Development of patient-centric digital biobank for world-leading chronic disease research: using AI to uniquely map genomic, clinical and longitudinal health data

COVID-19危机使全球无法预测个性化疾病进展的问题成为人们关注的焦点，即为什么有些人比其他人遭受更严重的痛苦或对治疗的反应不同。原因很复杂，是遗传、人口统计学和表型因素共同作用的结果。对于几乎每一种常见和罕见疾病，都没有中央存储库或工具/机制供研究人员快速访问/协调/查询这些数据。这大大降低了我们准确识别高危人群和开发新药物/治疗方法的能力。为了解决这一问题，并进入快速增长的500亿英镑以上的市场（alliedmarketresearch:2018），我们打算建立一个以患者为中心的慢性病数字生物库，包括基因组、临床和纵向健康数据。随着个人基因测序市场（包括公共和私人市场，到2025年将达到192亿英镑；CAGR-Allied market Research:2017年复合年增长率为19%）的持续增长，存储的个人健康数据量也在不断增长，这反过来又为药物发现、更快地访问目标患者群体和降低医学研究中的患者招募成本开辟了新的机会。特别是人工智能和机器学习有可能极大地加速发现用于患者分层的新型生物标志物和使用多种数据类型（包括基因组、临床和患者生成的纵向数据）作为输入的新型药物靶点。基因组学支持的药物成功的可能性要高2-3倍，但获取正确的数据是一个重大挑战。利用人工智能和机器学习方法进行统计遗传学，结合大数据架构，Sano genetics将与Patients Know Best、C4X Discovery和BenevolentAI合作，开发以患者为中心的基因组、临床和纵向健康数据（TRL6）数字生物银行平台，该平台将允许用户快速识别表型-基因型模式，以发现新的生物标记物来对患者进行分层，并发现新的药物靶点和化合物。为了帮助建立平台，我们选择最初专注于溃疡性结肠炎（一种炎症性肠病）——到2026年，这一市场价值174亿英镑，全球有超过800万人受到影响，并且拥有大量可用数据——然后寻求扩展平台，以解决其他慢性病和新出现的威胁（如COVID-19）中相同的数据访问/分析挑战。我们预测疾病对患者的影响和严重程度的能力在规划、资源和政策方面带来了重大挑战。

Public descriptionThe COVID-19 crisis has thrown into sharp focus the global inability to predict personalised disease progression i.e. why do some people suffer more severely than others or respond differently to treatments. The reason is complex and due to a combination of genetic, demographic, and phenotypic factors. For nearly every common and rare disease, there is no central repository nor tools/mechanisms for researchers to quickly access/harmonise/interrogate this data. This drastically reduces our ability to accurately identify those most at risk and to develop new medicines/treatments.To address this and tap into a fast-growing £50bn+ market(alliedmarketresearch:2018), we intend to build a patient-centric, chronic disease digital biobank of genomic, clinical, and longitudinal health data.As the market for personal genetic sequencing- both public and private (£19.2bn by 2025;19% CAGR-Allied Market Research:2017)- continues to grow, as does the volume of personal health data stored, which in turn opens up new opportunities for drug discovery, quicker access to target patient groups and reducing patient recruitment costs in medical research.Artificial Intelligence and Machine Learning in particular have the potential to dramatically speed up the discovery of novel biomarkers for patient stratification and novel drug targets using multiple data-types including genomic, clinical, and patient-generated longitudinal data as input.Medicines supported by genomics are 2-3x more likely to succeed, but accessing the right data is a major challenge. Using AI and Machine Learning approaches to statistical genetics, combined with big-data architectures, Sano Genetics, in partnership with Patients Know Best, C4X Discovery and BenevolentAI will develop the patient-centric digital biobank platform of genomic, clinical, and longitudinal health data (TRL6) that will allow users and to quickly identify phenotype-genotype patterns, in order to discover novel biomarkers to stratify patients and to discover novel drug targets and compounds.To help build the platform, we have chosen to initially focus on Ulcerative Colitis(a form of Inflammatory Bowel Disease)- a market worth £17.4bn by 2026, that affects over 8m globally and that has significant available data- before looking to scale the platform to address the same data-access/analysis challenge in other chronic conditions and emerging threats such as COVID-19, where our ability to predict the diseases impact and severity in patients has created significant challenges in planning, resources and policy.