Laboratory variables-based artificial neural network models for predicting fatty liver disease: A retrospective study.

Background: The efficacy of artificial neural network (ANN) models employing laboratory variables for predicting fatty liver disease (FLD) remains inadequately established. The study aimed to develop ANN models to precisely predict FLD. Methods: Of 12,058 participants undergoing the initial FLD screening, 7,990 eligible participants were included. A total of 6,309 participants were divided randomly into the training (4,415 participants, 70%) and validation (1,894 participants, 30%) sets for developing prediction models. The performance of ANNs was additionally tested in the testing set (1,681 participants). The area under the receiver operating characteristic curve (AUROC) was employed to assess the models' performance. Results: The 18-variable, 11-variable, 3-variable, and 2-variable models each achieved robust FLD prediction performance, with AUROCs over 0.92, 0.91, and 0.89 in the training, validation, and testing, respectively. Although slightly inferior to the other three models in performance (AUROC ranges: 0.89-0.92 vs 0.91-0.95), the 2-variable model showed 80.3% accuracy and 89.7% positive predictive value in the testing. Incorporating age and gender increased the AUROCs of the resulting 20-variable, 13-variable, 5-variable, and 4-variable models each to over 0.93, 0.92, and 0.91 in the training, validation, and testing, respectively. Conclusions: Implementation of the ANN models could effectively predict FLD, with enhanced predictive performance via the inclusion of age and gender.

Several laboratory variables indicate severity and prognosis of COVID-19.

OBJECTIVE: While several laboratory variables have been used to assess COVID-19 disease, to our knowledge, no attempt has previously been made to compare differences across different patient groups. We attempted to evaluate the relationship between laboratory variables and severity of the disease as well as on prognosis. METHOD: We searched BioLINCC database and identified three studies which had separately included outpatients, inpatients, and ICU patients. For this re-analysis, we extracted data on general demography, laboratory variables and outcome. RESULT: In total, 2454 participants (496 outpatients [Study 1], 478 inpatients [Study 2], and 1480 ICU patients [Study 3]) were included in the analysis. We found three laboratory variables (i.e., creatinine, aspartate transferase, and albumin) were not only prognostic factors for outcome of inpatients with COVID-19, but also reflected disease severity as they were significantly different between inpatients and ICU patients. These three laboratory variables are an indication of kidney function, liver function, and nutritional status. CONCLUSION: For patients with COVID-19, in addition to monitoring infectious disease indicators, we need to pay attention to liver function, renal function, and take timely measures to correct them to improve prognosis.

Machine Learning-Based Routine Laboratory Tests Predict One-Year Cognitive and Functional Decline in a Population Aged 75+ Years.

BACKGROUND: Cognitive and functional decline are common problems in older adults, especially in those 75+ years old. Currently, there is no specific plasma biomarker able to predict this decline in healthy old-age people. Machine learning (ML) is a subarea of artificial intelligence (AI), which can be used to predict outcomes Aim: This study aimed to evaluate routine laboratory variables able to predict cognitive and functional impairment, using ML algorithms, in a cohort aged 75+ years, in a one-year follow-up study. METHOD: One hundred and thirty-two older adults aged 75+ years were selected through a community-health public program or from long-term-care institutions. Their functional and cognitive performances were evaluated at baseline and one year later using a functional activities questionnaire, Mini-Mental State Examination, and the Brief Cognitive Screening Battery. Routine laboratory tests were performed at baseline. ML algorithms-random forest, support vector machine (SVM), and XGBoost-were applied in order to describe the best model able to predict cognitive and functional decline using routine tests as features. RESULTS: The random forest model showed better accuracy than other algorithms and included triglycerides, glucose, hematocrit, red cell distribution width (RDW), albumin, hemoglobin, globulin, high-density lipoprotein cholesterol (HDL-c), thyroid-stimulating hormone (TSH), creatinine, lymphocyte, erythrocyte, platelet/leucocyte (PLR), and neutrophil/leucocyte (NLR) ratios, and alanine transaminase (ALT), leukocyte, low-density lipoprotein cholesterol (LDL-c), cortisol, gamma-glutamyl transferase (GGT), and eosinophil as features to predict cognitive decline (accuracy = 0.79). For functional decline, the most important features were platelet, PLR and NLR, hemoglobin, globulin, cortisol, RDW, glucose, basophil, B12 vitamin, creatinine, GGT, ALT, aspartate transferase (AST), eosinophil, hematocrit, erythrocyte, triglycerides, HDL-c, and monocyte (accuracy = 0.92). CONCLUSIONS: Routine laboratory variables could be applied to predict cognitive and functional decline in oldest-old populations using ML algorithms.

