External Cross-validation Research Articles

An Ensemble Machine Learning Model to Enhance Extrapolation Ability of Predicting Coarse Particulate Matter with High Resolutions in China.

Accurate exposure assessment is important for conducting PM10-2.5-related epidemiological studies, which have been limited thus far. In this study, we aimed to develop an ensemble machine learning method to estimate PM10-2.5 concentrations in mainland China during 2013-2020. The study was conducted in two stages. In the first stage, we developed two methods: the indirect method refers to developing models for PM2.5 and PM10 separately and subsequently calculating PM10-2.5 as the difference between them; and the direct method refers to establishing a model between PM10-2.5 measurements and relevant predictors directly. In the second stage, we employed an ensemble method by integrating predictions from both indirect and direct methods. Internal and external cross-validation (CV) were performed to validate the extrapolation capacity of models. The ensemble method demonstrated enhanced extrapolation accuracy in both internal and external CV compared to indirect and direct methods. The predictions produced by the ensemble method captured the spatiotemporal pattern of PM10-2.5, even in the sand and dust storm seasons. Our study introduces an ensemble strategy leveraging the strengths of both indirect and direct methods to estimate PM10-2.5 concentrations, which holds significant potential to support future epidemiological studies to address knowledge gaps in understanding the health effects of PM10-2.5.

A generalized machine learning framework to estimate fatigue life across materials with minimal data

In this research, a generalized machine learning (ML) framework is proposed to estimate the fatigue life of epoxy polymers and additively manufactured AlSi10Mg alloy materials, leveraging their failure surface void characteristics. An extreme gradient boosting algorithm-based ML framework encompassing Synthetic Minority Over-sampling TEchnique (SMOTE), categorical data encoding, and external loop cross-validation is developed to evaluate the fatigue life across materials. The influence of different training strategies based on materials, input features, encoding method, and data standardization on the model performance is explored. Additionally, the importance of anti-data-leakage and anti-overfitting measures over the ML model performance is addressed. The result shows that the data-leakage-free, external loop cross-validated model can estimate the fatigue life of selective epoxy polymers and metal alloys with an average R2 of 0.71 ± 0.06 using a mere 12 to 27 experimental data points per material category. Whereas the model trained with data-leakage and overfitting results in high R2 of 0.9.

Data-driven models to predict the water-to-cement ratio and initial setting time of cement grouts

ABSTRACT As a common defect, the water–cement ratio of grout can currently be monitored only during the maintenance phase, which limits the repair methods and misses the optimal opportunity for repairs. To overcome this limitation, this study integrated ultrasonic parameters previously used to characterise cement-based materials and developed a new Initial Setting Time and Water–Cement Ratio (IST_WCR) risk model to predict the setting time and water–cement ratio grout using machine learning (ML) algorithms. Experiments on grout involved four different water–cement ratios, ranging from 0.11 to 0.18. A data-driven method based on ML was used to extract predictive factors from eight ultrasonic parameters, including the speed, energy, main frequency, and main frequency amplitude of P-waves and S-waves, and to evaluate multiple ML classifiers to establish the IST_WCR risk prediction model. This model underwent internal and external cross-validations and demonstrated very strong performance with a Brier score of under 0.01. The dataset for ML classifiers contained a total of 956 signals and 7648 features. Compared with traditional methods, this method can automatically characterise the setting process of grout and identify defective water–cement ratios at a very early stage of curing.

Prioritising deteriorating patients using time-to-event analysis: prediction model development and internal–external validation

