Kerrie Matthews CHIA

Kerrie Matthews CHIA

Clinical Midwifery Consultant, Queensland Health

B1 Information science concepts / B2 Design and development / C1 System design concepts / E1 Sociotechnical concepts / E2 Problem solving / E5 User experience

This article explores the concept of fairness relating to artificial intelligence (AI) in healthcare. Fairness is guided by ethical principles, which if not honoured may result in unintentional biased algorithmic decision support, which can be further compounded by biased training data or erroneous assumptions underlying algorithm design.

The authors acknowledge a clash between fairness principles and predefined metrics can lead to conflict between stakeholders and suggest a model for fairness consideration. They argue that ongoing evaluation by ethicists, clinicians, algorithm developers, patients, policy makers and the general public is required to agree on an ethical framework that considers fairness and preferences when applying AI in health care.

The Lancet Digital Health is one of an increasing number of open access journals dedicated to digital health. Check it out here.

References
Naher, A; Krumpal, I, Antao, E; Ong, E;  Rojo, M; Kaggwa, F; Balzer, F; Celi, A; Braune, K;  Wieler, L; Agha-Mir-Salim, L: ‘Measuring fairness preferences is important for artificial intelligence in health care’, The Lancet Digital Health, published May 2024, Available at: Measuring fairness preferences is important for artificial intelligence in health care – The Lancet Digital Health