Developing artificial intelligence (AI) tools for healthcare is a collaborative effort, bringing data scientists, clinicians, patients and other disciplines together. In this paper, we explore the collaborative data practices of research consortia tasked with applying AI tools to understand and manage multiple long-term conditions in the UK. Through an inductive thematic analysis of 13 semi-structured interviews with participants of these consortia, we aimed to understand how collaboration happens based on the tools used, communication processes and settings, as well as the conditions and obstacles for collaborative work. Our findings reveal the adaptation of tools that are used for sharing knowledge and the tailoring of information based on the audience, particularly those from a clinical or patient perspective. Limitations on the ability to do this were also found to be imposed by the use of electronic healthcare records and access to datasets. We identified meetings as the key setting for facilitating exchanges between disciplines and allowing for the blending and creation of knowledge. Finally, we bring to light the conditions needed to facilitate collaboration and discuss how some of the challenges may be navigated in future work.

翻译：开发用于医疗健康的人工智能（AI）工具是一项协作性的工作，需要数据科学家、临床医生、患者等多学科人员的共同参与。本文探索了受委托运用AI工具理解和管理英国多种慢性病的研究联盟中的协作数据实践。通过对这些联盟参与者13次半结构化访谈的归纳主题分析，我们旨在理解基于所用工具、沟通过程和场景以及协作工作条件与障碍的协作方式。研究结果揭示了用于知识共享的工具的适应性调整，以及根据受众（特别是临床或患者视角的受众）定制信息的过程。同时，电子医疗记录的使用和数据集的获取也被发现限制了这种能力。我们确定会议是促进学科间交流、实现知识融合与创新的关键场景。最后，我们揭示了促进协作所需的条件，并探讨了未来工作中克服某些挑战的可能途径。