛查未满足的社会需求,并将患者与适当资源联系起来。与社会工作者、社区健康工作者和基于社区的组织合作,解决食物不安全、住房不稳定、交通挑战和经济限制。倡导扩大获得负担得起的医疗保健、健康食品和安全社区的政策27, 33。

  • 响应文化差异的护理:认识到并解决结构性种族主义和歧视对健康结果的影响。确保提供专业的翻译服务,将医疗信息翻译成患者的首选语言,并尽可能提供文化相符的提供者。通过一致、尊重、不带偏见的互动建立信任。

大多数审查的证据来自高收入国家,特别是美国。在低收入和中等收入国家(LMICs),SLE的发病率较高,疾病表现往往更为严重,由于诊断延迟和治疗可及性有限,实施挑战更大。障碍包括风湿病专家的有限可用性、免疫抑制疗法的受限访问、碎片化的或不存在的健康保险系统24,以及对AI部署的信息技术基础设施不足。拉丁美洲GLADEL队列数据表明,在LMIC背景下,这些社会决定因素通过不同于北美队列所记录的途径加剧了心血管疾病风险19。未来的研究和临床指南制定必须明确解决这些全球现实,以确保SLE-CVD预防的进步能够公平地惠及所有受影响的人群。

SLE-CVD风险预测中的人工智能和精准医疗

机器学习在心血管疾病风险分层中的潜力

虽然AI和机器学习方法在SLE患者的CVD风险分层方面显示出前景,但目前的证据基础仍处于初步阶段。针对SLE特异性模型(如SLECRISK和梯度提升模型)是在单中心学术队列中开发的,据我们所知,这些模型尚未在独立的SLE人群中进行前瞻性外部验证36, 37。这些工具尚未准备好广泛应用于临床。

在一般人群队列中,包括多族群动脉硬化研究(MESA),机器学习模型的表现优于传统的方程,如ACC/AHA混合队列方程38, 39。这些发现表明可行性,但不能替代SLE特异性验证。提高区分度(例如,更高的AUC)是必要的,但不足以用于临床应用。SLE特异性模型是否能通过AI指导的风险分层改善患者预后,仍需前瞻性评估,目前这些模型尚未进行此类评估36, 37。

Earlier work in the Multi-Ethnic Study of Atherosclerosis (MESA) demonstrated that ML models outperformed the ACC/AHA risk calculator for cardiovascular risk prediction. Importantly, improved discriminative performance (e.g., higher area under the curve [AUC] values) is a necessary but not sufficient condition for clinical utility39. Whether AI-guided cardiovascular risk stratification translates into improved patient outcomes or changes in clinical practice requires prospective evaluation evidence that does not yet exist for any of the SLE-specific models reviewed17, 36, 37, 39 establishing the foundation for SLE-specific AI approaches. This synthesis examines the evidence with these caveats in view.

Recent advances in SLE-specific CVD risk prediction

Recent advances in SLE-specific cardiovascular risk prediction have accelerated the development of machine learning models tailored to the unique clinical and immunologic features of systemic lupus erythematosus. Choi et al. introduced SLECRISK, a novel ML-based tool trained on longitudinal data from 1,243 patients with SLE followed for 10 year36. The model integrates traditional CVD risk factors with SLE-specific variables including disease activity, disease duration, serum creatinine, and autoantibody profiles (anti-dsDNA, anti-RNP, lupus anticoagulant, and anti-Ro), as well as complement levels. SLECRISK achieved a sensitivity of 0.74 for identifying patients at moderate-to-high CVD risk (≥ 7.5%), outperforming the ACC/AHA Pooled Cohort Equations (sensitivity 0.38). Notably, it identified 3.4-fold higher-risk patients, particularly younger women with severe SLE who lacked traditional CVD risk factors but exhibited significant disease-related vulnerability.

Building on this progress, Liu et al. developed an interpretable gradient boosting machine (GBM) model to predict early cardiovascular involvement in SLE37. Using data from 1,023 patients with a median follow-up exceeding 10 years, the model identified arthritis, hypertension, HDL-cholesterol, LDL-cholesterol, C-reactive protein, erythrocyte sedimentation rate, and total cholesterol as influential predictors. By incorporating the DALEX interpretability framework, the authors visualized individualized risk profiles and each variable’s relative contribution, addressing the common “black box” critique of machine-learning approaches.

