Learning to predict in-hospital mortality risk in the intensive care unit with attention-based temporal convolution network.

BACKGROUND: Dynamic prediction of patient mortality risk in the ICU with time series data is limited due to high dimensionality, uncertainty in sampling intervals, and other issues. A new deep learning method, temporal convolution network (TCN), makes it possible to deal with complex clinical time series data in ICU. We aimed to develop and validate it to predict mortality risk using time series data from MIMIC III dataset. METHODS: A total of 21,139 records of ICU stays were analysed and 17 physiological variables from the MIMIC III dataset were used to predict mortality risk. Then we compared the model performance of the attention-based TCN with that of traditional artificial intelligence (AI) methods. RESULTS: The area under receiver operating characteristic (AUCROC) and area under precision-recall curve (AUC-PR) of attention-based TCN for predicting the mortality risk 48 h after ICU admission were 0.837 (0.824 -0.850) and 0.454, respectively. The sensitivity and specificity of attention-based TCN were 67.1% and 82.6%, respectively, compared to the traditional AI method, which had a low sensitivity (< 50%). CONCLUSIONS: The attention-based TCN model achieved better performance in the prediction of mortality risk with time series data than traditional AI methods and conventional score-based models. The attention-based TCN mortality risk model has the potential for helping decision-making for critical patients. TRIAL REGISTRATION: Data used for the prediction of mortality risk were extracted from the freely accessible MIMIC III dataset. The project was approved by the Institutional Review Boards of Beth Israel Deaconess Medical Center (Boston, MA) and the Massachusetts Institute of Technology (Cambridge, MA). Requirement for individual patient consent was waived because the project did not impact clinical care and all protected health information was deidentified. The data were accessed via a data use agreement between PhysioNet, a National Institutes of Health-supported data repository (https://www.physionet.org/), and one of us (Yu-wen Chen, Certification Number: 28341490). All methods were carried out in accordance with the institutional guidelines and regulations.

A Simple and Quick Screening Method for Intrapulmonary Vascular Dilation in Cirrhotic Patients Based on Machine Learning.

BACKGROUND AND AIMS: Screening for hepatopulmonary syndrome in cirrhotic patients is limited due to the need to perform contrast enhanced echocardiography (CEE) and arterial blood gas (ABG) analysis. We aimed to develop a simple and quick method to screen for the presence of intrapulmonary vascular dilation (IPVD) using noninvasive and easily available variables with machine learning (ML) algorithms. METHODS: Cirrhotic patients were enrolled from our hospital. All eligible patients underwent CEE, ABG analysis and physical examination. We developed a two-step model based on three ML algorithms, namely, adaptive boosting (termed AdaBoost), gradient boosting decision tree (termed GBDT) and eXtreme gradient boosting (termed Xgboost). Noninvasive variables were input in the first step (the NI model), and for the second step (the NIBG model), a combination of noninvasive variables and ABG results were used. Model performance was determined by the area under the curve of receiver operating characteristics (AUCROCs), precision, recall, F1-score and accuracy. RESULTS: A total of 193 cirrhotic patients were ultimately analyzed. The AUCROCs of the NI and NIBG models were 0.850 (0.738-0.962) and 0.867 (0.760-0.973), respectively, and both had an accuracy of 87.2%. For both negative and positive cases, the recall values of the NI and NIBG models were both 0.867 (0.760-0.973) and 0.875 (0.771-0.979), respectively, and the precisions were 0.813 (0.690-0.935) and 0.913 (0.825-1.000), respectively. CONCLUSIONS: We developed a two-step model based on ML using noninvasive variables and ABG results to screen for the presence of IPVD in cirrhotic patients. This model may partly solve the problem of limited access to CEE and ABG by a large numbers of cirrhotic patients.

Effects of circadian rhythm on Narcotrend index and target-controlled infusion concentration of propofol anesthesia.

