Personalizing Medication Recommendation with a Graph-Based Approach

The broad adoption of electronic health records (EHRs) has led to vast amounts of data being accumulated on a patient’s history, diagnosis, prescriptions, and lab tests. Advances in recommender technologies have the potential to utilize this information to help doctors personalize the prescribed medications. However, existing medication recommendation systems have yet to make use of all these information sources in a seamless manner, and they do not provide a justification on why a particular medication is recommended. In this work, we design a two-stage personalized medication recommender system called PREMIER that incorporates information from the EHR. We utilize the various weights in the system to compute the contributions from the information sources for the recommended medications. Our system models the drug interaction from an external drug database and the drug co-occurrence from the EHR as graphs. Experiment results on MIMIC-III and a proprietary outpatient dataset show that PREMIER outperforms state-of-the-art medication recommendation systems while achieving the best tradeoff between accuracy and drug-drug interaction. Case studies demonstrate that the justifications provided by PREMIER are appropriate and aligned to clinical practices.

Effects of Transthoracic Echocardiography on the Prognosis of Patients with Acute Respiratory Distress Syndrome: A Propensity Score Matched Analysis of the MIMIC-III Database

Introduction: Acute respiratory distress syndrome (ARDS) has high mortality and is mainly related to the circulatory failure.Therefore, real-time monitoring of cardiac function and structural changes has important clinical significance.Transthoracic echocardiography (TTE) is a simple and noninvasive real-time cardiac examination which is widely used in intensive care unit (ICU) patients.The purpose of this study was to analyze the effect of TTE on the prognosis of ICU patients with ARDS.Methods: The data of ARDS patients were retrieved from the MIMIC-III v1.4 database and patients were divided into the TTE group and non-TTE group. Then, the baseline data were compared between the two groups, and the effect of TTE on the prognosis of ARDS patients was analyzed through multivariate logistic analysis and the propensity score (PS).Results: A total of 1,346 ARDS patients were enrolled, including 519 (38.6%) cases in the TTE group and 827 (61.4%) cases in the non-TTE group. Compared with the non-TTE group, the 28-day mortality of patients in the TTE group was greatly improved (OR=0.64 95%CI: 0.48-0.86, P=0.003). The length of ICU stay in the TTE group was significantly shorter than that in the non-TTE group (17d vs.14d, P=0.0001). The infusion volume in the TTE group was significantly less than that of the non-TTE group (6.2L vs.5.5L on day 1, P=0.0012). Importantly, the patients in the TTE group were weaned ventilators earlier than those in the non-TTE group (ventilator-free days within 28 d: 21 d vs. 19.8 d, respectively, P = 0.071).Conclusion: TTE can lower the risk of 28-d mortality in patients with ARDS.

Favorable Outcomes of Anticoagulation With Unfractioned Heparin in Sepsis-Induced Coagulopathy: A Retrospective Analysis of MIMIC-III Database

Backgrounds: Anticoagulation in sepsis-associated disseminated intravascular coagulation (DIC) remains uncertain. The aim of this study was to investigate whether unfractioned heparin (UFH) could improve clinical outcomes in patients with sepsis-induced coagulopathy (SIC).Methods: Septic patients with SIC were identified from the Medical Information Mart for Intensive Care (MIMIC)-III database. Cox-proportional hazards model, logistic regression model and linear regression were used to assess the associations between UFH administration and 28-day mortality, hospital mortality, occurrence of bleeding complications and length of stay, respectively. Propensity score matching (PSM) analysis was used to match the imbalance between patients in the UFH group and the control group. Patients were further stratified according to SIC score and Simplified Acute Physiology Score II (SAPS II).Results: A total of 1,820 septic patients with SIC were included in the data analysis. After PSM, 652 pairs of patients were matched between the patients in the UFH group and the control group. UFH was significantly associated with reduced 28-day mortality (HR, 0.323, 95% CI, 0.258–0.406; p < 0.001) and hospital mortality (HR, 0.380, 95% CI, 0.307–0.472; p < 0.001) without increasing the risks of intracranial hemorrhage (OR, 1.480, 95% CI, 0.955–2.294; p = 0.080) or gastrointestinal bleeding (OR, 1.094, 95% CI, 0.503–2.382; p = 0.820). For subgroup analysis, it didn't change the favorable results of UFH on mortality and UFH didn't increase the risk of hemorrhage in patients with severe disease.Conclusions: The analysis of MIMIC-III database indicated that anticoagulant therapy with UFH may be associated with a survival benefit in patients with SIC.