BackgroundBinary classification models are frequently used to predict clinical deterioration, however they ignore information on the timing of events. An alternative is to apply time-to-event models, augmenting clinical workflows by ranking patients by predicted risks. This study examines how and why time-to-event modelling of vital signs data can help prioritise deterioration assessments using lift curves, and develops a prediction model to stratify acute care inpatients by risk of clinical deterioration.MethodsWe developed and validated a Cox regression for time to in-hospital mortality. The model used time-varying covariates to estimate the risk of clinical deterioration. Adult inpatient medical records from 5 Australian hospitals between 1 January 2019 and 31 December 2020 were used for model development and validation. Model discrimination and calibration were assessed using internal–external cross validation. A discrete-time logistic regression model predicting death within 24 h with the same covariates was used as a comparator to the Cox regression model to estimate differences in predictive performance between the binary and time-to-event outcome modelling approaches.ResultsOur data contained 150,342 admissions and 1016 deaths. Model discrimination was higher for Cox regression than for discrete-time logistic regression, with cross-validated AUCs of 0.96 and 0.93, respectively, for mortality predictions within 24 h, declining to 0.93 and 0.88, respectively, for mortality predictions within 1 week. Calibration plots showed that calibration varied by hospital, but this can be mitigated by ranking patients by predicted risks.ConclusionTime-varying covariate Cox models can be powerful tools for triaging patients, which may lead to more efficient and effective care in time-poor environments when the times between observations are highly variable.

Relationship between body weight and hip width in dairy buffaloes (Bubalus bubalis).

The objective of the present study was to evaluate the relationship between body weight (BW) and hip width (HW) in dairy buffaloes (Bubalus bubalis). HW was measured in 215 Murrah buffaloes with a BW of 341 ± 161.6 kg, aged between three months and five years, and raised in southeastern Mexico. Linear and non-linear regressions were used to construct the prediction models. The goodness of fit of the models was evaluated using the Akaike information criterion (AIC), Bayesian information criterion (BIC), coefficient of determination (R2), mean squared error (MSE), and root MSE (RMSE). Additionally, the developed models were evaluated through internal and external cross-validation (k-folds) using independent data. The ability of the fitted models to predict the observed values was assessed based on the root mean square error of prediction (RMSEP), R2, and mean absolute error (MAE). The relationship between BW and HW showed a high correlation coefficient (r = 0.96, P < 0.001). The chosen fitted model to predict BW was: -176.33 (± 40.83***) + 8.74 (± 1.79***) × HW + 0.04 (± 0.01*) × HW2, because it presented the lowest MSE, RMSE, and AIC values, which were 1228.64, 35.05 and 1532.41, respectively. Therefore, with reasonable accuracy, the quadratic model using hip width may be suitable for predicting body weight in buffaloes.

Predicting the risk of pancreatic cancer in adults with new-onset diabetes: development and internal–external validation of a clinical risk prediction model

BackgroundThe National Institute for Health and Care Excellence (NICE) recommends that people aged 60+ years with newly diagnosed diabetes and weight loss undergo abdominal imaging to assess for pancreatic cancer. More nuanced stratification could lead to enrichment of these referral pathways.MethodsPopulation-based cohort study of adults aged 30–85 years at type 2 diabetes diagnosis (2010–2021) using the QResearch primary care database in England linked to secondary care data, the national cancer registry and mortality registers. Clinical prediction models were developed to estimate risks of pancreatic cancer diagnosis within 2 years and evaluated using internal–external cross-validation.ResultsSeven hundred and sixty-seven of 253,766 individuals were diagnosed with pancreatic cancer within 2 years. Models included age, sex, BMI, prior venous thromboembolism, digoxin prescription, HbA1c, ALT, creatinine, haemoglobin, platelet count; and the presence of abdominal pain, weight loss, jaundice, heartburn, indigestion or nausea (previous 6 months). The Cox model had the highest discrimination (Harrell’s C-index 0.802 (95% CI: 0.797–0.817)), the highest clinical utility, and was well calibrated. The model’s highest 1% of predicted risks captured 12.51% of pancreatic cancer cases. NICE guidance had 3.95% sensitivity.DiscussionA new prediction model could have clinical utility in identifying individuals with recent onset diabetes suitable for fast-track abdominal imaging.

