Abstract:

As one of the three core elements of AI， data quality has increasingly become a focal point in AI research. Technical defects in data collection and other processes often result in dirty data in the dataset， necessitating the application of big data cleansing techniques to mitigate risks such as biases in AI algorithms. However， non-compliant big data cleansing behaviors may have a negative impact on data quality. Focusing on AI medical devices， an interdisciplinary approach is employed to examine the normative challenges and practical dilemmas of big data cleansing from both technical and legal perspectives. Furthermore， comparative methods are applied， drawing on legal practices such as the European Union’s Medical Device Regulation （MDR） and Artificial Intelligence Act， while taking into account China’s national context. Such pathways for regulatory innovation are proposed， including promoting the integration and interaction between technical standards for dataset quality and legal regulations， clarifying obligations for big data cleansing within the framework of AI legislation， and improving the intervention and adaptation of product liability systems.

Key words: artificial intelligence（AI）, medical device, data quality, big data cleansing, legal regulation