Development and Internal Validation of a Nomogram to Predict Mortality During the ICU Stay of Thoracic Fracture Patients Without Neurological Compromise: An Analysis of the MIMIC-III Clinical Database

Background: This study aimed to develop and validate a nomogram for predicting mortality in patients with thoracic fractures without neurological compromise and hospitalized in the intensive care unit.Methods: A total of 298 patients from the Medical Information Mart for Intensive Care III (MIMIC-III) database were included in the study, and 35 clinical indicators were collected within 24 h of patient admission. Risk factors were identified using the least absolute shrinkage and selection operator (LASSO) regression. A multivariate logistic regression model was established, and a nomogram was constructed. Internal validation was performed by the 1,000 bootstrap samples; a receiver operating curve (ROC) was plotted, and the area under the curve (AUC), sensitivity, and specificity were calculated. In addition, the calibration of our model was evaluated by the calibration curve and Hosmer-Lemeshow goodness-of-fit test (HL test). A decision curve analysis (DCA) was performed, and the nomogram was compared with scoring systems commonly used during clinical practice to assess the net clinical benefit.Results: Indicators included in the nomogram were age, OASIS score, SAPS II score, respiratory rate, partial thromboplastin time (PTT), cardiac arrhythmias, and fluid-electrolyte disorders. The results showed that our model yielded satisfied diagnostic performance with an AUC value of 0.902 and 0.883 using the training set and on internal validation. The calibration curve and the Hosmer-Lemeshow goodness-of-fit (HL). The HL tests exhibited satisfactory concordance between predicted and actual outcomes (P = 0.648). The DCA showed a superior net clinical benefit of our model over previously reported scoring systems.Conclusion: In summary, we explored the incidence of mortality during the ICU stay of thoracic fracture patients without neurological compromise and developed a prediction model that facilitates clinical decision making. However, external validation will be needed in the future.

Prevalence and Mortality of Hypochloremia Among Patients Suffering From Coronary Artery Disease and Congestive Heart Failure: An Analysis of Patients in CIN-I and MIMIC-III Databases

Background: Hypochloremia is an independent predictor for mortality in patients with coronary artery disease (CAD) but whether the same correlation exists in CAD patients with congestive heart failure (CHF) is unclear.Methods: This is an analysis of data stored in the databases of the CIN-I [a registry of Cardiorenal Improvement (NCT04407936) in China from January 2007 to December 2018] and Medical Information Mart for Intensive Care (MIMIC)-III. CAD patients with CHF were included. The outcome measures were 90-day all-cause mortality (ACM) and long-term ACM.Results: Data from 8,243 CAD patients with CHF were analyzed. We found that 10.2% of the study population had hypochloremia (Cl− <98 mmol/L) in CIN-I (n = 4,762) and 20.1% had hypochloremia in MIMIC-III (n = 3,481). Patients suffering from hypochloremia were, in general, older and had a higher prevalence of comorbidities. After adjustment for confounders, hypochloremia remained a significant predictor of short-term mortality risk [90-day ACM: adjusted hazard ratio (aHR), 1.69; 95% CI, 1.27–2.25; P < 0.001 in CIN-I, and 1.36 (1.17–1.59); P < 0.001 in MIMIC-III]. Hypochloremia was also associated with long-term mortality [aHR, 1.26; 95% CI, 1.06–1.50; P = 0.009 in CIN-I, and 1.48 (1.32–1.66); P < 0.001 in MIMIC-III]. Prespecified subgroup analyses revealed an association of hypochloremia with long-term ACM to be attenuated slightly in the women of the two databases (P interaction < 0.05).Conclusions: Hypochloremia is independently associated with higher short-term and long-term ACM. Further studies are needed to determine if early preventive measurements and active intervention of hypochloremia can reduce the mortality risk of CAD patients with CHF.

Effects of Transthoracic Echocardiography on the Prognosis of Patients with Acute Respiratory Distress Syndrome: a Propensity Score Matched Analysis of the MIMIC-III Database