Prospective virtual screening combined with bio-molecular simulation enabled identification of new inhibitors for the KRAS drug target

Lung cancer is a disease with a high mortality rate and it is the number one cause of cancer death globally. Approximately 12–14% of non-small cell lung cancers are caused by mutations in KRASG12C. The KRASG12C is one of the most prevalent mutants in lung cancer patients. KRAS was first considered undruggable. The sotorasib and adagrasib are the recently approved drugs that selectively target KRASG12C, and offer new treatment approaches to enhance patient outcomes however drug resistance frequently arises. Drug development is a challenging, expensive, and time-consuming process. Recently, machine-learning-based virtual screening are used for the development of new drugs. In this study, we performed machine-learning-based virtual screening followed by molecular docking, all atoms molecular dynamics simulation, and binding energy calculations for the identifications of new inhibitors against the KRASG12C mutant. In this study, four machine learning models including, random forest, k-nearest neighbors, Gaussian naïve Bayes, and support vector machine were used. By using an external dataset and 5-fold cross-validation, the developed models were validated. Among all the models the performance of the random forest (RF) model was best on the train/test dataset and external dataset. The random forest model was further used for the virtual screening of the ZINC15 database, in-house database, Pakistani phytochemicals, and South African Natural Products database. A total of 100 ns MD simulation was performed for the four best docking score complexes as well as the standard compound in complex with KRASG12C. Furthermore, the top four hits revealed greater stability and greater binding affinities for KRASG12C compared to the standard drug. These new hits have the potential to inhibit KRASG12C and may help to prevent KRAS-associated lung cancer. All the datasets used in this study can be freely available at (https://github.com/Amar-Ajmal/Datasets-for-KRAS).

Development and validation of a prediction score for failure to casirivimab/imdevimab in hospitalized patients with COVID-19 pneumonia.

Casirivimab and imdevimab (CAS/IMV) are two non-competing, high-affinity human IgG1 anti-SARS-CoV-2 monoclonal antibodies, that showed a survival benefit in seronegative hospitalized patients with COVID-19. This study aimed to estimate the day-28 risk of mechanical ventilation (MV) and death in individuals hospitalized for severe COVID-19 pneumonia and receiving CAS/IMV. Additionally, it aimed to identify variables measured at the time of hospital admission that could predict these outcomes and derive a prediction algorithm. This is a retrospective, observational cohort study conducted in 12 hospitals in Italy. Adult patients who were consecutively hospitalized from November 2021 to February 2022 receiving CAS/IMV were included. A multivariable logistic regression model was used to identify predictors of MV or death by day 28 from treatment initiation, and β-coefficients from the model were used to develop a risk score that was derived by means of leave-one-out internal cross-validation (CV), external CV, and calibration. Secondary outcome was mortality. A total of 480 hospitalized patients in the training set and 157 patients in the test set were included. By day 28, 36 participants (8%) underwent MV and 28 died (6%) for a total of 58 participants (12%) experiencing the composite primary endpoint. In multivariable analysis, four factors [age, PaO2/FiO2 ratio, lactate dehydrogenase (LDH), and platelets] were independently associated with the risk of MV/death and were used to generate the proposed risk score. The accuracy of the score in the area under the curve (AUC) was 0.80 and 0.77 in internal validation and test for the composite endpoint and 0.87 and 0.86 for death, respectively. The model also appeared to be well calibrated with the raw data. The mortality risk reported in our study was lower than that previously reported. Although CAS/IMV is no longer used, our score might help in identifying which patients are not likely to benefit from monoclonal antibodies and may require alternative interventions.

In-silico design novel phenylsulfonyl furoxan and phenstatin derivatives as multi-target anti-cancer inhibitors based on 2D-QSAR, molecular docking, dynamics and ADMET approaches

ABSTRACT A series of 31 hybrid of phenylsulfonyl furoxan and phenstatin (1a-j, 2a-j, 3a-j, 4 and 5) derivatives, were computationally studied as potential anti-cancer inhibitors against four cell lines, i.e. A2780, MDA-MB-231, HCT-116 and A549. In this work, the 2D-QSAR approach combining the multiple linear regression (MLR) model, and internal and external cross-validation, showed a satisfactory quality factor: R 2 = 0.85, 0.74, 0.82 and 0.75 for the four cell lines, respectively. The binding affinity of the hybrid agents towards the four 4GL7, 6GUE, 1M17 and 4XL7 anti-tumoral targets was further evaluated using molecular docking and dynamics simulations (0–200 ns). The dynamics assessment parameters indicated the formation of satisfactorily stable complexes. In addition, all considered data sets show that the best binding affinity, including the highest docking score, hydrogen bond energy and amino acid steric interactions are well predicted for the best-selected complexes. The developed 2D-QSAR model was leveraged to design and predict the biological activity of 12 new hybrid compounds (N1–N12) based on the best in-vivo inhibitor, namely, the 3 h ligand of the formula: (4-((1-(2-((4-((3crylamidophenyl)amino)quinazolin-2-yl)thio)acetyl)piperidin-4-yl)oxy)-3-(phenylsulfonyl)-1,2,5-oxadiazole 2-oxide). Multitargeting docking scores and dynamics simulations show that they exhibit satisfactorily potent anti-tumoral inhibition abilities towards the four proteins. Our in-silico outcomes would be combined with in-vitro and in-vivo studies to provide a perspective on the validation of their anti-cancer activity. In particular, the ADMET predictions indicate that four new designed ligands have demonstrated a good drug-like profile and can be considered prospective candidates for future anti-cancer therapies.

