【摘要】 目的 了解潜在的人工智能(Artificial intelligence, AI)偏倚如何影响基层卫生医疗结果，以及如何识别和减轻医疗结果不平等现象，推进基层医疗发展。方法 结合基层医疗现状，全面调研AI算法模型全生命周期(Whole lifetime, WLT)中的偏倚以及影响基层医疗发展的制约因素，并提出相应的解决对策思考。结果 AI可能从研究方向提出、数据采集、变量选择、设计开发到临床实施的WLT制约基层医院医疗发展。医院管理者可采取相应的对策，包括加强自身学习、顶层指导，加强院内病历质量考核，加强基层医院信息化建设，加强数据监管、质控和标准化，制定人才培养引进政策等，减轻不平等现象，实现AI赋能基层医疗水平提升的目的。结论 AI算法模型在其WLT中的偏倚是不容忽视的，解决偏倚对于促进基层医院医疗决策的公平性至关重要。在公共卫生领域，AI对医疗结果公平伦理挑战的监管框架尚未得到充分解决，需要进一步关注。更多还原

【Abstract】 Objective To explore how potential artificial intelligence(AI) bias affects primary health care, and how inequities can be identified and mitigated to ensure primary healthcare development. Methods In combination with the current situation of primary health services, we investigated the bias in the whole lifetime(WLT) of AI algorithm model comprehensively and the constraint factors for the development, and put forward countermeasures. Results AI may exacerbate challenges to healthcare development in primary hospitals throughout the WLT from the proposal of research directions, data collection, variable selection, design and development to clinical practice. Hospital administrators could take countermeasures to reduce inequalities and realize AI-enabled primary care improvement, including strengthening their own learning capacity and top-level guidance, enhancing medical records quality assessment, enhancing informalization construction in primary hospitals, strengthening data supervision, quality control and standardization, and develop policies to introduce and cultivate talents. Conclusions The bias of AI algorithmic models throughout WLT cannot be ignored. The potential bias need to be addressed to promote equity in medical decision making in primary hospitals. In the field of public health, the regulatory framework for the ethical challenges to the equity in medical outcomes by AI has not been adequately addressed, which requires further attention.更多还原