Liu C et al. used weighted gene coexpression network analysis with LASSO, SVM-RFE, and random forest to identify five hub genes (SPI1, MMP9, C1QA, CX3CR1, and MNDA) associated with atherosclerosis risk in SLE. They reported nomogram AUCs of 0.900–0.981 in public transcriptomic datasets17. That result applies to gene-based atherosclerosis risk prediction; it does not establish a pooled AUC across clinical events, arrhythmias, or heart failure.

The two SLE-specific models (SLECRISK and the GBM model) have important limitations. Both were developed and internally validated in single-center academic cohorts (1,243 and 1,023 patients, respectively) and, to our knowledge, have not undergone prospective external validation in an independent, diverse SLE population36, 37. Sample sizes are modest relative to feature complexity, which increases the risk of overfitting, and training cohorts may not represent community, safety-net, or LMIC populations. Researchers should acknowledge these constraints before clinical deployment.

A promising research direction is integration of social determinants of health into CVD prediction models. Electronic health records increasingly capture structured data on insurance, education, neighborhood deprivation, housing, and food security, and natural language processing can recover related information from clinical notes. Models that combine biological, clinical, and social data may improve risk stratification and help identify patients who need social support as well as medical management. ²⁵ This approach remains conceptual until such models are developed and externally validated in SLE.

At the same time, natural language processing can extract additional information from unstructured clinical notes, including documentation of housing instability, food insecurity, or transportation barriers. Models that combine biological, clinical, and SDOH data may improve predictive performance while helping identify patients who could benefit from targeted social support interventions alongside medical management27. This integrative approach supports a more holistic strategy that triggers both clinical action and SDOH-informed care pathways, with the potential to improve cardiovascular risk management in SLE.

Implementation challenges and ethical considerations

Algorithmic bias and health equity

ML models can perpetuate or amplify existing health disparities if training datasets underrepresent minority populations, exclude relevant SDOH variables, or encode biased clinical decisions.36, 39 Rigorous attention to training-data representativeness, transparent reporting of model performance across demographic subgroups, and continuous monitoring for disparate impact are essential.36, 39 In addition, inclusive model development practices, such as involving frontline clinicians, patients, and community stakeholders who understand the lived realities and care contexts of the populations represented in the data, can help ensure that annotation, validation, and interpretation processes do not inadvertently reinforce structural inequities. Although empirical work in SLE-specific AI is still emerging, these principles are increasingly recognized as foundational to the equitable implementation of ML-based cardiovascular risk prediction.

Interpretability and clinical trust

Clinicians may be reluctant to act upon recommendations from opaque algorithms. Advances in explainable AI, including feature importance plots, individual prediction explanations, and counterfactual reasoning, can help bridge this gap and make model predictions more interpretable and actionable17, 36, 37 While explainability is essential, the successful implementation of AI-based CVD risk prediction tools also depends on broader considerations, including clinician trust, local validation, usability, and alignment with clinical workflows. Validating models in the populations they are intended to serve, minimizing alert burden, and providing adequate end-user training may support more effective adoption. Co-design with clinicians and other stakeholders can further enhance usability and point-of-care actionability, helping to ensure that explainable AI tools translate into meaningful improvements in cardiovascular risk management for patients with SLE35

External validation

Most ML models for SLE-CVD prediction have been developed in single-center cohorts, raising concerns about generalizability. Rigorous external validation in diverse, independent cohorts is essential before clinical deployment36, 37 While evidence specific to SLE remains limited, broader methodological guidance in cardiovascular AI highlights the importance of evaluating model performance across different geographic regions, healthcare systems, and care settings, as well as across racial/ethnic, socioeconomic, and age subgroups. Assessing calibration, subgroup performance, fairness, clinical utility, and temporal stability can further strengthen confidence in model reliability. Multi-site collaborations and data-sharing initiatives will be critical to enabling the robust external validation needed to ensure that ML-based CVD risk prediction tools are both accurate and equitable across diverse SLE populations.

