Random Forest Score Research Articles

Regression Forest Approaches to Gravity Wave Parameterization for Climate Projection

AbstractWe train random and boosted forests, two machine learning architectures based on regression trees, to emulate a physics‐based parameterization of atmospheric gravity wave momentum transport. We compare the forests to a neural network benchmark, evaluating both offline errors and online performance when coupled to an atmospheric model under the present day climate and in 800 and 1,200 ppm CO2 global warming scenarios. Offline, the boosted forest exhibits similar skill to the neural network, while the random forest scores significantly lower. Both forest models couple stably to the atmospheric model, and control climate integrations with the boosted forest exhibit lower biases than those with the neural network. Integrations with all three data‐driven emulators successfully capture the Quasi‐Biennial Oscillation (QBO) and sudden stratospheric warmings, key modes of stratospheric variability, with the boosted forest more accurate than the random forest in replicating their statistics across our range of carbon dioxide perturbations. The boosted forest and neural network capture the sign of the QBO period response to increased CO2, though both struggle with the magnitude of this response under the more extreme 1,200 ppm scenario. To investigate the connection between performance in the control climate and the ability to generalize, we use techniques from interpretable machine learning to understand how the data‐driven methods use physical information. We leverage this understanding to develop a retraining procedure that improves the coupled performance of the boosted forest in the control climate and under the 800 ppm CO2 scenario.

Radiomic Features at Contrast-Enhanced CT Predict Virus-Driven Liver Fibrosis: A Multi-Institutional Study.

Liver fibrosis is a major cause of morbidity and mortality among in patients with chronic hepatitis. Radiomics, particularly of the spleen, may improve diagnostic accuracy and treatment strategies. External validations are necessary to ensure reliability and generalizability. In this retrospective study, we developed 3 radiomics models using contrast-enhanced computed tomography scans from 167 patients with liver fibrosis (training group) between January 2020 and December 2021. Radiomic features were extracted from arterial venous, portal venous, and equilibrium phase images. Recursive feature selection random forest and the least absolute shrinkage and selection operator logistic regression were used for feature selection and dimensionality reduction. Performance was assessed by area under the curve, C-index, calibration plots, and decision curve analysis. External validation was performed on 114 patients from 2 institutions. Twenty-five radiomic features were significantly associated with fibrosis stage, with 80% of the top 10 features originating from portal venous phase spleen images. The radiomics models showed good performance in the validation cohort (C-indices 0.723-0.808) and excellent calibration. Decision curve analysis indicated clinical benefits, with machine learning-based radiomics models (Random Forest score and support vector machine based radiomics score) providing more significant advantages. Radiomic features offer significant benefits over existing serum indices for staging virus-driven liver fibrosis, underscoring the value of radiomics in enhancing diagnostic accuracy. Specifically, radiomics analysis of the spleen presents additional noninvasive options for assessing fibrosis, highlighting its potential in improving patient management and outcomes.

Effect of galectin-1 on prognosis and responsiveness of immune checkpoint plus tyrosine kinase inhibition in renal cell carcinoma.

In renal cell carcinoma (RCC), no clinically available biomarker has been utilized for checkpoint inhibitor immunotherapy (IO) + tyrosine kinase inhibitor (TKI) combinations. Galectin-1 overexpression is found in tumors, with potential immune-regulating roles. RNA-sequencing was performed in two cohorts of RCC treated with IO/TKI combination therapy (ZS-MRCC, JAVELIN-101). Immunohistochemistry and flow cytometry were performed to investigate immune cell infiltration and function in the tumor microenvironment of RCC. The RECIST criteria were used to define response and progression-free survival (PFS). Galectin-1 expression was elevated in RCC with higher stage (p < 0.001) and grade (p < 0.001). Galectin-1 expression was also elevated in non-responders of IO/TKI therapy (p = 0.047). High galectin-1 was related with shorter PFS in both ZS-MRCC cohort (p = 0.036) and JAVELIN-101 cohort (p = 0.005). Multivariate Cox analysis defined galectin-1 as an independent factor for PFS (HR 2.505; 95% CI 1.116-5.622; p = 0.026). In the tumor microenvironment, high galectin-1 was related with decreased GZMB+CD8+ T cells (Speraman's ρ = -0.31, p = 0.05), and increased PD1 + CD8+ T cells (Speraman's ρ = 0.40, p = 0.01). Besides, elevated number of regulatory T cells (p = 0.039) and fibroblasts (p = 0.011) was also found in high galectin-1 tumors. Finally, a random-forest score (RFscore) was built for predicting IO/TKI benefit. IO/TKI therapy showed benefit only in low-RFscore patients (HR 0.489, 95% CI 0.358-0.669, p < 0.001), rather than high-RFscore patients (HR 0.875, 95% CI 0.658-1.163, p = 0.357). High galectin-1 indicated therapeutic resistance and shorter PFS of IO/TKI therapy. High galectin-1 also indicated CD8+ T cell dysfunction. High galectin-1 could be applied for patient selection of IO/TKI therapy in RCC.

