Biases in Healthcare AI and Their Impact on Patient Empowerment: A Scoping Review

Artificial Intelligence (AI) is increasingly applied in healthcare and is often linked to patient empowerment. However, biases in data, algorithms, and design may hinder empowerment by reinforcing inequalities and limiting autonomy.

This scoping review examines how bias in healthcare AI impacts patient empowerment.

We searched PubMed and multiple databases via EBSCOHost. Studies were included if they addressed bias in healthcare AI in connection with patients, focusing on empowerment, participation or related concepts. AI with patient-centered focus was defined as AI applications directly involving patients (e.g., chatbots, decision aids). After screening 488 records, 13 studies met the inclusion criteria.

Various social, structural, and technical biases were reported. While patient involvement in AI applications was addressed (e.g., engagement in remote monitoring or chatbot use), explicit reference to empowerment, understood as autonomy, control, and decision-making, was rare. Most studies equated empowerment with patient engagement.

Bias in healthcare AI can limit the empowering potential of digital health tools, and research explicitly linking bias to empowerment outcomes is still at an early stage.