Outlining the Design Space of Explainable Intelligent Systems for\n Medical Diagnosis

Abstract

The adoption of intelligent systems creates opportunities as well as\nchallenges for medical work. On the positive side, intelligent systems have the\npotential to compute complex data from patients and generate automated\ndiagnosis recommendations for doctors. However, medical professionals often\nperceive such systems as black boxes and, therefore, feel concerned about\nrelying on system generated results to make decisions. In this paper, we\ncontribute to the ongoing discussion of explainable artificial intelligence\n(XAI) by exploring the concept of explanation from a human-centered\nperspective. We hypothesize that medical professionals would perceive a system\nas explainable if the system was designed to think and act like doctors. We\nreport a preliminary interview study that collected six medical professionals'\nreflection of how they interact with data for diagnosis and treatment purposes.\nOur data reveals when and how doctors prioritize among various types of data as\na central part of their diagnosis process. Based on these findings, we outline\nfuture directions regarding the design of XAI systems in the medical context.\n

Ähnliche Arbeiten