Effect of Annexin A2 on prognosis and sensitivity to immune checkpoint plus tyrosine kinase inhibition in metastatic renal cell carcinoma

BackgroundImmunotherapy (IO) plus tyrosine kinase inhibitor (TKI) therapy is the first-line recommendation for advanced renal cell carcinoma (RCC), but no biomarker has been approved for it. Annexin A2 (ANXA2) can induce immune escape in tumors.MethodsTwo independent cohorts of advanced RCC treated by IO + TKI were utilized for survival analysis (ZS-MRCC, n = 45; Javelin-101, n = 726). ANXA2 expression was determined by RNA-sequencing. The impact of ANXA2 on the tumor microenvironment was assessed by RNA-sequencing, flow cytometry and immunohistochemistry in two localized RCC datasets (ZS-HRRCC, n = 40; TCGA-KIRC, n = 530).ResultsANXA2 was upregulated in non-responders of IO + TKI therapy (p = 0.027). High-ANXA2 group showed poor progression-free survival (PFS) in both the ZS-MRCC cohort (HR, 2.348; 95% CI 1.084–5.085; P = 0.025) and the Javelin-101 cohort (HR, 1.472; 95% CI 1.043–2.077; P = 0.027). Multivariate Cox regression determined ANXA2 as an independent prognostic factor (HR, 2.619; 95% CI 1.194–5.746; P = 0.016). High-ANXA2 was correlated with decreased proportion of granzyme B+ CD8+ T cells (Spearman’s ρ = − 0.40, P = 0.01), and increased TIM-3+ (Spearman’s ρ = 0.43, P < 0.001) and CTLA4+ (Spearman’s ρ = 0.49, P < 0.001) tumor-infiltrating lymphocytes. A random forest (RF) score was further build by integrating ANXA2 and immune genes, which stratified patients who would benefit from IO + TKI therapy (low-RF score, IO + TKI vs TKI, HR = 0.453, 95% CI 0.328–0.626; high-RF score, IO + TKI vs TKI, HR = 0.877, 95% CI 0.661–1.165; interaction P = 0.003).ConclusionsUpregulated ANXA2 was associated with poor PFS and therapeutic resistance in RCC treated by IO + TKI therapy, and related with T cell exhaustion. The integrated RF score could stratify patients who would benefit from IO + TKI therapy.

CT-based radiomics signature for differentiating pyelocaliceal upper urinary tract urothelial carcinoma from infiltrative renal cell carcinoma.

