PT  - JOURNAL ARTICLE
AU  - Manzano, Toni
AU  - Fernandez, Cristina
AU  - Ruiz, Toni
AU  - Richard, Hugo
TI  - AI Algorithm Qualification
AID  - 10.5731/pdajpst.2019.011338
DP  - 2020 Jan 01
TA  - PDA Journal of Pharmaceutical Science and Technology
PG  - pdajpst.2019.011338
4099  - http://journal.pda.org/content/early/2020/08/14/pdajpst.2019.011338.short
4100  - http://journal.pda.org/content/early/2020/08/14/pdajpst.2019.011338.full
AB  - Quality is defined by the American Society for Quality (ASQ) as ″the totality of features and characteristics of a product or service that bears on its ability to satisfy given needs″. Therefore, quality is applicable to processes that supply outcomes which values can be measured. The statistical control is an effective methodology which provides the outcome of quality of goods, bringing an added value that other methods like the quality by inspection do not bring about. The statistical methods applied to process control have been thoroughly developed and the mathematics that supports them have been broadly demonstrated. Artificial Intelligence is a field where mathematics, statistics and programming play a joint role and its results could also be applied to disciplines like quality control. Nevertheless, its utilization is subordinated to the qualification of implemented algorithms. This research presents a standard procedure to qualify Artificial Intelligence algorithms, allowing their usage in regulated environments to grant the quality of the delivered products or services (e.g., in drugs and medicines manufacturing). The regulated principles are defined by the concept of Quality by Design (QbD), which is a notion introduced in the pharmaceutical industry as a good practice for process management under multivariate analysis. This study is intended to provide a guidance to qualify AI algorithms taking the QbD guidelines as foundation for this purpose.