人工智能囊胚形态评估数据集构建与质控专家共识
Expert consensus on the construction and quality control of datasets for artificial intelligence(AI)assisted blastocyst morphometric assessment
王浩 1张孝东 1孙莹璞 1孙海翔 1邓成艳 1黄学锋 1刘平 1周灿权 1冯云 1郝桂敏 1卢文红 1沈浣 1师娟子 1张松英 1滕晓明 1王晓红 1王秀霞 1伍琼芳 1全松 1曾勇 1钟影 1邵小光 1柯林楠 1毛歆 1韩倩倩 1黄国宁1
作者信息
- 1. 中华医学会生殖医学分会第五届委员会;中国食品药品检定研究院
- 折叠
摘要
囊胚形态人工智能(AI)评估是AI医疗器械发展的新兴方向,也是AI在辅助生殖领域的重要应用.AI在新领域应用的起步阶段,数据集的构建与质控对产品质量有重要影响.目前,囊胚形态学AI评估在数据采集、标注、质控等方面尚未形成统一的规范.在参考AI医疗器械、辅助生殖医疗器械现有国家行业标准的基础上,本文以囊胚形态AI评估数据集为主题,对数据集构建与质控要求进行了探讨,对数据集质量特性进行了解析,旨在指导数据集制造责任方加强数据集全生命周期管理,更好地为产品研发、测试、临床试验等环节提供质量保障,助力产业发展.
Abstract
Computer assisted assessment of blastocyst morphology is an emerging direction in the artificial intelligence(AI)medical devices and an important application of AI in the field of assisted reproduction.In the initial stage of the application of AI in new fields,the construction and quality control of data sets have an important impact on product quality.At present,AI-assisted blastocyst morphology assessment has not yet formed a unified specification in terms of data collection,labeling,and quality control.Based on the existing national industry standards for AI medical devices and assisted reproduction medical devices,this paper discusses the requirements for data set construction and quality control and analyzes the quality characteristics of data sets with the theme of blastocyst morphology assessment datasets,with the aim of guiding data set manufacturers to strengthen the management of datasets in the whole life cycle,and to better provide quality assurance for the product research and development,testing,and clinical trials in order to help the development of the industry.
关键词
人工智能(AI)/囊胚形态评估/数据集构建/数据集标注/数据集质量控制Key words
Artificial intelligence(AI)/Blastocyst morphology assessment/Data set construction/Data set annotation/Data set quality control引用本文复制引用
基金项目
重庆市技术创新与应用发展专项项目(CSTB2022TIAD-KPX0146)
国家自然科学基金面上项目(82371728)
出版年
2024
国家科技期刊平台
NSTL
万方数据