To develop a CT-based radiomics model and a combined model for preoperatively discriminating infiltrative renal cell carcinoma (RCC) and pyelocaliceal upper urinary tract urothelial carcinoma (UTUC), which invades the renal parenchyma. Eighty patients (37 pathologically proven infiltrative RCCs and 43 pathologically proven pyelocaliceal UTUCs) were retrospectively enrolled and randomly divided into a training set (n = 56) and a testing set (n = 24) at a ratio of 7:3. Traditional CT imaging characteristics in the portal venous phase were collected by two radiologists (SPH and ZXL, who have 4 and 30 years of experience in abdominal radiology, respectively). Patient demographics and traditional CT imaging characteristics were used to construct the clinical model. The radiomics score was calculated based on the radiomics features extracted from the portal venous CT images and the random forest (RF) algorithm to construct the radiomics model. The combined model was constructed using the radiomics score and significant clinical factors according to the multivariate logistic regression. The diagnostic efficacy of the models was evaluated using receiver operating characteristic (ROC) curve analysis and the area under the curve (AUC). The RF score based on the eight validated features extracted from the portal venous CT images was used to build the radiomics model. Painless hematuria as an independent risk factor was used to build the clinical model. The combined model was constructed using the RF score and the selected clinical factor. Both the radiomics model and combined model showed higher efficacy in differentiating infiltrative RCC and pyelocaliceal UTUC in the training and testing cohorts with AUC values of 0.95 and 0.90, respectively, for the radiomics model and 0.99 and 0.90, respectively, for the combined model. The decision curves of the combined model as well as the radiomics model indicated an overall net benefit over the clinical model. Both the radiomics model and the combined model achieved a notable reduction in false-positive and false-negativerates, resulting in significantly higher accuracy compared to the visual assessments in both the training and testing cohorts. The radiomics model and combined model had the potential to accurately differentiate infiltrative RCC and pyelocaliceal UTUC, which invades the renal parenchyma, and provide a new potentially non-invasive method to guide surgery strategies.

Abstract 18629: Discriminative Accuracy of Cha 2 Ds 2 Vasc Score, and Development of Predictive Accuracy Model Using Machine Learning for Ischemic Stroke in Cardiac Amyloidosis

Background: Cardiac amyloidosis (CA) in conjunction with atrial fibrillation (AF) presents unique management challenges. CHA2DS2VASc score in these patients is believed to underestimate the risk of ischemic stroke, necessitating a better predictive model in these patients. Methods: Data was obtained from the National Readmission Database (NRD). Outcomes between CA-AF and no-CA-AF were compared using multivariate regression analysis to calculate adjusted odds ratios (aOR). AutoScore; an interpretable machine learning framework, was used to develop a stroke risk prediction model, the predictive accuracy of which was evaluated with an area under the curve (AUC) using the receiver operating characteristic analysis. Results: A total of 11,860,804 (CA-AF 22,687 [0.19%] and no-CA-AF 11,838,117) patients were identified from 2015-2019. The adjusted odds of mortality (aOR 1.41 and 1.29), stroke (aOR 1.78 and 1.74), non-intracranial hemorrhage (aOR 2.10 and aOR 1.85), and intracranial hemorrhage (aOR 14.4 and aOR 4.26) were significantly higher in CA-AF compared with non-CA-AF at both index admission and 30-days, respectively. The CHA 2 DS 2 VASc score had a poor discriminative accuracy for stroke at 30-days in CA-AF (AUC 49%, 95%CI 47%-51%, p=0.54). The machine learning autoscore integrative model revealed that the predictive ability of our newly proposed E-CHADS score (end-stage renal disease (ESRD), congestive heart failure, hypertension, active cancer, dementia, and diabetes mellitus) for 30-day risk of ischemic stroke in CA-AF was excellent (for a cutoff of 52 points random forest score) with an AUC of 80% (95%CI 74%-86%) Conclusion: Cardiac amyloidosis carries a high risk of ischemic stroke that is not accurately predicted by the CHA 2 DS 2 VASc score. Our proposed model (E-CHADS) identifies 3 new variables (ESRD, dementia, and cancer) that have higher discriminative accuracy for ischemic stroke in these patients.

Implementation of computational strategies to combat drug resistance in Mycobacterium tuberculosis