Impact of Chart-Derived Frailty Index on 1-Year Mortality After Radical Cystectomy in 1004 Patients with Bladder Cancer.

BACKGROUND: Radical cystectomy is a major urological procedure with high morbidity and mortality. The chart-derived frailty index (CFI), a measure of preoperative frailty, can be calculated by using demographic and routine laboratory variables. We assessed the impact of CFI on 1-year mortality after radical cystectomy. METHODS: This retrospective study included patients with bladder cancer who underwent radical cystectomy between 2007 and 2021. The CFI was calculated as the sum of the presence of the following parameters: age > 70 years, body mass index < 18.5 kg/m2, hematocrit < 35%, albumin < 3.4 g/dL, and creatinine > 2.0 mg/dL. Patients were divided into those with low (0-2) and high (3-5) CFI. The 1-year, all-cause and cancer-specific mortalities after radical cystectomy were evaluated. RESULTS: Of the 1004 patients, 914 (91.0%) had a low CFI and 90 (9.0%) had a high CFI. The 1-year, all-cause mortality in the low and high CFI groups was 12.0% and 27.8%, respectively (P < 0.001). Multivariate Cox regression analysis revealed that high CFI (P < 0.001), tumor stage (P = 0.003), and red blood cell transfusion amount (P < 0.001) were significantly associated with 1-year, all-cause mortality after radical cystectomy. Kaplan-Meier survival analysis demonstrated significantly different 1-year, all-cause and cancer-specific mortalities after radical cystectomy between patients with a high CFI and those with a low CFI (log-rank test, both P < 0.001). CONCLUSIONS: High CFI is associated with higher 1-year mortality after radical cystectomy, suggesting that the CFI can effectively predict mortality after radical cystectomy.

The impacts of vaccination status and host factors during early infection on SARS-CoV-2 persistence:a retrospective single-center cohort study.

BACKGROUND: Viral persistence is a crucial factor that influences the transmissibility of SARS-CoV-2. However, the impacts of vaccination and physiological variables on viral persistence have not been adequately clarified. METHODS: We collected the clinical records of 377 COVID-19 patients, which contained unvaccinated patients and patients received two doses of an inactivated vaccine or an mRNA vaccine. The impacts of vaccination on disease severity and viral persistence and the correlations between 49 laboratory variables and viral persistence were analyzed separately. Finally, we established a multivariate regression model to predict the persistence of viral RNA. RESULTS: Both inactivated and mRNA vaccines significantly reduced the rate of moderate cases, while the vaccine related shortening of viral RNA persistence was only observed in moderate patients. Correlation analysis showed that 10 significant laboratory variables were shared by the unvaccinated mild patients and mild patients inoculated with an inactivated vaccine, but not by the mild patients inoculated with an mRNA vaccine. A multivariate regression model established based on the variables correlating with viral persistence in unvaccinated mild patients could predict the persistence of viral RNA for all patients except three moderate patients inoculated with an mRNA vaccine. CONCLUSION: Vaccination contributed limitedly to the clearance of viral RNA in COVID-19 patients. While, laboratory variables in early infection could predict the persistence of viral RNA.

Using machine learning models to predict HBeAg seroconversion in CHB patients receiving pegylated interferon-α monotherapy.

