Anzeige
Erschienen in:
31.10.2023 | Correspondence
SHAP Model Explainability in ECMO – PAL mortality prediction: A Critical Analysis. Author’s reply
Erschienen in: Intensive Care Medicine | Ausgabe 12/2023
Einloggen, um Zugang zu erhaltenExcerpt
We thank Valiente Fernández and colleagues for their comments on our work [1, 2]. We agree that the figure presented in our work is quite complex, but hopefully, something readers will get used to seeing in the near future as artificial intelligence (AI) becomes more prominent and transparent in medicine. Fernández and colleagues correctly interpret Fig. 1 of our original paper, indicating that positive values on the x-axis contribute to survival, and negative values contribute to mortality. Fernández and colleagues also point out that the primary outcome for this model is mortality, so the x-axis should be inverted (positive values driving mortality). We believe that intuitively, positive values should represent survival, and negative represent non-survival. However, we agree that this could be misconstrued from the text.
Fig. 1
Shapley additive explainer (SHAP) values show the influence of each variable on the model output. The colour of each dot represents the value of an individual patient data point, with pink being the maximum variable value and blue being the minimum variable value. The dot’s position on the x-axis represents that data point’s contribution to a patient-specific outcome in combination with all other variables for that patient. Positive values on the x-axis drive survival, while negative values drive mortality. The width of the violin shows the distribution. Variables are ranked in descending order of predictive importance
× ![]()
…