Overview

Current Machine learning (ML) methods can achieve high levels of performance on clinical data that matches or even exceeds human clinicians, e.g., by leveraging the power of deep neural networks and large data repositories. However, most existing methods are limited: they are either black-box models that are often incomprehensible for physicians and thus deemed too complex or not trustworthy, they show limited robustness and addressing the gap between the ML community and the clinical community remains a challenge for implementing the methods in clinical practice. As a result, the majority of ML research in the medical domain is mostly stalled at the research paper level and remains difficult to act on. To achieve the overarching goal of realizing the promise of cutting-edge ML techniques and bring this exciting research to fruition, we must bridge the gap between research and clinics. In this workshop, we aim to bring together ML researchers and clinicians to discuss the challenges and potential solutions on how to enable the use of state-of-the-art ML techniques in the daily clinical practice and ultimately improve healthcare.

We are specifically interested (but are not limited to) the following areas:

• Procedures that bring humans-in-the-loop for auditing ML healthcare systems to improve human performance, machine performance, or both.

• Methods that are robust to changes in population, distribution shifts, or other types of biases.

• Properties of ML methods/systems to be fulfilled to successfully deploy them in the clinic where the feasibility of these properties should also be taken into account.

• Analyses of how to assess the failure modes of ML models for healthcare and reduce over-reliance.

• Developing methods for improved interpretability of ML predictions in the context of healthcare.

• Translational and implementational aspects: challenges and lessons learned from integrating an ML system into clinical workflow.

The workshop will feature three main sessions, each with two invited talks (from ML and clinical experts), followed by a moderated Q&A with a specific focus and an informal round-table discussion that encourages a broader conversation and open discussions between the audience and the speakers of varying backgrounds. With these interactive sessions, we want to encourage productive exchange between machine learning researchers and clinical experts. There will be two spotlight presentations and poster sessions via gather.town for contributed submissions. One session will be in the middle of the day, and one at the end of the day to enable participation from different time zones.

Sponsors