The path forward

Realizing AI’s potential for precision cardiovascular care in SLE will require sustained collaboration among rheumatologists, cardiologists, data scientists, implementation scientists, patient advocates, and health equity researchers to define how AI models can be integrated into detection and treatment pathways. Priority areas for future investigation include: (1) multi-center prospective studies validating AI prediction tools across diverse patient populations; (2) investigation of whether AI-guided risk stratification may improve clinical outcomes; (3) development of decision support tools combining AI risk predictions with evidence-based treatment recommendations; (4) incorporation of patient-reported outcomes and wearable device data into dynamic risk models; (5) ethical frameworks ensuring AI promotes rather than undermines health equity; and (6) cost-effectiveness analyses of AI-based approaches37.

Limitations

This review has several inherent limitations. As a narrative review without a pre-registered protocol, it does not meet the standards of a systematic review, and we did not apply formal PRISMA reporting. The search was restricted to English-language publications, which may introduce language bias and underrepresent findings from non-English-speaking regions. Study selection relied on reviewer judgment rather than independent dual-reviewer assessment with formal inter-rater reliability, introducing potential selection bias. The rapidly evolving AI and machine learning landscape means the review does not capture studies published after March 2026. Finally, the predominance of evidence from high-income, primarily North American settings limits direct generalizability to low- and middle-income country contexts.

Conclusions

Cardiovascular disease is one of the leading causes of mortality among individuals with SLE, driven by a complex interplay of disease-specific pathophysiology, traditional cardiovascular risk factors, and social determinants of health. This narrative review synthesizes evidence demonstrating that socioeconomic disadvantage, limited healthcare access, education inequities, and exposure to structural racism systematically elevate CVD risk in vulnerable SLE populations, particularly Black and Hispanic individuals. Lifestyle factors, including physical inactivity, tobacco use, and poor dietary patterns, further compound this vulnerability. Healthcare providers must adopt comprehensive, equity-focused approaches to CVD prevention in SLE that extend beyond traditional biomedical interventions to address social determinants, cultural factors, and structural barriers to health20, 32, 33.

Critical evidence gaps define the priorities for future research in SLE-CVD prevention. Prospective multi-center validation of SLE-specific AI risk prediction models in diverse populations is urgently needed before clinical deployment. Randomized controlled trials examining whether AI-guided, SDOH-informed interventions improve major cardiovascular outcomes in SLE have not yet been conducted. Longitudinal biomarker research is needed to understand how socioenvironmental exposures interact with SLE disease activity to amplify CVD risk. Implementation science research is needed to understand equitable deployment of AI tools in under-resourced and LMIC settings. Addressing these gaps should be a priority for future precision prevention research in SLE. Integrating lifestyle and SDOH data with clinical variables may improve risk discrimination beyond conventional risk assessment and could identify high-risk individuals who may benefit from targeted preventive interventions. However, careful attention to algorithmic bias, health equity, external validation, and ethical implementation is essential. Future research should prioritize external validation of AI tools, clinical outcome studies, and interventions that address social determinants among disadvantaged SLE populations. Only through sustained commitment to both scientific innovation and health equity can we reduce the burden of cardiovascular disease and improve quality of life for all individuals living with SLE.

Data availability

No datasets were generated or analysed during the current study.

Abbreviations

SLE:

systemic lupus erythematosus

AI:

artificial intelligence

CVD:

cardiovascular disease

References

  1. Centers for Disease Control and Prevention, Lupus & Published (2024). Available from: https://www.cdc.gov/lupus. Accessed 20 October 2025.

  2. Alammari, Y. et al. Cardiovascular morbidity in systemic lupus erythematosus: a single-center retrospective study. Cureus. 16 (4), e57842. https://doi.org/10.7759/cureus.57842 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

  3. Barbhaiya, M. et al. Race/ethnicity and cardiovascular events among patients with systemic lupus erythematosus. Arthritis Rheumatol. 69 (9), 1823–1831. https://doi.org/10.1002/art.40174 (2017).