Introduction: Acute respiratory distress syndrome (ARDS) has high mortality and is mainly related to the circulatory failure.Therefore, real-time monitoring of cardiac function and structural changes has important clinical significance.Transthoracic echocardiography (TTE) is a simple and noninvasive real-time cardiac examination which is widely used in intensive care unit (ICU) patients.The purpose of this study was to analyze the effect of TTE on the prognosis of ICU patients with ARDS.Methods: The data of ARDS patients were retrieved from the MIMIC-III v1.4 database and patients were divided into the TTE group and non-TTE group. Then, the baseline data were compared between the two groups, and the effect of TTE on the prognosis of ARDS patients was analyzed through multivariate logistic analysis and the propensity score (PS).Results: A total of 1,346 ARDS patients were enrolled, including 519 (38.6%) cases in the TTE group and 827 (61.4%) cases in the non-TTE group. Compared with the non-TTE group, the 28-day mortality of patients in the TTE group was greatly improved (OR=0.64 95%CI: 0.48-0.86, P=0.003). The length of ICU stay in the TTE group was significantly shorter than that in the non-TTE group (17d vs.14d, P=0.0001). The infusion volume in the TTE group was significantly less than that of the non-TTE group (6.2L vs.5.5L on day 1, P=0.0012). Importantly, the patients in the TTE group were weaned ventilators earlier than those in the non-TTE group (ventilator-free days within 28 d: 21 d vs. 19.8 d, respectively, P = 0.071).Conclusion: TTE can lower the risk of 28-d mortality in patients with ARDS.

Serum Lactate/Albumin as a Predictor of In-Hospital Mortality in ICU Pancreatitis Patients: an Analysis of the MIMIC-III Database

Background: Acute pancreatitis (AP) is a common serious illness, and is characterized by rapid deterioration and a high mortality rate. Several biomarkers can evaluate and guide the treatment of acute pancreatitis, but there is currently no consensus on which markers are the most effective, simple, and economical for treating early-onset AP. In this study, we used the MIMIC III database to conduct a retrospective study on the relationship between early lactate/albumin (LAC/ALB), in-hospital mortality, and complication rates in patients with acute pancreatitis in the ICU.Methods: Basic data and indicators of laboratory tests, hospital deaths, and hospitalization days of acute pancreatitis patients were extracted from the database, after which the relationship between LAC/ALB and hospital mortality, ICU hospitalization days, and organ failure were evaluated using a t-test, a rank-sum test, a chi-square test or Fisher's exact probability method, and a Cox proportional hazard model.Results: 894 patients met the requirements and were selected from the MIMIC III database. They were subsequently grouped according to the lower limit ratio of the LAC/ALB normal value of 0.7. The group with LAC/ALB>0.7 showed higher hospital mortality rates, and the Lac, Inr, nitrogen, blood sugar, AKI incidence, Tbil, Sapsii score, and Sofa scores were all higher than the group with LAC/ALB<0.7. A multivariate Cox regression analysis model was used to explore the relationship between LAC/ALB levels and inpatient mortality. After including different adjustment variables, we determined that LAC/ALB is a risk factor for in-hospital death. The results of the subgroup analysis of LAC/ALB levels and mortality of hospitalized patients indicate that higher levels of LAC/ALB are risk factors for in-hospital deaths in patients with acute pancreatitis.

Does Diabetes Mellitus Increase the Short- and Long-Term Mortality in Patients With Critical Acute Myocardial Infarction? Results From American MIMIC-III and Chinese CIN Cohorts

BackgroundThe harmful effect of diabetes mellitus (DM) on mortality in patients with acute myocardial infarction (AMI) remains controversial. Furthermore, few studies focused on critical AMI patients. We aimed to address whether DM increases short- and long-term mortality in this specific population.MethodsWe analyzed AMI patients admitted into coronary care unit (CCU) with follow-up of ≥1 year from two cohorts (MIMIC-III, Medical Information Mart for Intensive Care III; CIN, Cardiorenal ImprovemeNt Registry) in the United States and China. Main outcome was mortality at 30-day and 1-year following hospitalization. Kaplan-Meier curves and Cox proportional hazards models were constructed to examine the impact of DM on mortality in critical AMI patients.Results1774 critical AMI patients (mean age 69.3 ± 14.3 years, 46.1% had DM) were included from MIMIC-III and 3380 from the CIN cohort (mean age 62.2 ± 12.2 years, 29.3% had DM). In both cohorts, DM group was older and more prevalent in cardio-renal dysfunction than non-DM group. Controlling for confounders, DM group has a significantly higher 30-day mortality (adjusted odds ratio (aOR) (95% CI): 2.71 (1.99-3.73) in MIMIC-III; aOR (95% CI): 9.89 (5.81-17.87) in CIN), and increased 1-year mortality (adjusted hazard ratio (aHR) (95% CI): 1.91 (1.56-2.35) in MIMIC-III; aHR (95% CI): 2.62(1.99-3.45) in CIN) than non-DM group.ConclusionsTaking into account cardio-renal function, critical AMI patients with DM have a higher 30-day mortality and 1-year mortality than non-DM group in both cohorts. Further studies on prevention and management strategies for DM are needed for this population.Clinical Trial Registrationclinicaltrials.gov, NCT04407936.