Clinical Evaluation Framework Using Behavioural &amp;amp; Visual Attention Read-Outs for Explainable AI (XAI)

Nagendran, Myura; Festor, Paul; Komorowski, Matthieu; Gordon, Anthony; Faisal, A. Aldo

doi:10.2139/ssrn.4496127

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The most important estimand during evaluation studies is the performance of the augmented clinical workflow as a whole, rather than the isolated system's technical performance (49). Augmented workflows can cause unanticipated harm (50) and will only be effective if system recommendations are trusted and actioned by the end user; therefore, a thorough evaluation of safety (51), interpretability (52), and acceptability ( 53) is essential. The unforeseen poor performance of current sepsis decision support systems (54) emphasizes the importance of post-deployment surveillance (55), including surveillance of how system predictions are operationalized, which may change after initial trials.…”

Section: Barriers To Translationmentioning

confidence: 99%

Can Machine Learning Personalize Cardiovascular Therapy in Sepsis?

Catling,

Nagendran,

Festor

et al. 2024

Critical Care Explorations

Self Cite

View full text Add to dashboard Cite

Large randomized trials in sepsis have generally failed to find effective novel treatments. This is increasingly attributed to patient heterogeneity, including heterogeneous cardiovascular changes in septic shock. We discuss the potential for machine learning systems to personalize cardiovascular resuscitation in sepsis. While the literature is replete with proofs of concept, the technological readiness of current systems is low, with a paucity of clinical trials and proven patient benefit. Systems may be vulnerable to confounding and poor generalization to new patient populations or contemporary patterns of care. Typical electronic health records do not capture rich enough data, at sufficient temporal resolution, to produce systems that make actionable treatment suggestions. To resolve these issues, we recommend a simultaneous focus on technical challenges and removing barriers to translation. This will involve improving data quality, adopting causally grounded models, prioritizing safety assessment and integration into healthcare workflows, conducting randomized clinical trials and aligning with regulatory requirements.

show abstract