Prognostic Assessment of COVID-19 in the Intensive Care Unit by Machine Learning Methods: Model Development and Validation.

BACKGROUND: Patients with COVID-19 in the intensive care unit (ICU) have a high mortality rate, and methods to assess patients' prognosis early and administer precise treatment are of great significance. OBJECTIVE: The aim of this study was to use machine learning to construct a model for the analysis of risk factors and prediction of mortality among ICU patients with COVID-19. METHODS: In this study, 123 patients with COVID-19 in the ICU of Vulcan Hill Hospital were retrospectively selected from the database, and the data were randomly divided into a training data set (n=98) and test data set (n=25) with a 4:1 ratio. Significance tests, correlation analysis, and factor analysis were used to screen 100 potential risk factors individually. Conventional logistic regression methods and four machine learning algorithms were used to construct the risk prediction model for the prognosis of patients with COVID-19 in the ICU. The performance of these machine learning models was measured by the area under the receiver operating characteristic curve (AUC). Interpretation and evaluation of the risk prediction model were performed using calibration curves, SHapley Additive exPlanations (SHAP), Local Interpretable Model-Agnostic Explanations (LIME), etc, to ensure its stability and reliability. The outcome was based on the ICU deaths recorded from the database. RESULTS: Layer-by-layer screening of 100 potential risk factors finally revealed 8 important risk factors that were included in the risk prediction model: lymphocyte percentage, prothrombin time, lactate dehydrogenase, total bilirubin, eosinophil percentage, creatinine, neutrophil percentage, and albumin level. Finally, an eXtreme Gradient Boosting (XGBoost) model established with the 8 important risk factors showed the best recognition ability in the training set of 5-fold cross validation (AUC=0.86) and the verification queue (AUC=0.92). The calibration curve showed that the risk predicted by the model was in good agreement with the actual risk. In addition, using the SHAP and LIME algorithms, feature interpretation and sample prediction interpretation algorithms of the XGBoost black box model were implemented. Additionally, the model was translated into a web-based risk calculator that is freely available for public usage. CONCLUSIONS: The 8-factor XGBoost model predicts risk of death in ICU patients with COVID-19 well; it initially demonstrates stability and can be used effectively to predict COVID-19 prognosis in ICU patients.

Two novel mutations in the BCKDHB gene that cause maple syrup urine disease.

BACKGROUND: Maple syrup urine disease (MSUD) is a rare metabolic disorder of autosomal recessive inheritance caused by decreased activity of branched-chain α-ketoacid dehydrogenase complex (BCKD). Mutations in the three genes (BCKDHA, BCKDHB and DBT) are associated with MSUD. Here, we describe the presenting symptoms, clinical course and gene mutation analysis of a Chinese boy with MSUD. METHODS: Plasma amino acid analysis was performed by tandem mass spectrometry and the levels of organic acids in urine were measured with gas chromatography-mass spectrometry. The BCKDHB gene was sequenced by Sanger method. Furthermore, the significance of the novel mutations was predicted by Polyphen and Mutationtaster. After diagnosis, the patient was fed with protein-restricted diet to reduce intake of BCAA and was treated with l-carnitine. Metabolic parameters, clinical presentation and mental development were followed up. RESULTS: The patient was diagnosed as MSUD. Two novel BCKDHB mutations (c.523 T > C and c.478-25_552del100) were identified. In silico analysis predicted that the two mutations were "disease causing". The boy tolerated the treatment well and had symptomatic improvement. He presented with mild hypotonia and had nearly normal DQ scores at the age of 10 months. The two novel mutations resulted in the clinical manifestations of MSUD. Our results may reflect the heterogeneity of the pathogenic variants found in patients with MSUD.

Correlation study of cough variant asthma and mycoplasma pneumonia infection in children.

Cough variant asthma (CVA), as one of bronchitis diseases, features with repeated cough. In clinics, CAV does not show significant abnormal signs, therefore antibiotic therapy can hardly achieve satisfactory treatment effect. With the development of scientific technologies, the correlation between mycoplasma pneumoniae infection and CVA has become a hot research topic in clinics. In clinics, mycoplasma pneumoniae is extensively regarded as major cause for CVA, with complex pathogenic mechanism. The symptoms of CVA is characterized by chronic non abnormal inflammation, normally accompanied with bronchospasm and intestinal infection. Clinical practices show that about 6% of CVA children mainly show continuous cough till midnight during onset period, which is easily to misdiagnosed as bronchitis. Mycoplasma pneumoniae infection is a common disease threatening Children's health in China. With the annual increase of incidence of infantile pneumonia in China, mycoplasma pneumoniae infection has become a major reason leading to the death of child in China. More such mycoplasma pneumoniae infection is a sporadic disease spreading in a small range, and can onset in four seasons, making great impact on living quality and health of children. On this basis, this paper analyzes the correlation between Children's CVA and mycoplasma pneumoniae infection, in the hope of providing valuable reference for clinical treatment.

