Production lines in pharmaceutical manufacturing generate numerous heterogeneous data sets from various embedded systems which control the multiple processes of medicine production. Such data sets should arguably ensure end-to-end traceability and data integrity in order to release a medicine batch, which is uniquely identified and tracked by its batch number/code. Consequently, auditable computerised systems are crucial on pharmaceutical production lines, since the industry is becoming increasingly regulated for product quality and patient health purposes. This paper describes the EU-funded SPuMoNI project, which aims to ensure the quality of large amounts of data produced by computerised production systems in representative pharmaceutical environments. Our initial results include significant progress in: (i) end-to-end verification taking advantage of blockchain properties and smart contracts to ensure data authenticity, transparency, and immutability; (ii) data quality assessment models to identify data behavioural patterns that can violate industry practices and/or international regulations; and (iii) intelligent agents to collect and manipulate data as well as perform smart decisions. By analysing multiple sensors in medicine production lines, manufacturing work centres, and quality control laboratories, our approach has been initially evaluated using representative industry-grade pharmaceutical manufacturing data sets generated at an IT environment with regulated processes inspected by regulatory and government agencies.

药品制造中的生产线从各种嵌入式系统生成大量异构数据集，这些数据集控制着药品生产的多个过程。这样的数据集应该可以确保端到端的可追溯性和数据完整性，以便发布药品批次，该药品批次通过其批次号/代码进行唯一标识和跟踪。因此，可审计的计算机系统对于制药生产线至关重要，因为该行业正变得越来越受产品质量和患者健康目的的监管。本文介绍了欧盟资助的SPuMoNI项目，该项目旨在确保在代表性制药环境中计算机化生产系统产生的大量数据的质量。我们的初步结果包括：（i）利用区块链属性和智能合约的端到端验证，以确保数据的真实性，透明性和不变性；（ii）数据质量评估模型，以识别可能违反行业惯例和/或国际法规的数据行为模式；（iii）智能代理收集和处理数据以及执行智能决策。通过分析药品生产线，制造工作中心和质量控制实验室中的多个传感器，我们的方法已通过在IT环境中生成的具有代表性的工业级制药制造数据集进行了初步评估，该数据集由监管机构和政府机构进行了检查。