Development and validation of a machine learning-based predictive model for assessing the 90-day prognostic outcome of patients with spontaneous intracerebral hemorrhage

BackgroundSpontaneous intracerebral hemorrhage (sICH) is associated with significant mortality and morbidity. Predicting the prognosis of patients with sICH remains an important issue, which significantly affects treatment decisions. Utilizing readily available clinical parameters to anticipate the unfavorable prognosis of sICH patients holds notable clinical significance. This study employs five machine learning algorithms to establish a practical platform for the prediction of short-term prognostic outcomes in individuals afflicted with sICH.MethodsWithin the framework of this retrospective analysis, the model underwent training utilizing data gleaned from 413 cases from the training center, with subsequent validation employing data from external validation center. Comprehensive clinical information, laboratory analysis results, and imaging features pertaining to sICH patients were harnessed as training features for machine learning. We developed and validated the model efficacy using all the selected features of the patients using five models: Support Vector Machine (SVM), Logistic Regression (LR), Random Forest (RF), XGboost and LightGBM, respectively. The process of Recursive Feature Elimination (RFE) was executed for optimal feature screening. An internal five-fold cross-validation was employed to pinpoint the most suitable hyperparameters for the model, while an external five-fold cross-validation was implemented to discern the machine learning model demonstrating the superior average performance. Finally, the machine learning model with the best average performance is selected as our final model while using it for external validation. Evaluation of the machine learning model’s performance was comprehensively conducted through the utilization of the ROC curve, accuracy, and other relevant indicators. The SHAP diagram was utilized to elucidate the variable importance within the model, culminating in the amalgamation of the above metrics to discern the most succinct features and establish a practical prognostic prediction platform.ResultsA total of 413 patients with sICH patients were collected in the training center, of which 180 were patients with poor prognosis. A total of 74 patients with sICH were collected in the external validation center, of which 26 were patients with poor prognosis. Within the training set, the test set AUC values for SVM, LR, RF, XGBoost, and LightGBM models were recorded as 0.87, 0.896, 0.916, 0.885, and 0.912, respectively. The best average performance of the machine learning models in the training set was the RF model (average AUC: 0.906 ± 0.029, P < 0.01). The model still maintains a good performance in the external validation center, with an AUC of 0.817 (95% CI 0.705–0.928). Pertaining to feature importance for short-term prognostic attributes of sICH patients, the NIHSS score reigned supreme, succeeded by AST, Age, white blood cell, and hematoma volume, among others. In culmination, guided by the RF model’s variable importance weight and the model's ROC curve insights, the NIHSS score, AST, Age, white blood cell, and hematoma volume were integrated to forge a short-term prognostic prediction platform tailored for sICH patients.ConclusionWe constructed a prediction model based on the results of the RF model incorporating five clinically accessible predictors with reliable predictive efficacy for the short-term prognosis of sICH patients. Meanwhile, the performance of the external validation set was also more stable, which can be used for accurate prediction of short-term prognosis of sICH patients.

Investigating Factors Associated with the Development of Postnatal Depression After Cesarean Delivery: A Validation Cohort Study.

