ROAD2H: Resource Optimisation, Argumentation, Decision Support and Knowledge Transfer to Create Value via Learning Health Systems

ROAD2H：资源优化、论证、决策支持和知识转移，通过学习健康系统创造价值

• 批准号: EP/P029558/1
• 负责人: Ara Darzi
• 金额: $ 193.16万
• 依托单位: Imperial College London
• 依托单位国家: 英国
• 项目类别: Research Grant
• 财政年份: 2017
• 资助国家: 英国
• 起止时间: 2017 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=EP%2FP029558%2F1
• 关键词: ROAD2H; Resource; Optimisation; Argumentation; Decision

ROAD2H will develop novel Learning Health System (LHS) techniques to facilitate uptake of health interventions to achieve Universal Health Coverage (UHC) in Low- and Middle-Income Countries (LMICs). UHC is a specific target in the UN Sustainable Development Goals (SDG 3.8), with many LMICs attempting to extend population coverage of quality healthcare via public-subsidised health insurance schemes (HISs). Two issues are key to achieving UHC via these schemes: i) how budgets should be best allocated among competing clinical interventions (allocative efficiency) ii) under what timing and circumstances particular interventions should be offered for what kinds of patients in which settings and to what standard of care (quality).Current attempts in LMICs to address efficiency and quality issues include Health Technology Assessment (HTA) based on cost-effectiveness analysis of individual healthcare interventions, and evidence-based guidelines/pathways recommending best practice within whole disease areas. However there have been limited uptake of system-wide mathematical optimisation techniques (in part due to significant data requirements); no efforts to using data routinely collected by HISs for billing and claims purposes, as well as Electronic Health Records (EHRs), for improving efficiency and quality, even though existing attempts among LMICs to use such data retrospectively as inputs into HTA and guidelines show promise. We hypothesise that a LHS framework for national HISs, combining resource optimization, argumentation (as understood in AI), clinical decision support and clinical and policy analytics, will facilitate the uptake of evidence-based, cost-effective and data-driven healthcare interventions, and improve the overall efficiency and quality of integrated care systems in LMICs transitioning to UHC. ROAD2H will deliver DSS tools that integrate individual patient data (drawn from EHRs) with localized clinical guidelines (including HTA, clinical pathways) obtained via policy and clinical analytics from best available clinical and cost-effectiveness evidence as well as HIS billing/claim data. We will use argumentation to parameterize optimisation problems drawn from cost-effectiveness and resource optimization considerations, and then use standard methods in mathematical optimisation to maximize efficiency of clinical interventions. We will use argumentation also to integrate optimization for maximal efficiency and reasoning with localized guidelines, and resolve conflicts as well as transparently explain recommendations for clinical interventions. The DSS will make use of these explained recommendations as well as data provenance techniques to generate and transparently collect patient-tailored recommendations to enable further learning, in line with the LHS methodology. The DSS tools will operate at the EHR/clinician/patient interaction and integrate data from multiple sources as well as have localised interfaces (by country and possibly hospital/clinician).The project will be carried out in three phases:- Phase 1: To review existing HISs and infrastructures in three exemplar countries (China, Serbia, Myanmar), against the requirements for LHSs- Phase 2: Iterative development of a prototype digital methodology (encompassing policy, data, cost-effectiveness analytics, optimization, argumentation, DSS and provenance) to be embedded into existing HISs - Phase 3: To evaluate, by means of proof-of-concept prototype DSS tools in Serbia and China and by scoping in Myanmar, the technical and clinical feasibility of the ROAD2H methodology, its acceptability among end users, and its potential and preconditions for generalization across disease areas and health systems settings in order to support UHC.