    Article  PubMed  PubMed Central  Google Scholar 

  4. Joyce, D. P. et al. Prevalence of cardiovascular events in a population-based registry of patients with systemic lupus erythematosus. Arthritis Res. Ther. 26 (1), 160. https://doi.org/10.1186/s13075-024-03395-6 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

  5. Bello, N., Meyers, K. J., Workman, J., Hartley, L. & McMahon, M. Cardiovascular events and risk in patients with systemic lupus erythematosus: systematic literature review and meta-analysis. Lupus. 32 (3), 325–341. https://doi.org/10.1177/09612033221147471 (2023).

    Article  CAS  PubMed  Google Scholar 

  6. Wong, C. Y. et al. Cardiovascular risk factors and complications in patients with systemic lupus erythematosus with and without nephritis: a systematic review and meta-analysis. Lupus Sci. Med. 11 (1), e001152. https://doi.org/10.1136/lupus-2024-001152 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

  7. Sircana, M. C., Erre, G. L., Castagna, F. & Manetti, R. Crosstalk between inflammation and atherosclerosis in rheumatoid arthritis and systemic lupus erythematosus: is there a common basis? Life (Basel). 14 (6), 716. https://doi.org/10.3390/life14060716 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  8. Oliveira, C. B. & Kaplan, M. J. Cardiovascular disease risk and pathogenesis in systemic lupus erythematosus. Semin Immunopathol. 44 (3), 309–324. https://doi.org/10.1007/s00281-022-00922-y (2022).

    Article  PubMed  PubMed Central  Google Scholar 

  9. Kostopoulou, M., Nikolopoulos, D., Parodis, I. & Bertsias, G. Cardiovascular disease in systemic lupus erythematosus: recent data on epidemiology, risk factors, and prevention. Curr. Vasc Pharmacol. 18 (6), 549–565. https://doi.org/10.2174/1570161118666191227101636 (2020).

    Article  CAS  PubMed  Google Scholar 

  10. Papazoglou, N., Kravvariti, E., Konstantonis, G., Sfikakis, P. P. & Tektonidou, M. G. The impact of traditional cardiovascular risk factor control on 7-year follow-up atherosclerosis progression in systemic lupus erythematosus. Rheumatol. (Oxford). 63 (1), 50–57. https://doi.org/10.1093/rheumatology/kead184 (2024).

    Article  CAS  Google Scholar 

  11. Lu, M. C. & Koo, M. Effects of exercise intervention on health-related quality of life in patients with systemic lupus erythematosus: a systematic review and meta-analysis of controlled trials. Healthc. (Basel). 9 (9), 1215. https://doi.org/10.3390/healthcare9091215 (2021).

    Article  Google Scholar 

  12. Lu, X. et al. Patients with systemic lupus erythematosus face a high risk of cardiovascular disease: systematic review and meta-analysis. Int. Immunopharmacol. 94, 107466. https://doi.org/10.1016/j.intimp.2021.107466 (2021).

    Article  CAS  PubMed  Google Scholar 

  13. Blaess, J. et al. Recommendations for physical activity and exercise in persons living with systemic lupus erythematosus: consensus by an international task force. RMD Open. 10 (2), e004171. https://doi.org/10.1136/rmdopen-2024-004171 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

  14. Zhang, W. T. et al. Effects of tobacco smoking on cardiovascular disease in patients with systemic lupus erythematosus: a systematic review and meta-analysis. Front. Immunol. 13, 967506. https://doi.org/10.3389/fimmu.2022.967506 (2022).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  15. Garg, S. et al. Timing and predictors of incident cardiovascular disease in systemic lupus erythematosus: risk occurs early and highlights racial disparities. J. Rheumatol. 50 (1), 84–92. https://doi.org/10.3899/jrheum.220279 (2023).

    Article  PubMed  Google Scholar 

  16. Falasinnu, T., Chaichian, Y., Palaniappan, L. & Simard, J. F. Unraveling race, socioeconomic factors, and geographical context in the heterogeneity of lupus mortality in the United States. ACR Open. Rheumatol. 1 (3), 164–172. https://doi.org/10.1002/acr2.1024 (2019).

    Article  PubMed  PubMed Central  Google Scholar 

  17. Liu, C. et al. Identification of crucial genes for predicting the risk of atherosclerosis with system lupus erythematosus based on comprehensive bioinformatics analysis and machine learning. Comput. Biol. Med. 152, 106388. https://doi.org/10.1016/j.compbiomed.2022.106388 (2023).

