Conectar
To develop and validate a machine learning-based risk prediction model for delirium in older inpatients.
A prospective cohort study.
A prospective cohort study was conducted. Eighteen clinical features were prospectively collected from electronic medical records during hospitalisation to inform the model. Four machine learning algorithms were employed to develop and validate risk prediction models. The performance of all models in the training and test sets was evaluated using a combination of the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, Brier score, and other metrics before selecting the best model for SHAP interpretation.
A total of 973 older inpatient data were utilised for model construction and validation. The AUC of four machine learning models in the training and test sets ranged from 0.869 to 0.992; the accuracy ranged from 0.931 to 0.962; and the sensitivity ranged from 0.564 to 0.997. Compared to other models, the Random Forest model exhibited the best overall performance with an AUC of 0.908 (95% CI, 0.848, 0.968), an accuracy of 0.935, a sensitivity of 0.992, and a Brier score of 0.053.
The machine learning model we developed and validated for predicting delirium in older inpatients demonstrated excellent predictive performance. This model has the potential to assist healthcare professionals in early diagnosis and support informed clinical decision-making.
By identifying patients at risk of delirium early, healthcare professionals can implement preventive measures and timely interventions, potentially reducing the incidence and severity of delirium. The model's ability to support informed clinical decision-making can lead to more personalised and effective care strategies, ultimately benefiting both patients and healthcare providers.
This study was reported in accordance with the TRIPOD statement.
No patient or public contribution.
Taking a dimensional view, this study aims to understand, among professional caregivers after patient deaths, the symptom distribution and development of the short-term bereavement reaction (SBR) network and the node-level links between the meaning of patient death (MPD) and the SBR network.
A cross-sectional secondary analysis was conducted with existing data from 220 Chinese urban hospital nurses and physicians who experienced the most recent patient death within a month. MPD was measured by the 10 formative items of the meaning of patient death model, and SBR was measured by the Short-term Bereavement Reactions Subscale of the Professional Bereavement Scale. Both Gaussian graphical network analysis and Bayesian network analysis were applied to the SBR network, and Gaussian graphical network analysis was used to estimate the MPD-SBR network.
Frustrated and guilty are central nodes in the regularized partial correlation SBR network. Meanwhile, a traumatic event and failure at work are important bridge nodes between the MPD network and the SBR network. In the Bayesian SBR network, moved by the family's understanding, moved by the family's gratitude and sad mainly drive other nodes.
After a patient death, nurses' and physicians' SBR networks feature professional-dimension symptoms at their core, while they follow ‘personal to professional’ and ‘concrete to abstract’ symptom development patterns. The personal meaning of a traumatic event and the professional meaning of a failure at work play key roles in bridging the MPD and SBR networks, and meanings of both the personal and the professional dimensions can link to professional-dimension reactions.
The manuscript followed the STROBE checklist for reporting cross-sectional studies.
No patient or public contribution.
FreshRSS 