Mycobacterium tuberculosis (MTB) is an infectious pathogen that causes tuberculosis (TB). Even though it is curable, the first-line anti-TB medication rifampicin has been a source of rising concern for human health across the globe. However, rifampicin resistance results from mutations in the RNA polymerase's binding site of the target protein. This infectious pathogen can survive in macrophages and can induce a long-lasting latent infection. The prolonged survival of mycobacterial species was achieved with the aid of serine/threonine protein kinase G (PknG) which performs a preventive role in this situation. Thus, PknG is believed to be an essential target to prevent the pathogen from entering the latency stage. The present investigation was oriented towards molecular-based virtual screening carried out using AutoDock for the FDA-approved and nutraceutical subsets of compounds from the DrugBank repository. Initially, a total of 3035 compounds were screened based on their absorption, distribution, metabolism, excretion and toxicity (ADMET) properties. The screened molecules were taken for docking purposes after which post-docking analysis was performed by calculating the random forest (RF) score. Further, the structural characterization of hit compounds was validated using interaction studies and in silico bioactivity prediction studies. Finally, the anti-myotubular activity prediction was carried out in the search for potential inhibitors. Importantly, four potential ligands, namely DB00080, DB00807, DB06249 and DB06274 were preferred after a thorough investigation. These ligands showed significant structural and protein-ligand interactions, suggesting that they might be potentially used as therapeutic candidates to combat MTB.

The prognostic significance of CDK6 expression in renal cell carcinoma treated by immune checkpoint plus tyrosine kinase inhibition.

Checkpoint inhibitor immunotherapy plus tyrosine kinase inhibitor (IO/TKI) has become the first-line treatment for metastatic renal cell carcinoma (RCC), despite the lack of biomarkers. Cyclin-dependent kinase 6 (CDK6) has shown a regulatory role in antitumour response. The study enrolled two cohorts of metastatic RCC treated by IO/TKI (Zhongshan Hospital [ZS]-MRCC, n = 45; JAVELIN-101, n = 726) and two cohorts of localized RCC (ZS-HRRCC, n = 40; TCGA-KIRC, n = 530). CDK6 was evaluated by RNA-sequencing. Progression-free survival (PFS) was the primary endpoint. The prognostic role of CDK6 was evaluated by survival analysis. The correlation between CDK6 and tumour microenvironment was assessed by immunohistochemistry and flow cytometry. The high-CDK6 group displayed a lower response rate (13.6%) than the low-CDK6 group (56.5%) (P = .002). High-CDK6 was associated with poor PFS in both the ZS-MRCC cohort (high-CDK6, median PFS 6.4 months; low-CDK6, median PFS not reached; P = .010) and JAVELIN-101 cohort (high-CDK6, median PFS 10.0 months; low-CDK6, median PFS 13.3 month; P = .033). High-CDK6 was associated with increased PD1+ CD8+ T cells (Spearman's ρ = .47, P < .001) and decreased Granzyme B+ CD8+ T cells (Spearman's ρ = -.35, P = .030). Finally, a random forest score (RFscore) was built by integrating CDK6 and immunologic genes, which was associated with survival benefits of IO/TKI (RFscore-low, TKI vs IO/TKI, HR = 2.47, 95% CI 1.82-3.35, P < .001; RFscore-high, TKI vs IO/TKI, HR = 0.99, 95% CI 0.75-1.32, P = .963). Elevated CDK6 expression indicated resistance and poor PFS under IO/TKI therapy, which was related to exhausted CD8+ T cells. Integrated RFscore could evaluate the benefits of IO/TKI.

Metabolite Profiling of the Gut–Renal–Cerebral Axis Reveals a Particular Pattern in Early Diabetic Kidney Disease in T2DM Patients

Type 2 diabetes mellitus (T2DM) represents an important microvascular disease concerning the kidney and the brain. Gut dysbiosis and microbiota-derived metabolites may be in relation with early pathophysiological changes in diabetic kidney disease (DKD). The aim of the study was to find new potential gut-derived biomarkers involved in the pathogenesis of early DKD, with a focus on the complex interconnection of these biomarkers with podocyte injury, proximal tubule dysfunction, renal and cerebrovascular endothelial dysfunction. The study design consisted of metabolite profiling of serum and urine of 90 T2DM patients (subgroups P1-normoalbuminuria, P2-microalbuminuria, P3-macroalbuminuria) and 20 healthy controls (group C), based on ultra-high-performance liquid chromatography coupled with electrospray ionization-quadrupole-time of flight-mass spectrometry analysis (UHPLC-QTOF-ESI+-MS). By multivariate and univariate analyses of serum and urine, which included Partial Least Squares Discriminant Analysis (PLSDA), Variable Importance Plots (VIP), Random Forest scores, One Way ANOVA and Biomarker analysis, there were discovered metabolites belonging to nitrogen metabolic pathway and retinoic acid signaling pathway which differentiate P1 group from P2, P3, C groups. Tyrosine, phenylalanine, indoxyl sulfate, serotonin sulfate, and all-trans retinoic acid express the metabolic fingerprint of P1 group vs. P2, P3, C groups, revealing a particular pattern in early DKD in T2DM patients.