This study aimed to validate a proposed association model previously published to determine the clinical relevance of pre-operative determinants in the development of PND after Cesarean delivery (CD). Parturients undergoing elective CD under neuraxial anesthesia were recruited for a prospective cohort study between Oct 2021 and Oct 2022 at KK Women's and Children's Hospital, Singapore. Predelivery pain, psychological and mechanical temporal summation, and demographic data were recorded. A follow-up survey was conducted at 6 to 10 weeks after CD. The primary outcome was the incidence of PND, defined as an Edinburgh Postnatal Depression Scale (EPDS) score ≥ 10. A total of 180 patients were recruited for validation. PND 6 to 10 weeks post-delivery occurred in 18.9% of recruited parturients. Multivariate regression analyses showed that higher pre-operative CSI scores (p=0.0156), higher anxiety levels about upcoming surgery (p=0.0429), increased pre-operative pain scores on movement (p=0.0110), and higher pre-operative HADS subscale scores on anxiety (p=0.0041) were independently associated with the development of PND weeks post-CD. Lower anticipation of pain medication needs (p=0.0038) was independently associated with the development of PND post-CD. The area under curve (AUC) of this multivariable model (training cohort), internal cross validation (training cohort) and external cross validation (validation cohort) were 0.818 (95% CI, 0.746 to 0.889), 0.785 (95% CI, 0.707 to 0.864) and 0.604 (95% CI, 0.497 to 0.710) respectively. The proposed model performed well in a local population. Further refinement is necessary to test the proposed model in populations with social and cultural differences.

Estimating reference crop evapotranspiration using improved convolutional bidirectional long short-term memory network by multi-head attention mechanism in the four climatic zones of China

Accurate reference crop evapotranspiration (ET0) estimation is essential for agricultural water management, crop productivity, and irrigation systems. As the standard ET0 estimation method, the Penman-Monteith equation has been widely recommended worldwide. However, its application is still restricted to comprehensive meteorological data deficiency, making the exploration of alternative simpler models for acceptable ET0 estimation highly meaningful. Concerning the aforementioned requirement, this study developed the novel deep learning model (MA-CNN-BiLSTM), which incorporates Multi-Head Attention mechanism (MA), Convolutional Neural Network (CNN), and Bidirectional Long Short-Term Memory network (BiLSTM) as intricate relationship processor, feature extractor, and regression component, to estimate ET0 based on radiation-based (Rn-based), humidity-based (RH-based), and temperature-based (T-based) input combinations at 600 stations during 1961–2020 throughout China under internal and external cross-validation strategies. Besides, through a comparative evaluation among MA-CNN-BiLSTM, CNN-BiLSTM, BiLSTM, LSTM, Multivariate Adaptive Regression Splines (MARS), and empirical models, the result indicated that MA-CNN-BiLSTM achieved superior precision, with values of Determination Coefficient (R2), Nash–Sutcliffe efficiency coefficient (NSE), Relative Root Mean Square Error (RRMSE), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE) ranging 0.877–0.972, 0.844–0.962, 0.129–0.292, 0.294–0.644 mm d−1, 0.244–0.566 mm d−1 for internal strategy and 0.797–0.927, 0.786–0.920, 0.162–0.335, 0.409–0.969 mm d−1, 0.294–0.699 mm d−1 for external strategy. Specifically, Rn-based MA-CNN-BiLSTM excelled in the temperate continental zone (TCZ) and mountain plateau zone (MPZ), while RH-based MA-CNN-BiLSTM yielded best precision in others. Furthermore, the internal strategy was superior to external strategy by 2.74–106.04% for R2, 1.11–120.49% for NSE, 1.41–40.27% for RRMSE, 1.68–45.53% for RMSE, and 1.21–38.87% for MAE, respectively. In summary, the main contribution of the present study is the proposal of a novel LSTM-type ET0 model (MA-CNN-BiLSTM) to cope with various data-missing scenarios throughout China, which can provide effective support for decision-making in regional agriculture water management.

