S. Tonekaboni, S. Joshi, M.D. McCradden & A. Goldenberg write in their publication: “What Clinicians Want: Contextualizing Explainable Machine Learning for Clinical End Use.” about the evolution of applied machine learning in clinical settings.
They explain how well adopted rule based assistive tools are currently being advanced by machine learning (ML) based tools. ML tools however don’t necessarily provide the transparency that was obtainable from their predecessors. The adoption of the ML tools is challenged by the desire to validate the process and the outcomes by the clinician or end user. This has prompted the trend of developing more interpretable ML tools. Bringing the black box to light so to say.
The ability to provide this insight is predicted to drive the acceptability of the models and will determine if algorithms will survive in the real world.
The researchers propose strategies for enhancing buy in and trust and executed an interesting experiment in the ICU and Emergency Department. You can read the publication by clicking the image below.
