Fig. 4: Example visual representation for a particular patient. | Nature Communications

Fig. 4: Example visual representation for a particular patient.

From: Personalising intravenous to oral antibiotic switch decision making through fair interpretable machine learning

Fig. 4

`Traffic light' suggestions are initially displayed in a temporal manner as the patient progresses. If required clinicians can obtain more information for any given day of the patient’s stay. Simple textual descriptions are provided alongside detailed tabular graphics to maximise clarity. The text quickly gets across key points, while tables show similar example patients and their features, both with their relative importance and contribution respectively. Furthermore, switch labels and predictions across both thresholds are displayed. Note that the patients label would obviously not be available during clinical use, but is shown here to be comprehensive. Finally, the limitations and use cases for the system are clearly labeled. ICU Intensive care unit, IV Intravenous.

Back to article page