Development of 2-dimensional and 3-dimensional QSAR models of Indazole derivatives as TTK inhibitors having Anticancer potential

The study aimed to explore the anticancer efficacy of indazole pharmacophore by analyzing a series of 109 derivatives of indazole as Tyrosine Threonine Kinase (TTK) inhibitors through quantitative activity relationship analysis using 2D and 3D QSAR techniques. The best 2D-QSAR model was generated by the MLR method, showing a high correlation coefficient (r2) of 0.9512, and good internal (q2), and external (pred_r2) cross-validation regression coefficients of 0.8998, and 0.8661, respectively. The residual values were modest, indicating good agreement between the observed and predicted pIC50 values, which suggested that the chosen model was predictably accurate. The 3D QSAR model, built using the SWF kNN approach, displayed a high internal cross-validation regression coefficient (q2) of 0.9132. Essential structural features/considerations in developing indazole as prospective anticancer medicines have been suggested. The study provides a reliable and predictive model for the prediction of anticancer activity of indazole derivatives. The identified essential structural features/considerations may be useful for the development of prospective anticancer medicines.

Research on the Anti-tumor Activity of Novel Histone Deacetylase Inhibitors based on 3D QSAR Model

Background: Histone deacetylases (HDACs) pertain to the category of Zn2 + or nicotinamide adenine dinucleotide (NAD+)- dependent proteolytic enzymes. While the antitumor effect of HDAC inhibitors alone has been demonstrated and the effect of HDAC inhibitors on solid tumors is not ideal, which considerably limits their clinical use. Therefore, the search for novel HDAC inhibitors equipped with specific inhibitors is extremely urgent and necessary. Methods: 3D-QSAR was employed to investigate insights into the crucial structural element that effect the activity of novel HDAC small molecule inhibitors. The best saliency CoMFA and CoMSIA models are obtained using 55 molecules in the training set and 16 molecules in the test set. Results: The statistical quality of the generated model is demonstrated by internal and external cross-validations. The CoMFA model obtained satisfactory values (q2 = 0.664, r2 = 0.917, SEE = 0.217) while optimized CoMSIA model exceed with (q2 = 0.672, r2 = 0.948, SEE = 0.175). Conclusion: The statistical parameters from 3D-QSAR models reveal that the results are reliable and significant with strong predictive ability. These theoretical results may contribute to the design of novel HDAC small molecule inhibitors with enhanced activity for the treatment of cancer.

The use of simple structural parameters of organic compounds to assess their PUF-air partition coefficients

A novel approach is introduced for the reliable prediction of PUF-air partition coefficients of organic compounds, which can determine the environmental fate of organic compounds during interactions with air, soil, and water. The biggest accessible measured data of PUF-air partition coefficients for 170 chemicals are used to develop and test the novel model. In comparison to available quantitative structure-property relationship (QSPR) methods for the prediction of PUF-air partition coefficients that need complex descriptors, the here used descriptors are simpler. The assessed various statistical factors of the simple method containing 147 (training) and 23 (test) organic compounds can verify the external and internal cross-validations. Various statistical parameters confirm the high reliability of the novel model as compared with the outputs of complex multiple linear regression (MLR), artificial neural network (ANN) and support vector machine (SVM) methods. The values of R-squared (R2), and root mean square error (RMSE) of the new model are for training/test sets are 0.924/0.894 and 0.374/0.318, respectively. Meanwhile, R2 and RMSE values for three comparative models training/test sets are (i) MLR: 0.848/0.670 (R2) and 0.531/0.573 (RMSE); (ii) ANN: 0.902/0.664 (R2) and 0.425/0.560 (RMSE); (iii) SVM: 0.935/0.794 (R2) and 0.351/0.419 (RMSE). Thus, the new model the simplest approach with higher reliability in comparison to the best available methods.

Development and internal validation of a clinical prediction model for serious complications after emergency laparotomy