Roles of T-cell Immunoglobulin and Mucin Domain Genes and Toll-like Receptors in Wheezy Children with Mycoplasma pneumoniae Pneumonia.

BACKGROUND: The study aimed to explore possible factors influencing wheezing in children with Mycoplasma pneumoniae pneumonia (MPP). METHODS: The study included 84 children with MPP, who were divided into two groups: wheezy group (n=40) and non-wheezy group (n=44), along with 30 age-matched healthy controls. T-cell immunoglobulin and mucin domain gene (Tim) 1, 3 and Toll-like receptor (TLR) 2, 4 were evaluated using RT-PCR. Serum IL-10, TNF-α, IFN-γ and IgE were assessed by enzyme-linked immunosorbent assay. Peripheral blood eosinophil (EOS) was measured by an automated haematology. RESULTS: Children with MPP had markedly increased TLR2, TLR4, Tim1, IL-10, TNF-α, IgE and EOS, and decreased IFN-γ than the healthy controls. In the presence of MPP, wheezy children had significantly elevated TLR2, Tim1, Tim3, TNF-α, IgE and EOS than non-wheezy children. In wheezy children with MPP, MP-specific antibody titre was positively correlated with TLR2 and TIM1, and negatively correlated with IFN-γ. IgE was positively correlated with TLR2, TLR4 and Tim1, while EOS was positively correlated with Tim1 and Tim3. CONCLUSION: TLR2, Tim1, Tim3, TNF-α, IgE and EOS play a role in MPP-related wheezing in children. The role of IgE might be associated with TLR2 and Tim1, and the role of EOS might be associated with Tim1 and Tim3.

α-1-Acid glycoprotein as a biomarker for the early diagnosis and monitoring the prognosis of sepsis.

OBJECTIVE: To explore the value of α-1-acid glycoprotein (AGP) for the early diagnostic and prognostic assessment of patients with sepsis. METHODS: Eighty-five patients with systemic inflammatory response syndrome (SIRS) and 192 patients with sepsis were enrolled. White blood cell counts and serum levels of AGP, C-reactive protein, and procalcitonin were tested on the day of admission to intensive care unit (ICU; day 1) and the following days 3, 5, 7, and 10. RESULTS: The sepsis group exhibited significantly higher levels of AGP than did the SIRS group on day 1 (P < .05); the area under the curve (AUC) of AGP was 0.869 with a specificity of 0.902 on diagnosis of sepsis. The differences were statistically significant among sepsis subgroups. On prognostic assessment, the areas under the curve of AGP, Sequential Organ Failure Assessment (SOFA) scores, and SOFA + AGP on ICU admission were 0.793, 0.813, and 0.878, respectively. The results of logistic regression showed that the odds ratios of AGP, SOFA, Acute Physiology and Chronic Health Evaluation II, and the length of ICU stay were 1.450, 1.212, 1.673, and 1.130. CONCLUSIONS: α-1-Acid glycoprotein could distinguish sepsis from SIRS and also be used to effectively assess the severity of sepsis. In addition, combined AGP and SOFA scores had a great predicting value in prognosis of sepsis.

Diagnosis, treatment and follow-up of patients with tetrahydrobiopterin deficiency in Shandong province, China.

OBJECTIVES: To summarize the clinical and biochemical data, mutation analysis, treatment, outcome and the follow-up data of patients with BH4 deficiency from 2004 to 2012 in Shandong province, China. METHODS: We analyzed the clinical, biochemical and treatment data of 40 patients with BH4 deficiency. Urinary neopterin and biopterin were analyzed. Further BH4 loading tests were performed in suspected patients with abnormal urinary pterin profiles. The patients with BH4 deficiency were treated with BH4 and neurotransmitter after diagnosis. Blood phenylalanine level, clinical symptoms and mental development were followed up. RESULTS: 40 cases with BH4 deficiency were identified and all classified as PTPS deficiency between 2004 and 2012 in Shandong province, China. They were diagnosed at the age of 20d - 41m and most patients received treatment with BH4, l-dopa and 5-HTP after diagnosis. Seven different mutations (P87S, K91R, T106M, D96N, N52S, S21R, and L127F) were detected in 11 patients. But outcome assessments were not always available. We obtained 19 records of DQ/IQ assessment. In 9 patients (7 early and 2 late diagnosed) no development delay is observed, while in 10 patients (8 early and 2 late diagnosed) development was delayed. CONCLUSIONS: Our study emphasized that screening for BH4 deficiency should be carried out in all patients with HPA in order to minimize misdiagnosis. Although the outcomes of BH4 deficiency are highly variable, early diagnosis and treatment is essential for good outcomes.

