Uncovering Clinical Evidence in COVID-19 Publications: An Integrated Search via Text & Images

项目摘要： 在COVID 19出版物中发现临床证据：通过文本和图像进行综合检索 拟议的研究旨在开发和推进使用科学图像数据的工具， 除文本外，我们亦会提供更多有关出版物的资料，以加快有效获取COVID-19公布的信息。 目前旨在应对COVID-19大流行的努力包括设计治疗，了解病毒 机制，检测感染和抗体，并最终开发疫苗。 所有这些努力都需要有效获取与该病毒有关的生物医学信息。艾伦研究所 最近发布了CORD-19数据集-一个大型的，不断更新的科学文献集合， COVID-19和冠状病毒。该数据集包含数万篇全文文章，为 文本挖掘工具，将支持访问与COVID-19有关的信息。 值得注意的是，这些出版物中的许多证据都是以数字的形式提供的。此外，区域 这些证据图像出现的地方信息丰富。 虽然生物医学基于文本的挖掘工具正在快速开发并提供用于访问该数据集， 不考虑包含关键临床和生物信息的图像。即使在COVID-19之外 在这一领域，迄今为止，在出版物中利用图像的工作很少，尽管它们提供了重要的信息， 关于文章中嵌入的信息的相关性的线索。 我们的前提，这是由我们自己和其他信息学家和临床医生的经验支持，是， 从图像中获得的信息可以（也应该）直接纳入生物医学-特别是 COVID-19文档检索和提取。这样做将改善对相关信息的准确访问 文章，同时指出其中的重要证据，并加快获得急需的关键 信息.该项目的工作将产生利用图像和图像的方法和工具 文本数据，促进更有效和更有针对性的检索和挖掘，从而更好地支持快速数据- 在COVID-19背景下的密集发现。

Project Summary: Uncovering clinical evidence in COVID19 publications: An integrated search via text & images The proposed research aims to develop and advance tools for using image-data appearing in scientific publications, in addition to text, in order to expedite effective access to COVID-19 published information. Current efforts aiming to address the COVID-19 pandemic include devising treatment, understanding virus mechanisms, detecting infection and antibodies, and ultimately – developing a vaccine. All these efforts require effective access to biomedical information related to the virus. The Allen Institute has recently released the CORD-19 dataset – a large, continually updated collection of scientific literature pertaining to COVID-19 and Corona viruses. This dataset comprises tens of thousands full text articles, forming a basis for text-mining tools that will support access to information pertaining to COVID-19. Notably, much of the evidence within these publications is provided in the form of figures. Furthermore, regions where such evidential images occur are rich in information. While biomedical text-based mining tools are being quickly developed and offered for accessing this dataset, images, which contain key clinical and biological information, are not considered. Even outside the COVID-19 realm, little has been done so far to utilize images within publications, despite the fact that they provide important cues about the relevance of the information embedded in articles. Our premise, which is supported by our own and by other informaticians and clinicians experience, is that information derived from images can (and should) be directly incorporated into the biomedical – and specifically into the COVID-19 – document retrieval and extraction. Doing so will improve accurate access to relevant articles, while pin-pointing significant evidence within them, and expediting access to much-needed critical information. The work on this project will result in methods and tools that take advantage of both image- and text-data, facilitating more effective and focused retrieval and mining, thus better supporting speedy data- intensive discovery in the context of COVID-19.