PurposeEmergency laparotomy (EL) is a common operation with high risk for postoperative complications, thereby requiring accurate risk stratification to manage vulnerable patients optimally. We developed and internally validated a predictive model of serious complications after EL.MethodsData for eleven carefully selected candidate predictors of 30-day postoperative complications (Clavien-Dindo grade > = 3) were extracted from the HELAS cohort of EL patients in 11 centres in Greece and Cyprus. Logistic regression with Least Absolute Shrinkage and Selection Operator (LASSO) was applied for model development. Discrimination and calibration measures were estimated and clinical utility was explored with decision curve analysis (DCA). Reproducibility and heterogeneity were examined with Bootstrap-based internal validation and Internal–External Cross-Validation. The American College of Surgeons National Surgical Quality Improvement Program’s (ACS-NSQIP) model was applied to the same cohort to establish a benchmark for the new model.ResultsFrom data on 633 eligible patients (175 complication events), the SErious complications After Laparotomy (SEAL) model was developed with 6 predictors (preoperative albumin, blood urea nitrogen, American Society of Anaesthesiology score, sepsis or septic shock, dependent functional status, and ascites). SEAL had good discriminative ability (optimism-corrected c-statistic: 0.80, 95% confidence interval [CI] 0.79–0.81), calibration (optimism-corrected calibration slope: 1.01, 95% CI 0.99–1.03) and overall fit (scaled Brier score: 25.1%, 95% CI 24.1–26.1%). SEAL compared favourably with ACS-NSQIP in all metrics, including DCA across multiple risk thresholds.ConclusionSEAL is a simple and promising model for individualized risk predictions of serious complications after EL. Future external validations should appraise SEAL’s transportability across diverse settings.

Denoising Autoencoder Normalization for Large-Scale Untargeted Metabolomics by Gas Chromatography-Mass Spectrometry.

Large-scale metabolomics assays are widely used in epidemiology for biomarker discovery and risk assessments. However, systematic errors introduced by instrumental signal drifting pose a big challenge in large-scale assays, especially for derivatization-based gas chromatography-mass spectrometry (GC-MS). Here, we compare the results of different normalization methods for a study with more than 4000 human plasma samples involved in a type 2 diabetes cohort study, in addition to 413 pooled quality control (QC) samples, 413 commercial pooled plasma samples, and a set of 25 stable isotope-labeled internal standards used for every sample. Data acquisition was conducted across 1.2 years, including seven column changes. In total, 413 pooled QC (training) and 413 BioIVT samples (validation) were used for normalization comparisons. Surprisingly, neither internal standards nor sum-based normalizations yielded median precision of less than 30% across all 563 metabolite annotations. While the machine-learning-based SERRF algorithm gave 19% median precision based on the pooled quality control samples, external cross-validation with BioIVT plasma pools yielded a median 34% relative standard deviation (RSD). We developed a new method: systematic error reduction by denoising autoencoder (SERDA). SERDA lowered the median standard deviations of the training QC samples down to 16% RSD, yielding an overall error of 19% RSD when applied to the independent BioIVT validation QC samples. This is the largest study on GC-MS metabolomics ever reported, demonstrating that technical errors can be normalized and handled effectively for this assay. SERDA was further validated on two additional large-scale GC-MS-based human plasma metabolomics studies, confirming the superior performance of SERDA over SERRF or sum normalizations.

Interpretable machine learning for predicting 28-day all-cause in-hospital mortality for hypertensive ischemic or hemorrhagic stroke patients in the ICU: a multi-center retrospective cohort study with internal and external cross-validation.

Timely and accurate outcome prediction plays a critical role in guiding clinical decisions for hypertensive ischemic or hemorrhagic stroke patients admitted to the ICU. However, interpreting and translating the predictive models into clinical applications are as important as the prediction itself. This study aimed to develop an interpretable machine learning (IML) model that accurately predicts 28-day all-cause mortality in hypertensive ischemic or hemorrhagic stroke patients. A total of 4,274 hypertensive ischemic or hemorrhagic stroke patients admitted to the ICU in the USA from multicenter cohorts were included in this study to develop and validate the IML model. Five machine learning (ML) models were developed, including artificial neural network (ANN), gradient boosting machine (GBM), eXtreme Gradient Boosting (XGBoost), logistic regression (LR), and support vector machine (SVM), to predict mortality using the MIMIC-IV and eICU-CRD database in the USA. Feature selection was performed using the Least Absolute Shrinkage and Selection Operator (LASSO) algorithm. Model performance was evaluated based on the area under the curve (AUC), accuracy, positive predictive value (PPV), and negative predictive value (NPV). The ML model with the best predictive performance was selected for interpretability analysis. Finally, the SHapley Additive exPlanations (SHAP) method was employed to evaluate the risk of all-cause in-hospital mortality among hypertensive ischemic or hemorrhagic stroke patients admitted to the ICU. The XGBoost model demonstrated the best predictive performance, with the AUC values of 0.822, 0.739, and 0.700 in the training, test, and external cohorts, respectively. The analysis of feature importance revealed that age, ethnicity, white blood cell (WBC), hyperlipidemia, mean corpuscular volume (MCV), glucose, pulse oximeter oxygen saturation (SpO2), serum calcium, red blood cell distribution width (RDW), blood urea nitrogen (BUN), and bicarbonate were the 11 most important features. The SHAP plots were employed to interpret the XGBoost model. The XGBoost model accurately predicted 28-day all-cause in-hospital mortality among hypertensive ischemic or hemorrhagic stroke patients admitted to the ICU. The SHAP method can provide explicit explanations of personalized risk prediction, which can aid physicians in understanding the model.

