144
Views
0
CrossRef citations to date
0
Altmetric
ORIGINAL RESEARCH

Explainable Deep Learning Model for Predicting Serious Adverse Events in Hospitalized Geriatric Patients Within 72 Hours

, ORCID Icon, , , , , & show all
Pages 1051-1063 | Received 29 Feb 2024, Accepted 04 Jun 2024, Published online: 12 Jun 2024

Figures & data

Figure 1 Patient inclusion flow chart and the development of the deep learning model.

Figure 1 Patient inclusion flow chart and the development of the deep learning model.

Table 1 Characteristics of Patients in Development and Test Data Set

Table 2 Clinical Features Associated with Adverse Event During Admission

Figure 2 The Receiver Operating Characteristic (ROC) curve of model prediction.

Figure 2 The Receiver Operating Characteristic (ROC) curve of model prediction.

Figure 3 Global Explanation of feature importance by SHapley Additive exPlanations (SHAP) value. Sum_gcs_e: The sum of the Glasgow Coma Scale scores when the patient left the emergency department to ward admission. Shock_index_e: Shock index (dividing the heart rate by the systolic blood pressure) when the patient left the emergency department to ward admission.

Abbreviations: Na, sodium level; DM, diabetes mellitus; DBP_e, Diastolic blood pressure when the patient left the emergency department to ward admission; Hb, Hemoglobin; WBC, White blood cells; RR_a, Respiratory rate during emergency department triage; K, potassium level.
Figure 3 Global Explanation of feature importance by SHapley Additive exPlanations (SHAP) value. Sum_gcs_e: The sum of the Glasgow Coma Scale scores when the patient left the emergency department to ward admission. Shock_index_e: Shock index (dividing the heart rate by the systolic blood pressure) when the patient left the emergency department to ward admission.

Data Sharing Statement

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.