Responsible Innovation: Exploring the Ethical Implications of AI-Powered Suicide Risk Prediction

Abstract: Artificial Intelligence (AI) may help save lives by proactively identifying patients at high risk for suicide. In a recently published study of 120,398 patients visits over 15 months, a newly created AI model, combined with face-to-face screening, accurately predicted patients who attempted suicide. This panel presentation will focus on ethical issues raised by implementing such models in healthcare. The first panelist is a physician, data scientist, and the lead researcher who developed this AI tool. He will discuss the development and current research using this model. The second panelist is an anthropologist who will focus on the human factors elements that are important to the implementation and ethics of such models and on the importance of value-centered design, which directly aligns to the conference theme: What Does It Mean to be Human? The third panelist is a lawyer, physician, and ethicist who will evaluate the ethical issues related to clinicians' perspectives on the use of such models. Finally, the moderator is an ethicist and clinical social worker who specializes in AI implementation and bioethics. She will facilitate discussion on the following themes: patient privacy, patient autonomy, and ethical concerns regarding bias and fairness, with an emphasis on practical recommendations. The ethics of healthcare systems limitations in mental health care will also be discussed. This panel will be engaging, encouraging attendees to participate and to ask questions. The presentation will end with a final set of recommendations that will be generalizable to other AI models in healthcare.