Prognosis and weaning of elderly multiple organ dysfunction syndrome patients with invasive mechanical ventilation.

BACKGROUND: Elderly multiple organ dysfunction syndrome (MODS) patients receiving invasive mechanical ventilation have poor prognosis in intensive care units (ICUs). We studied the usefulness of four commonly used severity scores and extrapulmonary factors that affected weaning to predict outcome of such patients. METHODS: Clinical data of 197 patients on admission to ICUs (from January 2009 to June 2012) were used retrospectively. The Acute Physiology and Chronic Health Evaluation (APACHE) II, APACHE III, Sample Acute Physiological Score (SAPS) II and MODS scores were calculated. All the patients were grouped into survivors and nonsurvivors according to the prognosis. Patients, who weaned from ventilator (n = 154), were subdivided into a successful weaning group and a failed weaning group. The receiver operating characteristic (ROC) curves and Logistic regression was used for prognostic and weaning assessment. RESULTS: Based on the outcomes, the areas under the ROC of APACHE II, APACHE III, SAPS II, and MODS were 0.837, 0.833, 0.824, and 0.837, respectively. The Logistic regression analysis revealed that the odds ratio (OR) of underlying lung diseases, serum albumin and creatinine, and the number of organ failures was 2.374, 0.920, 1.003, and 1.547. APACHE II scores on admission performed excellent (ROC: 0.921) on the weaning assessments. CONCLUSIONS: APACHE II and MODS systems were marginally better for evaluating the prognosis of elderly MODS patients who received invasive mechanical ventilation. Underlying lung diseases, serum albumin, serum creatinine and the number of organ failures were independent prognostic factors. Using the APACHE II scores on admission before weaning may increase the likelihood of successful weaning. (ClinicalTrial.gov identifier NCT01802983).

Heliox as a driving gas to atomize inhaled drugs on acute exacerbation of chronic obstructive pulmonary disease: a prospective clinical study.

BACKGROUND: Acute exacerbation of chronic obstructive pulmonary disease (AECOPD) is a common condition, which affects not only the quality of life of patients but also their prognosis. The purpose of this study was to explore the effects of an inhaled salbutamol sulfate solution and an inhalation suspension of the glucocorticoid budesonide that were atomized with heliox to treat patients with AECOPD. METHODS: Twenty-three patients with AECOPD were divided into a treatment group (He/O2 = 70%/30%) and a control group (N2/O2 = 70%/30%). The salbutamol sulfate and budesonide were administered by inhalation twice a day for 7 days. Vital signs, arterial blood gas levels, pulmonary function and the levels of serum myostatin (sMSTN) were measured and lung vibration imaging was performed. RESULTS: We found that the PaO2 and PaCO2 values were not significantly different between the two groups at the various time points (P > 0.05). There were also no significant differences in any of the parameters of pulmonary function between the two groups. However, after baseline correction, the increase rate of the forced expiratory volume in one second (FEV1), the forced vital capacity (FVC), and the maximum minute ventilation (MVV) appeared to be significantly increased at some time points compared with the baseline (before treatment) in both groups (P < 0.05). Although the values of quantitative lung distribution (QLD) for different regions and the levels of sMSTN were slightly different between the two groups, the repeated measures analysis of variance (ANOVA) revealed that there were no significant differences between the two groups or within any group (P > 0.05). CONCLUSION: Although the use of heliox as a driving gas can improve symptoms and benefit patients with AECOPD, the heliox treatment group did not have significant differences in arterial blood gases, lung function, lung vibration response imaging or the levels of sMSTN compared with the control group. (Chinese Clinical Trial Register Center ChiCTRTRC-00000273).

Analysis of the severity and prognosis assessment of aged patients with community-acquired pneumonia: a retrospective study.