BACKGROUND AND OBJECTIVE: Though there are many advantages of pegylated interferon-α (PegIFN-α) treatment to chronic hepatitis B (CHB) patients, the response rate of PegIFN-α is only 30 ~ 40%. Therefore, it is important to explore predictors at baseline and establish models to improve the response rate of PegIFN-α. METHODS: We randomly divided 260 HBeAg-positive CHB patients who were not previously treated and received PegIFN-α monotherapy (180 µg/week) into a training dataset (70%) and testing dataset (30%). The intersect features were extracted from 50 routine laboratory variables using the recursive feature elimination method algorithm, Boruta algorithm, and Least Absolute Shrinkage and Selection Operator Regression algorithm in the training dataset. After that, based on the intersect features, eight machine learning models including Logistic Regression, k-Nearest Neighbors, Support Vector Machine, Decision Tree, Random Forest, Gradient Boosting, Extreme Gradient Boosting (XGBoost), and Naïve Bayes were applied to evaluate HBeAg seroconversion in HBeAg-positive CHB patients receiving PegIFN-α monotherapy in the training dataset and testing dataset. RESULTS: XGBoost model showed the best performance, which had largest AUROC (0.900, 95% CI: 0.85-0.95 and 0.910, 95% CI: 0.84-0.98, in training dataset and testing dataset, respectively), and the best calibration curve performance to predict HBeAg seroconversion. The importance of XGBoost model indicated that treatment time contributed greatest to HBeAg seroconversion, followed by HBV DNA(log), HBeAg, HBeAb, HBcAb, ALT, triglyceride, and ALP. CONCLUSIONS: XGBoost model based on common laboratory variables had good performance in predicting HBeAg seroconversion in HBeAg-positive CHB patients receiving PegIFN-α monotherapy.

Routine laboratory parameters in patients with necrotizing pancreatitis by the time of operative pancreatic debridement: Food for thought.

BACKGROUND: Timing of invasive intervention such as operative pancreatic debridement (OPD) in patients with acute necrotizing pancreatitis (ANP) is linked to the degree of encapsulation in necrotic collections and controlled inflammation. Additional markers of these processes might assist decision-making on the timing of surgical intervention. In our opinion, it is logical to search for such markers among routine laboratory parameters traditionally used in ANP patients, considering simplicity and cost-efficacy of routine laboratory methodologies. AIM: To evaluate laboratory variables in ANP patients in the preoperative period for the purpose of their use in the timing of surgery. METHODS: A retrospective analysis of routine laboratory parameters in 53 ANP patients undergoing OPD between 2017 and 2020 was performed. Dynamic changes of routine hematological and biochemical indices were examined in the preoperative period. Patients were divided into survivors and non-survivors. Survivors were divided into subgroups with short and long post-surgery length of stay (LOS) in hospital. Correlation analysis was used to evaluate association of laboratory variables with LOS. Logistic regression was used to assess risk factors for patient mortality. RESULTS: Seven patients (15%) with severe acute pancreatitis (SAP) and 46 patients (85%) with moderately SAP (MSAP) were included in the study. Median age of participants was 43.2 years; 33 (62.3%) were male. Pancreatitis etiology included biliary (15%), alcohol (80%), and idiopathic/other (5%). Median time from diagnosis to OPD was ≥ 4 wk. Median postoperative LOS was at the average of 53 d. Mortality was 19%. Progressive increase of platelet count in preoperative period was associated with shortened LOS. Increased aspartate aminotransferase and direct bilirubin (DB) levels the day before the OPD along with weak progressive decrease of DB in preoperative period were reliable predictors for ANP patient mortality. CONCLUSION: Multifactorial analysis of dynamic changes of routine laboratory variables can be useful for a person-tailored timing of surgical intervention in ANP patients.

Individual outcome prediction models for patients with COVID-19 based on their first day of admission to the intensive care unit.

BACKGROUND: Currently, good prognosis and management of critically ill patients with COVID-19 are crucial for developing disease management guidelines and providing a viable healthcare system. We aimed to propose individual outcome prediction models based on binary logistic regression (BLR) and artificial neural network (ANN) analyses of data collected in the first 24 h of intensive care unit (ICU) admission for patients with COVID-19 infection. We also analysed different variables for ICU patients who survived and those who died. METHODS: Data from 326 critically ill patients with COVID-19 were collected. Data were captured on laboratory variables, demographics, comorbidities, symptoms and hospital stay related information. These data were compared with patient outcomes (survivor and non-survivor patients). BLR was assessed using the Wald Forward Stepwise method, and the ANN model was constructed using multilayer perceptron architecture. RESULTS: The area under the receiver operating characteristic curve of the ANN model was significantly larger than the BLR model (0.917 vs 0.810; p < 0.001) for predicting individual outcomes. In addition, ANN model presented similar negative predictive value than the BLR model (95.9% vs 94.8%). Variables such as age, pH, potassium ion, partial pressure of oxygen, and chloride were present in both models and they were significant predictors of death in COVID-19 patients. CONCLUSIONS: Our study could provide helpful information for other hospitals to develop their own individual outcome prediction models based, mainly, on laboratory variables. Furthermore, it offers valuable information on which variables could predict a fatal outcome for ICU patients with COVID-19.

