FAIR Data Competency and Machine Learning Readiness for Biomedical Scientists

项目摘要/摘要 大学博士后学术科学教育和研究培训(ASERT)计划 新墨西哥大学(UNM)将博士毕业生与杰出的生物医学研究导师和 通过与少数族裔的有效科学教育合作，让新墨西哥州的不同群体参与进来 服务机构(MSI)和国家部落学院合作伙伴。ASERT的总体目标是促进 全国各地受过培训并具有在各方面取得成功的经验的初级教职员工数量 在可能有高度多样化的学生群体的机构中的学术生活 跨学科、协作研究是取得成功的关键。本补编要求将会扩大 ASERT研究员和感兴趣的准成员在可查找、可访问、 使用机器学习进行可互操作和可重复使用(公平)的数据管理和数据分析。这个 拟议的培训是对生物医学科学研究人员的关键培训需求，因为 数据集的复杂性，需要改进数据管理以确保数据可用性，并支持新的 通过有效利用现有数据和方便地链接到新数据来发现数据。拥有广泛的教职员工 公平数据管理、机器学习和生物统计学方面的专业知识将开发和实施培训 由八个模块组成的计划，包含以下主要原则：1)获取数据/数据 管理和处理；2)模型培训；3)模型评估；4)模型改进。有很强的 为补充期后的培训方案提供机构支持，包括 更深入地研究现代AI/ML技术，并根据FIRE管理不同的数据集 先进的、有指导的生物医学数据科学课程提供的标准。研究员将分享信息 ASERT IRACDA与少数民族合作服务本科生在数据科学方面的机会 机构。在IRACDA全国会议上发表和介绍将有助于进一步传播 共享培训资源。

Project Summary/Abstract The Academic Science Education and Research Training (ASERT) program for postdoctoral scholars at the University of New Mexico (UNM) matches PhD graduates with outstanding biomedical research mentors and engages New Mexico’s diverse populations through effective science education collaborations with minority serving institution (MSI) and national tribal college partners. The overall objective of ASERT is to contribute to the number of junior faculty across the country with the training and experience to succeed in all facets of academic life at institutions that are likely to have highly diverse student populations and where interdisciplinary, collaborative research is essential for success. The present supplement request will expand the professional development of ASERT fellows and interested associate members in Findable, Accessible, Interoperable and Reusable (FAIR) data management and data analyses using machine learning. The proposed training is a critical training need for investigators in the biomedical sciences due to the increasing complexity of the datasets, a need for improved data stewardship to ensure data usability, and to enable new discoveries through effective use of existing data and facility in linking to new data. Faculty with extensive expertise in FAIR data management, machine learning, and biostatistics will develop and implement a training program consisting of eight modules that encompass the overarching principles of: 1) Obtaining data / Data management and processing; 2) model training; 3) model evaluation; 4) model improvement. There is strong institutional support for the training program beyond the supplement period, including opportunities for a deeper dive into modern AI/ML techniques, and managing diverse data sets in accordance with FAIR standards afforded by an advanced, mentored Biomedical Data Science course. Fellows will share information on opportunities in data science with undergraduate students at ASERT IRACDA partnered Minority Serving Institutions. Publication and presentation at the National IRACDA Conference will serve to further disseminate and share training resources.