BACKGROUND: The effects of circadian rhythms on drug metabolism and efficacy are being increasingly recognized. However, the extent to which they affect general anesthesia remains unclear. This study aims to investigate the effects of circadian rhythms on anesthetic depth and the concentrations of propofol target-controlled infusion (TCI). METHODS: Sixty patients undergoing laparoscopic surgeries were sequentially assigned to four groups. Group ND (n = 15): Propofol TCI with Narcotrend monitor during the day (8:00-18:00), Group NN (n = 15): Propofol TCI with Narcotrend monitor during the night (22:00-5:00), Group CLTD (n = 15): Propofol closed-loop TCI guided by bispectral index (BIS) during the day (8:00-18:00), Group CLTN (n = 15): Propofol closed-loop TCI guided by BIS during the night (22:00-5:00). The Narcotrend index, mean arterial pressure (MAP) and heart rate (HR) were compared between group ND and NN at 7 time points, from 5 min before induction to the end of operation. The propofol TCI concentrations, MAP and HR were compared between group CLTD and CLTN at 7 time points, from 5 min after induction to the end of operation. RESULTS: The Narcotrend index, MAP, and HR in group NN were lower than those in group ND from the beginning of mechanical ventilation to the end of operation (p < 0.05). The propofol TCI concentrations in group CLTN were lower than those in group CLTD from the beginning of operation to the end of operation (p < 0.05). CONCLUSION: Circadian rhythms have a significant effect on the depth of anesthesia and drug infusion concentrations during propofol TCI. When using general anesthesia during night surgery, the propofol infusion concentration should be appropriately reduced compared to surgery during the day. TRIAL REGISTRATION: The present study was registered on the ClinicalTrials.gov website ( NCT02440269 ) and approved by the Medical Ethics Committee of Southwest Hospital of Third Military Medical University (ethics lot number: 2016 Research No. 93). All patients provided informed written consent to participate in the study.

A better method for the dynamic, precise estimating of blood/haemoglobin loss based on deep learning of artificial intelligence.

BACKGROUND: Dynamic and precise estimation of blood loss (EBL) is quite important for perioperative management. To date, the Triton System, based on feature extraction technology (FET), has been applied to estimate intra-operative haemoglobin (Hb) loss but is unable to directly assess the amount of blood loss. We aimed to develop a method for the dynamic and precise EBL and estimate Hb loss (EHL) based on artificial intelligence (AI). METHODS: We collected surgical patients' non-recycled blood to generate blood-soaked sponges at a set gradient of volume. After image acquisition and preprocessing, FET and densely connected convolutional networks (DenseNet) were applied for EBL and EHL. The accuracy was evaluated using R2, the mean absolute error (MAE), the mean square error (MSE), and the Bland-Altman analysis. RESULTS: For EBL, the R2, MAE and MSE for the method based on DenseNet were 0.966 (95% CI: 0.962-0.971), 0.186 (95% CI: 0.167-0.207) and 0.096 (95% CI: 0.084-0.109), respectively. For EHL, the R2, MAE and MSE for the method based on DenseNet were 0.941 (95% CI: 0.934-0.948), 0.325 (95% CI: 0.293-0.355) and 0.284 (95% CI: 0.251-0.317), respectively. The accuracies of EBL and EHL based on DenseNet were more satisfactory than that of FET. Bland-Altman analysis revealed a bias of 0.02 ml with narrow limits of agreement (LOA) (-0.47 to 0.52 mL) and of 0.05 g with narrow LOA (-0.87 to 0.97 g) between the methods based on DenseNet and actual blood loss and Hb loss. CONCLUSIONS: We developed a simpler and more accurate AI-based method for EBL and EHL, which may be more fit for surgeries primarily using sponges and with a small to medium amount of blood loss.

AMD3100 treatment attenuates pulmonary angiogenesis by reducing the c-kit (+) cells and its pro-angiogenic activity in CBDL rat lungs.

Recent studies have shown that pulmonary angiogenesis is an important pathological process in the development of hepatopulmonary syndrome (HPS), and growing evidence has indicated that Stromal cell-derived factor 1/C-X-C chemokine receptor type 4 (SDF-1/CXCR4) axis is involved in pulmonary vascular disease by mediating the accumulation of c-kit+ cells. This study aimed to test the effect of AMD3100, an antagonist of CXCR4, in HPS pulmonary angiogenesis. Common bile duct ligation (CBDL) rats were used as experimental HPS model and were treated with AMD3100 (1.25mg/kg/day, i.p.) or 0.9% saline for 3weeks. The sham rats underwent common bile duct exposure without ligation. The c-kit+ cells accounts and its angiogenic-related functions, prosurvival signals, pulmonary angiogenesis and arterial oxygenation were analysed in these groups. Our results showed that pulmonary SDF-1/CXCR4, Akt, Erk and VEGF/VEGFR2 were significantly activated in CBDL rats, and the numbers of circulating and pulmonary c-kit+ cells were increased in CBDL rats compared with control rats. Additionally, the angiogenic-related functions of c-kit+ cells and pulmonary microvessel counts were also elevated in CBDL rats. CXCR4 inhibition reduced pulmonary c-kit+ cells and microvessel counts and improved arterial oxygenation within 3weeks in CBDL rats. The pulmonary prosurvival signals and pro-angiogenic activity of c-kit+ cells were also down-regulated in AMD3100-treated rats. In conclusion, AMD3100 treatment attenuated pulmonary angiogenesis in CBDL rats and prevented the development of HPS via reductions in pulmonary c-kit+ cells and inhibition of the prosurvival signals. Our study provides new insights in HPS treatment.