Comparison of the community-acquired pneumonia and COVID-19 at the early stage: findings from two cohort studies.

BACKGROUND: The current focus is largely on whole course medical management of coronavirus disease-19 (COVID-19) with real-time polymerase chain reaction (RT-PCR) and radiological features, while the mild cases are usually missed. Thus, combination of multiple diagnostic methods is urgent to understand COVID-19 fully and to monitor the progression of COVID-19. METHODS: laboratory variables of 40 mild COVID-19 patients, 30 patients with community-acquired pneumonia (CAP) and 32 healthy individuals were analyzed by principal component analysis (PCA), Kruskal test, Procrustes test, the vegan package in R, CCA package and receiver operating characteristic to investigate the characteristics of the laboratory variables and their relationships in COVID-19. RESULTS: The correlations between the laboratory variables presented a variety of intricate linkages in the COVID-19 group compared with the healthy group and CAP patient group. The prediction probability of the combination of lymphocyte count (LY), eosinophil (EO) and platelets (PLT) was 0.847, 0.854 for the combination of lactate (LDH), creatine kinase isoenzyme (CK-MB), and C-reactive protein (CRP), 0.740 for the combination of EO, white blood cell count (WBC) and neutrophil count (NEUT) and 0.872 for the combination of CK-MB and P. CONCLUSIONS: The correlations between the laboratory variables in the COVID-19 group could be a unique characteristic showing promise as a method for COVID-19 prediction and monitoring progression of COVID-19 infection.

Influence of general anaesthesia on the intravenous acetaminophen pharmacokinetics in Beagle dogs.

OBJECTIVE: To determine if general anaesthesia influences the intravenous (IV) pharmacokinetics (PK) of acetaminophen in dogs. STUDY DESIGN: Prospective, crossover, randomized experimental study. ANIMALS: A group of nine healthy Beagle dogs. METHODS: Acetaminophen PK were determined in conscious and anaesthetized dogs on two separate occasions. Blood samples were collected before, and at 5, 10, 15, 30, 45, 60 and 90 minutes and 2, 3, 4, 6, 8, 12 and 24 hours after 20 mg kg-1 IV acetaminophen administration. Haematocrit, total proteins, albumin, alanine aminotransferase, aspartate aminotransferase, urea and creatinine were determined at baseline and 24 hours after acetaminophen. The anaesthetized group underwent general anaesthesia (90 minutes) for dental cleaning. After the administration of dexmedetomidine (3 µg kg-1) intramuscularly, anaesthesia was induced with propofol (2-3 mg kg-1) IV, followed by acetaminophen administration. Anaesthesia was maintained with isoflurane in 50% oxygen (Fe'Iso 1.3-1.5%). Dogs were mechanically ventilated. Plasma concentrations were analysed with high-performance liquid chromatography. PK analysis was undertaken using compartmental modelling. A Wilcoxon test was used to compare PK data between groups, and clinical laboratory values between groups, and before versus 24 hours after acetaminophen administration. Data are presented as median and range (p < 0.05). RESULTS: A two-compartmental model best described time-concentration profiles of acetaminophen. No significant differences were found for volume of distribution values 1.41 (0.94-3.65) and 1.72 (0.89-2.60) L kg-1, clearance values 1.52 (0.71-2.30) and 1.60 (0.91-1.78) L kg-1 hour-1 or terminal elimination half-life values 2.45 (1.45-8.71) and 3.57 (1.96-6.35) hours between conscious and anaesthetized dogs, respectively. Clinical laboratory variables were within normal range. No adverse effects were recorded. CONCLUSIONS AND CLINICAL RELEVANCE: IV acetaminophen PK in healthy Beagle dogs were unaffected by general anaesthesia under the study conditions. Further studies are necessary to evaluate the PK in different clinical contexts.

Development and Validation of a LASSO Prediction Model for Better Identification of Ischemic Stroke: A Case-Control Study in China.

