Deep Learning for diagnosing patients with rare genetic diseases

There are over 7,000 unique rare diseases, some of which affecting 3,500 or fewer patients in the US. Due to clinicians' limited experience with such diseases and the considerable heterogeneity of their clinical presentations, many patients with rare genetic diseases remain undiagnosed. While artificial intelligence has demonstrated success in assisting diagnosis, its success is usually contingent on the availability of large annotated datasets. Here, we present SHEPHERD, a deep learning approach for multi-faceted rare disease diagnosis. To overcome the limitations of supervised learning, SHEPHERD performs label-efficient training by (1) training exclusively on simulated rare disease patients without the use of any real labeled data and (2) incorporating external knowledge of known phenotype, gene and disease associations via knowledge-guided deep learning. This repository houses (1) the preprocessed rare disease knowledge graph, (2) the simulated patients used for training SHEPHERD, and (3) the myGene2 rare disease patients used for evaluation. The accompanying github repository can be found at: https://github.com/mims-harvard/SHEPHERD.目前全球已识别出7000余种独特的罕见病，其中部分病种在美国的患者人数不足3500人。由于临床医生对这类疾病的诊疗经验相对匮乏，且其临床表现存在显著异质性，许多罕见遗传病患者始终无法获得明确诊断。尽管人工智能（Artificial Intelligence）在辅助诊断领域已展现出良好的应用潜力，但此类模型的性能通常依赖于大规模标注数据集的支撑。本文提出了SHEPHERD，一种面向多维度罕见病诊断的深度学习（Deep Learning）方法。为克服监督学习的局限性，SHEPHERD通过两种路径实现标签高效训练：(1) 仅基于模拟罕见病患者数据开展训练，无需使用任何真实标注数据；(2) 通过知识引导的深度学习技术，整合已知的表型、基因与疾病关联的外部知识。本数据集仓库包含以下三项内容：(1) 预处理后的罕见病知识图谱（Knowledge Graph），(2) 用于训练SHEPHERD的模拟患者数据集，以及(3) 用于模型评估的myGene2罕见病患者数据集。配套的GitHub仓库可通过以下链接获取：https://github.com/mims-harvard/SHEPHERD。