RESUMEN
Despite advancements in artificial intelligence-based decision-making, transitioning patients from intensive care units (ICUs) to low-acuity wards is challenging, especially in resource-limited settings. This study aimed to develop a simple scoring system to predict ICU discharge safety. We retrospectively analyzed patients admitted to a tertiary hospital's medical ICU (MICU) between July 2016 and December 2021. This period was divided into two phases for model development and validation. We identified risk factors associated with unexpected death within 14 days of MICU discharge and developed a predictive scoring system that incorporated these factors. We verified the system's performance using validation data. In the development cohort, 522 patients were discharged from the MICU, and 42 (8.04%) died unexpectedly. In multivariate analysis, the Sequential Organ Failure Assessment (SOFA) score (odds ratio [OR] 1.26, 95% confidence interval [CI] 1.13-1.41), red blood cell distribution width (RDW) (OR 1.20, 95% CI 1.07-1.36), and albumin (OR 0.37, 95% CI 0.16-0.84) were predictors of unexpected death. Each variable was assigned a weighted point in the scoring system, and the area under the curve (AUC) was 0.788 (95% CI 0.714-0.855). The scoring system was performed using an AUC of 0.738 (95% CI 0.653-0.822) in the validation cohort of 343 patients with 9.62% of unexpected deaths. When a cut-off of 0.032 was applied, a sensitivity and a specificity of 81.8% and 55.2%, respectively, were achieved. This simple bedside predictive score for ICU discharge uses the SOFA score, albumin level, and RDW to aid in timely decision-making and optimize critical care facility allocation in resource-limited settings.
RESUMEN
Invasive pulmonary aspergillosis (IPA) can occur in immunocompromised patients, and an early detection and intensive treatment are crucial. We sought to determine the potential of Aspergillus galactomannan antigen titer (AGT) in serum and bronchoalveolar lavage fluid (BALF) and serum titers of beta-D-glucan (BDG) to predict IPA in lung transplantation recipients, as opposed to pneumonia unrelated to IPA. We retrospectively reviewed the medical records of 192 lung transplant recipients. Overall, 26 recipients had been diagnosed with proven IPA, 40 recipients with probable IPA, and 75 recipients with pneumonia unrelated to IPA. We analyzed AGT levels in IPA and non-IPA pneumonia patients and used ROC curves to determine the diagnostic cutoff value. The Serum AGT cutoff value was 0.560 (index level), with a sensitivity of 50%, specificity of 91%, and AUC of 0.724, and the BALF AGT cutoff value was 0.600, with a sensitivity of 85%, specificity of 85%, and AUC of 0.895. Revised EORTC suggests a diagnostic cutoff value of 1.0 in both serum and BALF AGT when IPA is highly suspicious. In our group, serum AGT of 1.0 showed a sensitivity of 27% and a specificity of 97%, and BALF AGT of 1.0 showed a sensitivity of 60% and a specificity of 95%. The result suggested that a lower cutoff could be beneficial in the lung transplant group. In multivariable analysis, serum and BALF AGT, with a minimal correlation between the two, showed a correlation with a history of diabetes mellitus.
RESUMEN
BACKGROUND: Securing the representativeness of study populations is crucial in biomedical research to ensure high generalizability. In this regard, using multi-institutional data have advantages in medicine. However, combining data physically is difficult as the confidential nature of biomedical data causes privacy issues. Therefore, a methodological approach is necessary when using multi-institution medical data for research to develop a model without sharing data between institutions. OBJECTIVE: This study aims to develop a weight-based integrated predictive model of multi-institutional data, which does not require iterative communication between institutions, to improve average predictive performance by increasing the generalizability of the model under privacy-preserving conditions without sharing patient-level data. METHODS: The weight-based integrated model generates a weight for each institutional model and builds an integrated model for multi-institutional data based on these weights. We performed 3 simulations to show the weight characteristics and to determine the number of repetitions of the weight required to obtain stable values. We also conducted an experiment using real multi-institutional data to verify the developed weight-based integrated model. We selected 10 hospitals (2845 intensive care unit [ICU] stays in total) from the electronic intensive care unit Collaborative Research Database to predict ICU mortality with 11 features. To evaluate the validity of our model, compared with a centralized model, which was developed by combining all the data of 10 hospitals, we used proportional overlap (ie, 0.5 or less indicates a significant difference at a level of .05; and 2 indicates 2 CIs overlapping completely). Standard and firth logistic regression models were applied for the 2 simulations and the experiment. RESULTS: The results of these simulations indicate that the weight of each institution is determined by 2 factors (ie, the data size of each institution and how well each institutional model fits into the overall institutional data) and that repeatedly generating 200 weights is necessary per institution. In the experiment, the estimated area under the receiver operating characteristic curve (AUC) and 95% CIs were 81.36% (79.37%-83.36%) and 81.95% (80.03%-83.87%) in the centralized model and weight-based integrated model, respectively. The proportional overlap of the CIs for AUC in both the weight-based integrated model and the centralized model was approximately 1.70, and that of overlap of the 11 estimated odds ratios was over 1, except for 1 case. CONCLUSIONS: In the experiment where real multi-institutional data were used, our model showed similar results to the centralized model without iterative communication between institutions. In addition, our weight-based integrated model provided a weighted average model by integrating 10 models overfitted or underfitted, compared with the centralized model. The proposed weight-based integrated model is expected to provide an efficient distributed research approach as it increases the generalizability of the model and does not require iterative communication.