Planning for a pandemic: a global disease modelling platform

规划大流行：全球疾病建模平台

• 批准号: 59745
• 金额: $ 6.23万
• 依托单位: INTELLEGENS LIMITED
• 依托单位国家: 英国
• 项目类别: Feasibility Studies
• 财政年份: 2020
• 资助国家: 英国
• 起止时间: 2020 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=59745
• 关键词: Planning; pandemic; global; disease; modelling

The COVID-19 pandemic poses significant difficulties for the UK. Headlines are full of stories of the shortage of critical supplies that whip up a political whirlwind: personal protective equipment, critical drugs for patients in intensive care, ventilators, and sufficient hospital staff. The problem extends beyond the COVID-19 pandemic, with the same issues arising annually with diseases such as seasonal flu. More importantly, should we be faced with a similar pandemic in the future how can we manage it more efficiently. The government and companies have to plan how to meet peak demand for critical supplies and care providers must ensure that sufficient staff are in place and are properly equipped. However, they fly blind with little guidance, and so cannot produce supplies in advance or plan how to deploy resources. Over recent years AI and deep learning have become mainstream, delivering new insights and improvements across all parts of society. However, in most cases good quality data is needed to build these models, when there is little data, or it is not accurate, the models created will also be inaccurate and of little use. Intellegens, a spin out from Cambridge University, has developed a new type of machine learning to deal with these type of problems where data is limited and noisy. In these rapidly changing pandemic environments data quality and sparsity is making machine learning model development hard and unstable. In this project, Intellegens will develop an accurate machine learning model to predict the progression of pandemics and other diseases. Using all available data as it becomes available, which could include:* Real time patient case metrics including number of infections, recoveries, and deaths* Population metrics such as age distribution, housing density, and connectedness* Fraction of the population believed to have been infected but not tested* Number of unreported deaths outside the hospital system* Statistics from limited testing facilities* Profiling of most at risk groups based on age, ethnicity, or other conditions* Environmental, geographical, and economic indicatorsThis tool will ultimately allow policy makers and companies to plan ahead for difficult times, ensuring that appropriate supplies and staffing levels are in place for the given rapidly changing environment

COVID-19疫情为英国带来重大困难。报纸头条充斥着关键物资短缺的故事，这些物资引发了一场政治旋风：个人防护设备、重症监护病人的关键药物、呼吸机和足够的医院工作人员。这个问题不仅限于COVID-19大流行，季节性流感等疾病每年都会出现同样的问题。更重要的是，如果我们将来面临类似的大流行病，我们如何更有效地管理它。政府和公司必须计划如何满足对关键用品的高峰需求，护理提供者必须确保有足够的工作人员到位并配备适当的设备。然而，他们盲目飞行，几乎没有指导，因此无法提前生产补给或计划如何部署资源。近年来，人工智能和深度学习已经成为主流，为社会各个领域提供了新的见解和改进。然而，在大多数情况下，需要高质量的数据来构建这些模型，当数据很少或不准确时，创建的模型也会不准确且用处不大。从剑桥大学分拆出来的英特尔公司开发了一种新型的机器学习方法，用于处理数据有限且有噪声的问题。在这些快速变化的流行病环境中，数据质量和稀疏性使得机器学习模型的开发变得困难和不稳定。在这个项目中，英特尔将开发一个准确的机器学习模型来预测流行病和其他疾病的进展。使用所有可用的数据，其中可能包括：* 真实的患者病例指标，包括感染、康复和死亡人数 * 人口指标，如年龄分布、住房密度和连通性 * 据信已感染但未接受检测的人口比例 * 医院系统外未报告的死亡人数 * 来自有限检测设施的统计数据 * 根据年龄、种族或其他条件对最高危人群进行分析 * 环境、地理和经济指标该工具最终将使政策制定者和公司能够提前计划应对困难时期，确保为快速变化的环境提供适当的供应和人员配备