Pharmacovariome scanning using whole pharmacogene resequencing coupled with deep computational analysis and machine learning for clinical pharmacogenomics

BackgroundThis pilot study aims to identify and functionally assess pharmacovariants in whole exome sequencing data. While detection of known variants has benefited from pharmacogenomic-dedicated bioinformatics tools before, in this paper we have tested novel deep computational analysis in addition to artificial intelligence as possible approaches for functional analysis of unknown markers within less studied drug-related genes.MethodsPharmacovariants from 1800 drug-related genes from 100 WES data files underwent (a) deep computational analysis by eight bioinformatic algorithms (overall containing 23 tools) and (b) random forest (RF) classifier as the machine learning (ML) approach separately. ML model efficiency was calculated by internal and external cross-validation during recursive feature elimination. Protein modelling was also performed for predicted highly damaging variants with lower frequencies. Genotype–phenotype correlations were implemented for top selected variants in terms of highest possibility of being damaging.ResultsFive deleterious pharmacovariants in the RYR1, POLG, ANXA11, CCNH, and CDH23 genes identified in step (a) and subsequent analysis displayed high impact on drug-related phenotypes. Also, the utilization of recursive feature elimination achieved a subset of 175 malfunction pharmacovariants in 135 drug-related genes that were used by the RF model with fivefold internal cross-validation, resulting in an area under the curve of 0.9736842 with an average accuracy of 0.9818 (95% CI: 0.89, 0.99) on predicting whether a carrying individuals will develop adverse drug reactions or not. However, the external cross-validation of the same model indicated a possible false positive result when dealing with a low number of observations, as only 60 important variants in 49 genes were displayed, giving an AUC of 0.5384848 with an average accuracy of 0.9512 (95% CI: 0.83, 0.99).ConclusionWhile there are some technologies for functionally assess not-interpreted pharmacovariants, there is still an essential need for the development of tools, methods, and algorithms which are able to provide a functional prediction for every single pharmacovariant in both large-scale datasets and small cohorts. Our approaches may bring new insights for choosing the right computational assessment algorithms out of high throughput DNA sequencing data from small cohorts to be used for personalized drug therapy implementation.

Development of near-infrared reflectance spectroscopy (NIRS) model and genome-wide association study for glycogen and protein content in Pacific abalone

Quality traits have recently become the key objective in aquaculture selection breeding recently. The Pacific abalone (Haliotis discus hannai) is the most widely farmed abalone species in China. In this study, we investigated the potential of the near-infrared reflectance spectroscopy (NIRS) model to predict the glycogen and protein content of Pacific abalone. Additionally, we performed a genome-wide association study (GWAS) analysis to identify related SNPs and candidate genes for these two quality traits. Our results showed that the prediction accuracy of the NIRS model was 0.9929 and 0.9808 for glycogen and protein content, respectively. Moreover, the external cross-validation showed an accuracy above 0.95, indicating the reliability and robustness of the models. A total of 274 individuals were predicted by the NIRS model and genotyped with 69,530 high quality SNPs. Fourteen and fifteen genes were identified for glycogen and protein content, respectively, including slit2, fgfr2, gria4, hnrnpm, and megf11. These genes may affect glycogen and protein traits through biological processes such as energy metabolism, transmembrane transport, immune regulation, and signal transduction. This study not only constructed a rapid and reliable NIRS prediction model, but also helped us better understand the genetics of glycogen and protein content of Pacific abalone, which will improve the quality of Pacific abalone.