BACKGROUND: Timely diagnosis of ischemic stroke (IS) in the acute phase is extremely vital to achieve proper treatment and good prognosis. In this study, we developed a novel prediction model based on the easily obtained information at initial inspection to assist in the early identification of IS. METHODS: A total of 627 patients with IS and other intracranial hemorrhagic diseases from March 2017 to June 2018 were retrospectively enrolled in the derivation cohort. Based on their demographic information and initial laboratory examination results, the prediction model was constructed. The least absolute shrinkage and selection operator algorithm was used to select the important variables to form a laboratory panel. Combined with the demographic variables, multivariate logistic regression was performed for modeling, and the model was encapsulated within a visual and operable smartphone application. The performance of the model was evaluated on an independent validation cohort, formed by 304 prospectively enrolled patients from June 2018 to May 2019, by means of the area under the curve (AUC) and calibration. RESULTS: The prediction model showed good discrimination (AUC = 0.916, cut-off = 0.577), calibration, and clinical availability. The performance was reconfirmed in the more complex emergency department. It was encapsulated as the Stroke Diagnosis Aid app for smartphones. The user can obtain the identification result by entering the values of the variables in the graphical user interface of the application. CONCLUSION: The prediction model based on laboratory and demographic variables could serve as a favorable supplementary tool to facilitate complex, time-critical acute stroke identification.

The B-lymphocyte chemokine CXCL13 in the cerebrospinal fluid of children with Lyme neuroborreliosis: associations with clinical and laboratory variables.

Background: The B-lymphocyte chemokine CXCL13 is increasingly considered as a useful early phase diagnostic marker of Lyme neuroborreliosis (LNB). However, the large variation in level of CXCL13 in the cerebrospinal fluid (CSF) observed in LNB patients is still unexplained. We aimed to identify factors associated with the level of CXCL13 in children with LNB, possibly improving the interpretation of CXCL13 as a diagnostic marker of LNB. Methods: Children with confirmed and probable LNB were included in a prospective study on CXCL13 in CSF as a diagnostic marker of LNB. The variables age, sex, facial nerve palsy, generalized inflammation symptoms (fever, headache, neck-stiffness and/or fatigue), duration of symptoms, Borrelia antibodies in CSF, Borrelia antibody index (AI), CSF white blood cells (WBC), CSF protein and detection of the genospecies Borrelia garinii by PCR were included in simple and multivariable regression analyses to study the associations with the CXCL13 level. Results: We included 53 children with confirmed and 17 children with probable LNB. CXCL13 levels in CSF were positively associated with WBC, protein and Borrelia antibodies in CSF in both simple and multivariable analyses. We did not find any associations between CXCL13 and age, sex, clinical symptoms, duration of symptoms, AI or the detection of Borrelia garinii. Conclusions: High levels of CSF CXCL13 are present in the early phase of LNB and correlate with the level of CSF WBC and protein. Our results indicate that CSF CXCL13 in children evaluated for LNB can be interpreted independently of clinical features or duration of symptoms.

Development of a mortality score to assess risk of adverse drug reactions among hospitalized patients with moderate to severe chronic kidney disease.

BACKGROUND: Chronic kidney disease (CKD) is a significant health burden that increases the risk of adverse events. Currently, there is no validated models to predict risk of mortality among CKD patients experienced adverse drug reactions (ADRs) during hospitalization. This study aimed to develop a mortality risk prediction model among hospitalized CKD patients whom experienced ADRs. METHODS: Patients data with CKD stages 3-5 admitted at various wards were included in the model development. The data collected included demographic characteristics, comorbid conditions, laboratory tests and types of medicines taken. Sequential series of logistic regression models using mortality as the dependent variable were developed. Bootstrapping method was used to evaluate the model's internal validation. Variables odd ratio (OR) of the best model were used to calculate the predictive capacity of the risk scores using the area under the curve (AUC). RESULTS: The best prediction model included comorbidities heart disease, dyslipidaemia and electrolyte imbalance; psychotic agents; creatinine kinase; number of total medication use; and conservative management (Hosmer and Lemeshow test =0.643). Model performance was relatively modest (R square = 0.399) and AUC which determines the risk score's ability to predict mortality associated with ADRs was 0.789 (95% CI, 0.700-0.878). Creatinine kinase, followed by psychotic agents and electrolyte disorder, was most strongly associated with mortality after ADRs during hospitalization. This model correctly predicts 71.4% of all mortality pertaining to ADRs (sensitivity) and with specificity of 77.3%. CONCLUSION: Mortality prediction model among hospitalized stages 3 to 5 CKD patients experienced ADR was developed in this study. This prediction model adds new knowledge to the healthcare system despite its modest performance coupled with its high sensitivity and specificity. This tool is clinically useful and effective in identifying potential CKD patients at high risk of ADR-related mortality during hospitalization using routinely performed clinical data.