Modeling Job Satisfaction of Peruvian Basic Education Teachers Using Machine Learning Techniques

Teacher job satisfaction is an important aspect of academic performance, student retention, and teacher retention. We propose to determine the predictive model of job satisfaction of basic education teachers using machine learning techniques. The original data set consisted of 15,087 instances and 942 attributes from the national survey of teachers from public and private educational institutions of regular basic education (ENDO-2018) carried out by the Ministry of Education of Peru. We used the ANOVA F-test filter and the Chi-Square filter as feature selection techniques. In the modeling phase, the logistic regression algorithms, Gradient Boosting, Random Forest, XGBoost and Decision Trees-CART were used. Among the algorithms evaluated, XGBoost and Random Forest stand out, obtaining similar results in 4 of the 8 metrics evaluated, these are: balanced accuracy of 74%, sensitivity of 74%, F1-Score of 0.48 and negative predictive value of 0.94. However, in terms of the area under the ROC curve, XGBoost scores 0.83, while Random Forest scores 0.82. These algorithms also obtain the highest true-positive values (479 instances) and lowest false-negative values (168 instances) in the confusion matrix. Economic income, satisfaction with life, self-esteem, teaching activity, relationship with the director, perception of living conditions, family relationships; health problems related to depression and satisfaction with the relationship with colleagues turned out to be the most important predictors of job satisfaction in basic education teachers.

MetaScore: A Novel Machine-Learning-Based Approach to Improve Traditional Scoring Functions for Scoring Protein-Protein Docking Conformations.

Protein-protein interactions play a ubiquitous role in biological function. Knowledge of the three-dimensional (3D) structures of the complexes they form is essential for understanding the structural basis of those interactions and how they orchestrate key cellular processes. Computational docking has become an indispensable alternative to the expensive and time-consuming experimental approaches for determining the 3D structures of protein complexes. Despite recent progress, identifying near-native models from a large set of conformations sampled by docking-the so-called scoring problem-still has considerable room for improvement. We present MetaScore, a new machine-learning-based approach to improve the scoring of docked conformations. MetaScore utilizes a random forest (RF) classifier trained to distinguish near-native from non-native conformations using their protein-protein interfacial features. The features include physicochemical properties, energy terms, interaction-propensity-based features, geometric properties, interface topology features, evolutionary conservation, and also scores produced by traditional scoring functions (SFs). MetaScore scores docked conformations by simply averaging the score produced by the RF classifier with that produced by any traditional SF. We demonstrate that (i) MetaScore consistently outperforms each of the nine traditional SFs included in this work in terms of success rate and hit rate evaluated over conformations ranked among the top 10; (ii) an ensemble method, MetaScore-Ensemble, that combines 10 variants of MetaScore obtained by combining the RF score with each of the traditional SFs outperforms each of the MetaScore variants. We conclude that the performance of traditional SFs can be improved upon by using machine learning to judiciously leverage protein-protein interfacial features and by using ensemble methods to combine multiple scoring functions.

Stealthy attack detection method based on Multi-feature long short-term memory prediction model

Industrial control systems (ICS) face severe threats due to the inherent vulnerability of shared networks. Among the attacks against ICS, stealthy attack is an attack behavior in which an attacker injects false sensor measurements or drives signals into the control loop, evading detection by the intrusion detection system (IDS). They are highly destructive and difficult to detect because of concealment. In related studies, data-driven methods have the shortcomings of a large computation burden. Physics-based methods are usually difficult and expensive to change the system structure. Moreover, most of the current methods against stealthy attacks take up a considerable amount of system resources due to training or the pursuit of robustness. In this paper, using the importance and correlation of neglected feature data, an stealthy attack detection method based on the Multi-feature LSTM (MFLSTM) model is innovatively developed to predict and recover data attacked. Random Forest (RF) scores and heatmap are used to judge the importance and correlation of the feature data. Then the forget gates of the LSTM model are improved to construct an MFLSTM model that can learn predictive information from other feature data and decouples the attack conditions that stealthy attack relies on. The testing results indicated that MFLSTM has significant advantages in prediction accuracy, stability, and resource-saving. MFLSTM model saved 52.3% of the resources required for the same type of prediction. The single-point prediction mean square error (MSE) for the STEP 7 (S7) protocol attack signature prediction and secure water treatment (SWaT) dataset were 0.0471 and 0.0035, respectively, which also demonstrates the feasibility of our proposed method.