Deferoxamine regulates neuroinflammation and iron homeostasis in a mouse model of postoperative cognitive dysfunction.

BACKGROUND: Postoperative cognitive dysfunction (POCD) is a common complication after surgery, especially amongst elderly patients. Neuroinflammation and iron homeostasis are key hallmarks of several neurological disorders. In this study, we investigated the role of deferoxamine (DFO), a clinically used iron chelator, in a mouse model of surgery-induced cognitive dysfunction and assessed its neuroprotective effects on neuroinflammation, oxidative stress, and memory function. METHODS: A model of laparotomy under general anesthesia and analgesia was used to study POCD. Twelve to 14 months C57BL/6J male mice were treated with DFO, and changes in iron signaling, microglia activity, oxidative stress, inflammatory cytokines, and neurotrophic factors were assessed in the hippocampus on postoperative days 3, 7, and 14. Memory function was evaluated using fear conditioning and Morris water maze tests. BV2 microglia cells were used to test the anti-inflammatory and neuroprotective effects of DFO. RESULTS: Peripheral surgical trauma triggered changes in hippocampal iron homeostasis including ferric iron deposition, increase in hepcidin and divalent metal transporter-1, reduction in ferroportin and ferritin, and oxidative stress. Microglia activation, inflammatory cytokines, brain-derived neurotropic factor impairments, and cognitive dysfunction were found up to day 14 after surgery. Treatment with DFO significantly reduced neuroinflammation and improved cognitive decline by modulating p38 MAPK signaling, reactive oxygen species, and pro-inflammatory cytokines release. CONCLUSIONS: Iron imbalance represents a novel mechanism underlying surgery-induced neuroinflammation and cognitive decline. DFO treatment regulates neuroinflammation and microglia activity after surgery.

The angiogenic related functions of bone marrow mesenchymal stem cells are promoted by CBDL rat serum via the Akt/Nrf2 pathway.

Hepatopulmonary syndrome (HPS) is a complication of severe liver disease. It is characterized by an arterial oxygenation defect. Recent studies have demonstrated that pulmonary angiogenesis contributes to the abnormal gas exchange found in HPS. Additionally, mesenchymal stem cells (MSCs) are considered the stable source of VEGF-producing cells and have the potential to differentiate into multiple cell types. However, it has not been determined whether bone marrow mesenchymal stem cells (BM-MSCs) are mobilized and involved in the pulmonary angiogenesis in HPS. In this study, a CFU-F assay showed that the number of peripheral blood MSCs was increased in common bile duct ligation (CBDL) rats; however, there was no significant difference found in the number of BM-MSCs. In vitro, CBDL rat serum induced the overexpression of CXCR4 and PCNA in BM-MSCs. Consistently, the directional migration as well as the proliferation ability of BM-MSCs were enhanced by CBDL rat serum, as determined by a transwell migration and MTT assays. Moreover, the secretion of VEGF by BM-MSCs increased after treatment with CBDL rat serum. We also found that the expression of phospho-Akt, phospho-ERK, and Nrf2 in BM-MSCs was significantly up-regulated by CBDL rat serum in a time dependent manner, and the blockage of the Akt/Nrf2 signalling pathway with an Akt Inhibitor or Nrf2 siRNA, instead of an ERK inhibitor, attenuated the migration, proliferation and paracrine capacity of BM-MSCs. In conclusion, these findings indicated that the number of MSCs increased in the peripheral blood of CBDL rats, and the Akt/Nrf2 pathway plays a vital role in promoting the angiogenic related functions of BM-MSCs, which could be a potent contributor to pulmonary angiogenesis in HPS.

The involvement of aquaporin 1 in the hepatopulmonary syndrome rat serum-induced migration of pulmonary arterial smooth muscle cells via the p38-MAPK pathway.