BACKGROUND: Community-acquired pneumonia (CAP) is a prevalent and potentially life-threatening infection, and has poor prognosis in aged patients. The objective of this study was to compare the potential of admission N-terminal pro B-type natriuretic peptide (proBNP) levels and scoring models [CURB-65, Pneumonia Severity Index (PSI), and Acute Physiology and Chronic Health Evaluation (APACHE) II scores] to predict outcomes for aged patients with CAP admitted to Intensive Care Unit (ICU), and to explore the prognostic factors. METHODS: Clinical data of the patients were collected retrospectively, whose CURB-65, PSI, APACHE II scores were calculated and in whom measurements of proBNP was performed. The outcomes of interest were severity evaluation, prediction of need for mechanical ventilation and 28-day mortality. Receiver operating characteristic (ROC) curve was conducted to predict the assessment ability of proBNP and scoring models on different outcomes, and the logistic regression analysis was performed to screen factors affecting prognosis. RESULTS: 240 patients were enrolled, with the mean age of 75±8 years old. Admission levels of NT-proBNP, scoring models were significantly higher in SCAP patients, MV group, and non-survivors compared to non-SCAP patients, no-MV group, and 28-day survivors, respectively (P<0.001). PSI had the highest area under the curve (AUC) and specificity for the three outcomes considered (AUC: 0.868 and specificity: 0.906 for 28-day mortality, AUC: 0.864 and specificity: 0.831 for requirement of MV, and AUC: 0.888 and specificity: 0.894 for severity evaluation). NT-proBNP had the highest sensitivity of 0.987 and 0.903 on prediction of mortality and need for MV. And APACHE II scoring model with the highest sensitivity of 0.890 was used to evaluate severity. Logistic regression analysis showed that the odd ratio (OR) of systolic blood pressure, PSI, and APACHE II scores were 0.886, 1.019, and 1.249. CONCLUSIONS: PSI scores was the best indicator in predicting different clinical outcomes of aged patients with CAP among the proBNP and three scoring systems. Systolic blood pressure might be as a protective factor for prognosis while PSI and APACHE II scores as risk factors for prognosis of aged patients with CAP.

Value of soluble TREM-1, procalcitonin, and C-reactive protein serum levels as biomarkers for detecting bacteremia among sepsis patients with new fever in intensive care units: a prospective cohort study.

BACKGROUND: The purpose of this study was to explore the diagnostic value of soluble triggering receptor expressed on myeloid cells 1 (sTREM-1), procalcitonin (PCT), and C-reactive protein (CRP) serum levels for differentiating sepsis from SIRS, identifying new fever caused by bacteremia, and assessing prognosis when new fever occurred. METHODS: We enrolled 144 intensive care unit (ICU) patients: 60 with systemic inflammatory response syndrome (SIRS) and 84 with sepsis complicated by new fever at more than 48 h after ICU admission. Serum sTREM-1, PCT, and CRP levels were measured on the day of admission and at the occurrence of new fever (>38.3°C) during hospitalization. Based on the blood culture results, the patients were divided into a blood culture-positive bacteremia group (33 patients) and blood culture-negative group (51 patients). Based on 28-day survival, all patients, both blood culture-positive and -negative, were further divided into survivor and nonsurvivor groups. RESULTS: On ICU day 1, the sepsis group had higher serum sTREM-1, PCT, and CRP levels compared with the SIRS group (P <0.05). The areas under the curve (AUC) for these indicators were 0.868 (95% CI, 0.798-0.938), 0.729 (95% CI, 0.637-0.821), and 0.679 (95% CI, 0.578-0.771), respectively. With 108.9 pg/ml as the cut-off point for serum sTREM-1, sensitivity was 0.83 and specificity was 0.81. There was no statistically significant difference in serum sTREM-1 or PCT levels between the blood culture-positive and -negative bacteremia groups with ICU-acquired new fever. However, the nonsurvivors in the blood culture-positive bacteremia group had higher levels of serum sTREM-1 and PCT (P <0.05), with a prognostic AUC for serum sTREM-1 of 0.868 (95% CI, 0.740-0.997). CONCLUSIONS: Serum sTREM-1, PCT, and CRP levels each have a role in the early diagnosis of sepsis. Serum sTREM-1, with the highest sensitivity and specificity of all indicators studied, is especially notable. sTREM-1, PCT, and CRP levels are of no use in determining new fever caused by bacteremia in ICU patients, but sTREM-1 levels reflect the prognosis of bacteremia. TRIAL REGISTRATION: ClinicalTrial.gov identifier NCT01410578.