    Article  CAS  PubMed  Google Scholar 

  18. Baethge, C., Goldbeck-Wood, S. & Mertens, S. SANRA scale for the quality assessment of narrative review articles. Res. Integr. Peer Rev. 4, 5. https://doi.org/10.1186/s41073-019-0064-8 (2019).

    Article  PubMed  PubMed Central  Google Scholar 

  19. Pons-Estel, B. A. et al. The GLADEL multinational Latin American prospective inception cohort of 1,214 patients with systemic lupus erythematosus: ethnic and disease heterogeneity among Hispanics. Med. (Baltim). 2004; 83 (1), 1–17. https://doi.org/10.1097/01.md.0000104742.42401.e2

    Article  Google Scholar 

  20. Alarcón, G. S. et al. Systemic lupus erythematosus in a multiethnic cohort: LUMINA XXXV. Predictive factors of high disease activity over time. Ann. Rheum. Dis. 65 (9), 1168–1174. https://doi.org/10.1136/ard.2005.040774 (2006).

    Article  PubMed  PubMed Central  Google Scholar 

  21. Raymond, W. D., Hamdorf, M., Furfaro, M., Eilertsen, G. O. & Nossent, J. C. Smoking is associated with increased BAFF and decreased interferon-γ levels in patients with systemic lupus erythematosus. Lupus Sci. Med. 8 (1), e000537. https://doi.org/10.1136/lupus-2021-000537 (2021).

    Article  PubMed  PubMed Central  Google Scholar 

  22. Sieczkowska, S. M. et al. Effects of lifestyle intervention on cardiovascular risk factors in systemic lupus erythematosus patients: the study Living well with lupus. Clin. Rheumatol. 43 (3), 1003–1013. https://doi.org/10.1007/s10067-024-06870-2 (2024).

    Article  PubMed  Google Scholar 

  23. Duarte-García, A. & Yazdany, J. Cardiovascular disease disparities in systemic lupus erythematosus. J. Rheumatol. 50 (1), 9–10. https://doi.org/10.3899/jrheum.220933 (2023).

    Article  PubMed  Google Scholar 

  24. Maynard, J. W., Fang, H. & Petri, M. Low socioeconomic status is associated with cardiovascular risk factors and outcomes in systemic lupus erythematosus. J. Rheumatol. 39 (4), 777–783. https://doi.org/10.3899/jrheum.110724 (2012).

    Article  PubMed  PubMed Central  Google Scholar 

  25. Ha Dinh, T. T., Bonner, A., Clark, R., Ramsbotham, J. & Hines, S. The effectiveness of the teach-back method on adherence and self-management in health education for people with chronic disease: a systematic review. JBI Database Syst. Rev. Implement. Rep. 14 (1), 210–247. https://doi.org/10.11124/jbisrir-2016-2296 (2016).

    Article  Google Scholar 

  26. McMahon, M., Seto, R. & Skaggs, B. J. Cardiovascular disease in systemic lupus erythematosus. Rheumatol. Immunol. Res. 2 (3), 157–172. https://doi.org/10.2478/rir-2021-0022 (2021).

    Article  PubMed  PubMed Central  Google Scholar 

  27. Shantz, E. et al. Towards an understanding of the biopsychosocial determinants of CVD in SLE: a scoping review. Lupus Sci. Med. 11 (2), e001155. https://doi.org/10.1136/lupus-2024-001155 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

  28. Sun, K. et al. Implementation of a clinician-led medication adherence intervention among patients with systemic lupus erythematosus. J. Rheumatol. 51 (9), 884–890. https://doi.org/10.3899/jrheum.2024-0071 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

  29. Herndon, S. et al. A qualitative study of facilitators of medication adherence in systemic lupus erythematosus: perspectives from rheumatology providers/staff and patients. Lupus. 33 (2), 137–144. https://doi.org/10.1177/09612033231225843 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

  30. Emamikia, S., Gentline, C., Enman, Y. & Parodis, I. How can we enhance adherence to medications in patients with systemic lupus erythematosus? Results from a qualitative study. J. Clin. Med. 11 (7), 1857. https://doi.org/10.3390/jcm11071857 (2022).