Hepatopulmonary syndrome (HPS) is characterized by arterial oxygenation defects induced by intrapulmonary vascular dilation (IPVD). Pulmonary vascular remodeling (PVR) is an important pathological feature of IPVD; however, the details regarding the underlying mechanisms of this process remain undefined. Recent studies have determined that the abnormal migration of pulmonary arterial smooth muscle cells (PASMCs) plays a role in the pathogenesis of the PVR associated with HPS. Additionally, aquaporin 1 (AQP1) not only functions as a water channel molecule but also promotes cell migration by facilitating water transport in the lamellipodia of migrating cells. Common bile duct ligation (CBDL) rat is a well-accepted HPS model; we determined that the immunoperoxidase labeling of AQP1 was enhanced in the media of the pulmonary vessels in CBDL rats. HPS rat serum mediated the overexpression of AQP1 in PASMCs, and also upregulated PASMC migration. Small interfering RNAs (siRNAs) that targeted rat AQP1 caused significant downregulation of AQP1, which resulted in decreased PASMC migration. Furthermore, the inhibition of the p38-MAPK pathway abolished AQP1-dependent PASMC migration. In conclusion, this study demonstrated that AQP1 enhanced PASMC migration via the p38-MAPK pathway in rat with HPS and may represent a potential therapeutic strategy in the setting of pulmonary vascular remodeling associated with HPS.

Lung injury following lower extremity blast trauma in rats.

BACKGROUND: Lower extremity blast trauma is a common injury during armed conflict and after terrorist attacks with a high mortality, which is likely associated with distant vital organ injury. The current study aimed to investigate the underlying mechanisms of remote lung injury after blast lower extremity trauma. METHODS: Sprague-Dawley rats were randomly divided into two groups: sham and blast. The blast group underwent blast trauma to the left hind limb using chartaceous electricity detonators, which was then subdivided into the time at which they were sacrificed: 0.5, 1, 3, and 6 hours. The sham group was also subdivided into the baseline control and time course groups. The baseline group was sacrificed 0.5 hours after artery cannulation and the time course at 6 hours after sham blast. The lungs were harvested for histologic analysis and water content measurement. Blood samples were harvested at each end of experiment and analyzed for cytokines, myeloperoxidase, malondialdehyde, and superoxide dismutase and cystathionine γ-lyase activity and hydrogen sulfide. RESULTS: Blast hind limb trauma induced alveolar injury and cell infiltration, together with an increase in lung water content, in a time-dependent manner. Plasma and lung levels of proinflammatory cytokines, tumor necrosis factor-α and interleukin 6, and malondialdehyde, were found to be significantly increased in conjunction with a rise in myeloperoxidase and a concurrent fall in superoxide dismutase, cystathionine γ-lyase, and hydrogen sulfide. CONCLUSION: Our data demonstrated that blast limb trauma causes remote lung injury, which is likely associated with remarkable inflammatory response, oxidative stress, and depletion of protective mechanisms.

Low- and high-dose hydrogen peroxide regulation of transcription factor NF-E2-related factor 2.

BACKGROUND: Reactive oxygen species (ROS) may play both physiological and pathophysiological roles. Transcription factor NF-E2-related factor 2 (Nrf2) regulates antioxidant response element (ARE)-mediated genes expression and coordinates induction of chemoprotective proteins in response to physical and chemical stresses. The exact role of Nrf2 in cellular responses to different levels of oxidative stresses remains unknown. METHODS: Rat pulmonary microvascular endothelial cells were cultured and treated with 0 mmol/L, 0.125 mmol/L, 0.25 mmol/L, 0.5 mmol/L, 1.0 mmol/L and 2.0 mmol/L hydrogen peroxide solution for 2 hours. Nrf2 gene expression was assayed by reverse transcription-PCR, Nrf2-ARE binding activity was assayed with electrophoretic mobility shift assay (EMSA), and localization of Nrf2 was detected with immunohistochemistry. RESULTS: Low and moderate (0.125 mmol/L, 0.25 mmol/L and 0.5 mmol/L) doses hydrogen peroxide exposure of rat pulmonary microvascular endothelial cells led to the nuclear accumulation of Nrf2, increased activity of transcription regulation and up-regulation of ARE-medicated gene expression. In contrast, high doses of hydrogen peroxide (1 mmol/L, 2 mmol/L) exposure of the cells led to the nuclear exclusion of Nrf2, decreased activity transcription regulation and down-regulation of ARE-mediated gene expression. CONCLUSION: Low and moderate doses of hydrogen peroxide play protective roles by increasing transcription activity of Nrf2, whereas high- dose hydrogen peroxide plays a deleterious role by decreasing transcription activity of Nrf2.