A multiparameter model predicting in-hospital mortality in malignant cerebral infarction

Interpretable machine learning models for predicting in-hospital death in patients in the intensive care unit with cerebral infarction

BACKGROUND

Research on patients with cerebral infarction in the Intensive Care Unit (ICU) is still lacking. Our study aims to develop and validate multiple machine-learning (ML) models using two large ICU databases-Medical Information Mart for Intensive Care version III (MIMIC-III) and eICU Research Institute Database (eRI)-to guide clinical practice.

METHODS

We collected clinical data from patients with cerebral infarction in the MIMIC-III and eRI databases within 24 h of admission. The opinion of neurologists and the Least Absolute Shrinkage and Selection Operator regression was used to screen for relevant clinical features. Using eRI as the training set and MIMIC-III as the test set, we developed and validated six ML models. Based on the results of the model validation, we select the best model and perform the interpretability analysis on it.

RESULTS

A total of 4,338 patients were included in the study (eRI:3002, MIMIC-III:1336), resulting in a total of 18 clinical characteristics through screening. Model validation results showed that random forest (RF) was the best model, with AUC and F1 scores of 0.799 and 0.417 in internal validation and 0.733 and 0.498 in external validation, respectively; moreover, its sensitivity and recall were the highest of the six algorithms for both the internal and external validation. The explanatory analysis of the model showed that the three most important variables in the RF model were Acute Physiology Score-III, Glasgow Coma Scale score, and heart rate, and that the influence of each variable on the judgement of the model was consistent with medical knowledge.

CONCLUSION

Based on a large sample of patients and advanced algorithms, our study bridges the limitations of studies on this area. With our model, physicians can use the admission information of cerebral infarction patients in the ICU to identify high-risk groups among them who are prone to in-hospital death, so that they could be more alert to this group of patients and upgrade medical measures early to minimize the mortality of patients.

Outcomes and resource use of patients with large hemispheric infarction and cerebral edema: analysis of real-world data

OBJECTIVE

Large hemispheric infarction (LHI) is associated with a high likelihood of the evolution of life-threatening edema. Few studies have assessed real-world clinical outcomes and management strategies among patients with LHI. The objective of this study was to describe the management, in-hospital outcomes, and direct healthcare resource burden of patients with LHI, as well as those of patients with subsequent cerebral edema.

METHODS

This observational, retrospective cohort study analyzed de-identified data from US adult patients using the IBM MarketScan Hospital Drug Database (Q4-2015 to Q4-2017). Patients were included in the "Possible LHI" or the "Other Ischemic Strokes" cohorts using ICD-10 diagnosis codes. Patients with possible LHI were further categorized into "LHI with Edema" and "LHI without Edema" subgroups using diagnosis and procedure codes. Select clinical and economic outcomes were compared between cohorts and subgroups using multivariable regressions.

RESULTS

Of 79,201 eligible encounters with ischemic strokes, 11,772 unique patients were assigned to the Possible LHI cohort while 67,429 were assigned to the Other Ischemic Strokes cohort. Among patients with possible LHI, 869 (7%) were assigned to the LHI with Edema subgroup and 10,903 (93%) were assigned to the LHI without Edema subgroup. Patients in the Possible LHI cohort had longer hospital stays (mean difference [MD] [95%CI] = 2.6 [2.4;2.8] days), higher total facility charges (MD [95%CI] = $28,656 [26,794;30,524]), and higher odds of death (odds ratio [95%CI] = 2.2 [2.0;2.4]) than the Other Ischemic Strokes cohort. Among patients with possible LHI, the incremental clinical and resource burden was further exacerbated in the subgroup of patients with edema (hospital days: MD [95%CI] = 5.0 [3.9;6.2] days; total facility charges: MD [95%CI] = $59,585 [50,816;67,583]; mortality: odds ratio [95%CI] = 10.3 [8.5;12.4]).

CONCLUSIONS

Among patients with ischemic strokes, LHI was associated with increased clinical management and direct healthcare resource burden in real-world hospital settings. The burden was substantially increased among patients who developed cerebral edema.