    Article  PubMed  PubMed Central  Google Scholar 

  31. Fava, A. & Petri, M. Systemic lupus erythematosus: diagnosis and clinical management. J. Autoimmun. 96, 1–13. https://doi.org/10.1016/j.jaut.2018.11.001 (2019).

    Article  PubMed  Google Scholar 

  32. Yazdany, J., Trupin, L., Schmajuk, G., Katz, P. & Yelin, E. Quality of care for lupus in a safety-net clinic: using chart audit to identify disparities. Arthritis Care Res. (Hoboken). 64 (6), 872–879. https://doi.org/10.1002/acr.21600 (2012).

    Article  CAS  Google Scholar 

  33. Drenkard, C. & Lim, S. S. Update on lupus epidemiology: advancing health disparities research through the study of minority populations. Curr. Opin. Rheumatol. 31 (6), 689–696. https://doi.org/10.1097/BOR.0000000000000646 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. Pocovi-Gerardino, G. et al. Beneficial effect of Mediterranean diet on disease activity and cardiovascular risk in systemic lupus erythematosus patients: a cross-sectional study. Rheumatol. (Oxford). 60 (1), 160–169. https://doi.org/10.1093/rheumatology/keaa210 (2021).

    Article  CAS  Google Scholar 

  35. Damen, J. A. et al. Performance of the Framingham risk models and pooled cohort equations for predicting 10-year risk of cardiovascular disease: a systematic review and meta-analysis. BMC Med. 17 (1), 109. https://doi.org/10.1186/s12916-019-1340-7 (2019).

    Article  PubMed  PubMed Central  Google Scholar 

  36. Choi, M. Y. et al. Personalizing cardiovascular risk prediction for patients with systemic lupus erythematosus. Semin Arthritis Rheum. 67, 152468. https://doi.org/10.1016/j.semarthrit.2024.152468 (2024).

    Article  PubMed  PubMed Central  Google Scholar 

  37. Liu, S. et al. Developing an interpretable machine learning model for early prediction of cardiovascular involvement in systemic lupus erythematosus. J. Inflamm. Res. 18, 8629–8641. https://doi.org/10.2147/JIR.S526608 (2025).

    Article  PubMed  PubMed Central  Google Scholar 

  38. Ward, A. et al. Machine learning and atherosclerotic cardiovascular disease risk prediction in a multi-ethnic population. npj Digit. Med. 3, 125. https://doi.org/10.1038/s41746-020-00331-1 (2020).

    Article  PubMed  PubMed Central  Google Scholar 

  39. Kakadiaris, I. A. et al. Machine learning outperforms the ACC/AHA CVD risk calculator in the Multi-Ethnic Study of Atherosclerosis (MESA) cohort. J. Am. Heart Assoc. 7 (22), e009476. https://doi.org/10.1161/JAHA.118.009476 (2018).

    Article  PubMed  PubMed Central  Google Scholar 

Download references

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Author information

Authors and Affiliations

  1. College of Nursing and Health Sciences, Seattle University, Seattle, WA, USA

    Fatemeh Choupani

  2. School of Nursing, University of Washington, Seattle, WA, USA

    Fatemeh Choupani

  3. Biomedical Engineering Department, William A. Brookshire Department of Chemical and Biomolecular Engineering, Cullen College of Engineering, University of Houston, Houston, TX, USA

    Renita E. Horton

  4. Natural Sciences, College of Sciences and Technology, University of Houston Downtown, Houston, TX, USA

    Xyanthine Parillon

  5. Department of Medicine, Division of Infectious Diseases and Immunology, University of Massachusetts Chan Medical School, Worcester, MA, USA

    Zaida G. Ramirez-Ortiz

  6. Department of Congenital Heart Surgery, Baylor College of Medicine, Houston, TX, USA

    Lalita Wadhwa

  7. Environmental Sciences Graduate Program, Oregon State University, Corvallis, OR, USA

    Joyonna Gamble-George

Authors

  1. Fatemeh Choupani
  2. Renita E. Horton
  3. Xyanthine Parillon
  4. Zaida G. Ramirez-Ortiz
  5. Lalita Wadhwa
  6. Joyonna Gamble-George

Contributions

FC: Conceptualization, writing original draft, writing review & editing, Supervision, REH: Writing original draft, writing review & editing. XP: Writing original draft, writing review & editing. ZGR-O: writing original draft, writing review & editing. LW: Writing review & editing. JGG: writing review & editing, Supervision. All authors read and approved the final manuscript.