ROAD2H将开发新的学习卫生系统（LHS）技术，以促进采用卫生干预措施，在低收入和中等收入国家实现全民健康覆盖。全民健康覆盖是联合国可持续发展目标（可持续发展目标3.8）中的一项具体目标，许多中低收入国家试图通过公共补贴医疗保险计划（HISs）扩大高质量医疗保健的人口覆盖范围。两个问题是通过这些计划实现全民健康覆盖的关键：1)如何在相互竞争的临床干预措施之间最佳分配预算（分配效率）2)在什么时间和情况下，在什么环境下，为什么样的患者提供特定的干预措施，以及采取什么样的护理标准（质量）。低收入和中等收入国家目前为解决效率和质量问题所做的尝试包括：基于个人卫生保健干预措施成本效益分析的卫生技术评估（HTA），以及在整个疾病领域推荐最佳做法的循证指南/途径。然而，系统范围的数学优化技术的采用有限（部分原因是由于重要的数据需求）；没有努力将HISs常规收集的数据用于计费和索赔目的，以及电子健康记录（EHRs），以提高效率和质量，尽管中低收入国家正在尝试将此类数据回顾性地用作HTA和指南的输入，但这显示出了希望。我们假设，将资源优化、论证（如人工智能所理解的）、临床决策支持以及临床和政策分析相结合的国家HISs的LHS框架，将促进以证据为基础、具有成本效益和数据驱动的医疗干预措施的采用，并提高中低收入国家向全民健康覆盖过渡的综合医疗系统的整体效率和质量。ROAD2H将提供DSS工具，将个人患者数据（来自电子病历）与通过政策和临床分析获得的本地化临床指南（包括HTA，临床路径）整合在一起，这些分析来自最佳可用的临床和成本效益证据以及HIS账单/索赔数据。我们将使用论证来参数化从成本效益和资源优化考虑中得出的优化问题，然后在数学优化中使用标准方法来最大化临床干预的效率。我们还将使用论证来整合最大化效率的优化和本地化指南的推理，并解决冲突以及透明地解释临床干预的建议。DSS将利用这些已解释的建议以及数据来源技术，生成和透明地收集针对患者的建议，以便根据LHS方法进行进一步学习。DSS工具将在电子病历/临床医生/患者互动中运行，整合来自多个来源的数据，并具有本地化接口（按国家和可能的医院/临床医生）。该项目将分三个阶段进行：-第一阶段：根据lhs的要求，审查三个示范国家（中国、塞尔维亚、缅甸）现有的hss和基础设施-第二阶段：迭代开发将嵌入现有hss的原型数字方法（包括政策、数据、成本效益分析、优化、论证、DSS和来源）-第三阶段：通过塞尔维亚和中国的概念验证原型DSS工具以及缅甸的范围界定，评估ROAD2H方法的技术和临床可行性，其在最终用户中的可接受性，以及其在疾病领域和卫生系统环境中推广的潜力和先决条件，以支持全民健康覆盖。

项目成果

Argumentation for Explainable Scheduling

可解释调度的论证

• DOI: 10.1609/aaai.v33i01.33012752
• 发表时间: 2019
• 期刊: Proceedings of the AAAI Conference on Artificial Intelligence
• 作者: Cyras K

PRIMA 2017: Principles and Practice of Multi-Agent Systems

PRIMA 2017：多智能体系统原理与实践

• DOI: 10.1007/978-3-319-69131-2_25
• 发表时间: 2017
• 作者: Bao Z

From fine-grained properties to broad principles for gradual argumentation: A principled spectrum

• DOI: 10.1016/j.ijar.2018.11.019
• 发表时间: 2019-02-01
• 期刊: INTERNATIONAL JOURNAL OF APPROXIMATE REASONING
• 影响因子: 3.9
• 作者: Baroni, Pietro;Rago, Antonio;Toni, Francesca
• 通讯作者: Toni, Francesca

Resolving Conflicts in Clinical Guidelines using Argumentation

使用论证解决临床指南中的冲突

• 发表时间: 2019
• 作者: Cyras K

Computational complexity of flat and generic Assumption-Based Argumentation, with and without probabilities

平面和通用的基于假设的论证的计算复杂性，有或没有概率

• DOI: 10.1016/j.artint.2020.103449
• 发表时间: 2021
• 期刊: Artificial Intelligence
• 影响因子: 14.4
• 作者: Cyras K