Controlling Safety of Artificial Intelligence-Based Systems in Healthcare

• DOI: 10.3390/sym13010102
• Publication Date: 2021-01-11T04:03:42Z
• Authors: Mohammad Reza Dav...; Waldemar Karwowski; Krzysztof Fiok; Thomas Wan; Hamid R. Parsaei
• Keywords: Emergency plan

Artificial intelligence (AI)-based systems have achieved significant success in healthcare since 2016, and AI models accomplished medical tasks, at or above the performance levels of humans. Despite these achievements, various challenges exist application healthcare. One main is safety, which related to unsafe incorrect actions recommendations by algorithms. In response need address safety challenges, this research aimed develop a controlling system (SCS) framework reduce risk potential healthcare-related incidents. The was developed adopting multi-attribute value model approach (MAVT), comprises four symmetrical parts: extracting attributes, generating weights for developing rating scale, finalizing system. represents set attributes different layers can be used as checklist institutions with implemented models. Having will lead high scores SCS, indicates safe proposed provides basis implementing monitoring legislation, identifying risks models’ activities, improving human-AI interactions, preventing incidents from occurring, having an emergency plan remaining risks.