Author Biographies

Fatemeh Choupani (FC) is a Tenure-Track Assistant Professor in the College of Nursing and Health Sciences at Seattle University and an Affiliate Faculty member at the University of Washington School of Nursing. She is a registered nurse with extensive clinical experience in critical care, oncology, and psychiatric–mental health nursing. Her research focuses on health equity, social determinants of health, cardiovascular risk in autoimmune diseases, particularly systemic lupus erythematosus, and the ethical integration of artificial intelligence in nursing education and precision health. https://orcid.org/0000-0002-1679-8729

Renita E. Horton (REH) serves as an Assistant Professor in Biomedical Engineering and holds an affiliated appointment with the William A. Brookshire Department of Chemical and Biomolecular Engineering at the University of Houston, Houston, TX, USA. Her work has focused on organs-on-chips, autoimmunity and lupus, and cardiovascular disease modeling. Academic profile: ORCID: https://orcid.org/0000-0001-7589-9485

Xyanthine Parillon (XP) is a researcher affiliated with Natural Sciences, College of Sciences and Technology, University of Houston Downtown, Houston, TX, USA. Her research interests include lifestyle determinants of chronic disease, behavioral health interventions, autoimmunity, AI/ML modeling, and preventive medicine in autoimmune disease populations. Academic profile: ORCID: https://orcid.org/0000-0002-4780-508X

Zaida G. Ramirez-Ortiz (ZGR-O) is a researcher and Tenure-track Assistant Professor in the Department of Medicine, University of Massachusetts Chan Medical School, Worcester, MA, USA. Her scholarly interests include clinical immunology and autoimmunity, clinical informatics, artificial intelligence in rheumatology and cardiovascular medicine, and health equity in AI model development. Academic profile: https://orcid.org/0000-0002-0253-0473

Lalita Wadhwa (LW) is a clinician-researcher affiliated with the Department of Congenital Heart Surgery, Baylor College of Medicine, Houston, TX, USA. Her expertise encompasses cardiovascular medicine and interdisciplinary cardiovascular risk management in complex patient populations. Academic profile: ORCID ID: https://orcid.org/0000-0003-4506-7826

Joyonna Gamble-George (JGG) is a researcher affiliated with the Environmental Sciences Graduate Program, Oregon State University, Corvallis, OR, USA. Her research focuses on environmental health sciences, structural determinants of health equity, translational research methodology, multi-omics, and artificial intelligence. Academic profile: https://orcid.org/0000-0002-5492-9216

Corresponding author

Correspondence to Fatemeh Choupani.

Ethics declarations

Ethics approval and consent to participate

This study did not require ethics approval. This manuscript is a narrative review synthesizing previously published peer-reviewed research; no primary data involving human participants were collected or analyzed.

Professional Relationships

All authors are research collaborators with pre-existing professional relationships established across multiple projects addressing health equity, chronic disease prevention, and clinical innovation. Fatemeh Choupani (FC) is the connecting collaborator through whose network all author relationships are established. FC and JGG also have a mentor–mentee relationship. These professional relationships do not raise any conflicts of interest

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Choupani, F., Horton, R.E., Parillon, X. et al. Social determinants of health, lifestyle factors, and artificial intelligence in cardiovascular risk stratification for systemic lupus erythematosus: a narrative review. Sci. Rev. 1, 34 (2026). https://doi.org/10.1038/s44573-026-00037-7

Download citation

  • Received: 27 May 2026

  • Revised: 30 September 2026

  • Accepted: 16 October 2026

  • Published: 20 November 2026

  • Version of record: 20 November 2026

  • DOI: https://doi.org/10.1038/s44573-026-00037-7

Keywords

来源:Nature 健康护理主题 · nature.com