Machine Learning Analysis of Handgun Transactions to Predict Firearm Suicide Risk

Evidence suggests that limiting access to firearms among individuals at high risk of suicide can be an effective means of suicide prevention, yet accurately identifying those at risk to intervene remains a key challenge. Firearm purchasing records may offer a large-scale and objective data source for the development of tools to predict firearm suicide risk. To test whether a statewide database of handgun transaction records, coupled with machine learning techniques, can be used to forecast firearm suicide risk. This prognostic study used the California database of 4 976 391 handgun transaction records from 1 951 006 individuals from January 1, 1996, to October 6, 2015. Transaction-level random forest classification was implemented to predict firearm suicide risk, and the relative predictive power of features in the algorithm was estimated via permutation importance. Analyses were performed from December 1, 2020, to May 19, 2022. The main outcome was firearm suicide within 1 year of a firearm transaction, derived from California death records (1996-2016). With the use of California's Dealer's Records of Sale (1996-2015), 41 handgun, transaction, purchaser, and community-level predictor variables were generated. There are a total of 4 976 391 transactions in the California's Dealer's Record of Sale database representing 1 951 006 individuals (1 525 754 men [78.2% of individuals]; mean [SD] age, 43.4 [13.9] years). Firearm suicide within 1 year occurred in 0.07% of handgun transactions (3278 transactions among 2614 individuals). A total of 38.6% of observed firearm suicides were among transactions classified in the highest-risk ventile (379 of 983 transactions), with 95% specificity. Among the small number of transactions with a random forest score above 0.95, more than two-thirds (24 of 35 [68.6%]) were associated with a purchaser who died by firearm suicide within 1 year. Important features included known risk factors, such as older age at first purchase, and previously unreported predictors, including distance to firearms dealer and month of purchase. This prognostic study presented the first large-scale machine learning analysis of individual-level handgun transaction records. The results suggested the potential utility of such records in identifying high-risk individuals to aid suicide prevention efforts. It also identified handgun, individual, and community characteristics that have strong predictive relationships with firearm suicide and may warrant further study.

Diagnostic Performance of a Noninvasive Breath Test for Colorectal Cancer: COBRA1 Study

Diagnostic Performance of a Noninvasive Breath Test for Colorectal Cancer: COBRA1 Study

Combined prognostic value of preoperative serum thyrotrophin and thyroid hormone concentration in papillary thyroid cancer.

BackgroundA growing number of studies have found a close association between thyroid hormones and thyrotrophin (TSH), and they also have prognostic significance in some cancer types; this study aimed to investigate the prognostic value of free triiodothyronine (fT3), free thyroxine (fT4), fT3/fT4, TSH, and their combination in patients with papillary thyroid carcinoma (PTC).MethodsThis study retrospectively analyzed the relevant data of 726 newly diagnosed PTC patients. Both univariate and multivariate analyses were used to predict the recurrence rate, and a risk score was established. In addition, with the use of a random survival forest, a random forest (RF) score was constructed. After calculating the area under the curve (AUC), the diagnostic efficacy of risk score, RF score, and four indicators was compared.ResultsfT3, fT4, fT3/fT4, and TSH were strongly associated with some invasive clinicopathological features and postoperative recurrence. Patients with high expression of fT4 and TSH have a high risk of recurrence. By contrast, patients with high expression of fT3 and fT3/fT4 have a low risk of recurrence. At the same time, the combined use of various indicators is more helpful for establishing an accurate diagnosis. By comparison, we found that the RF score was better than the risk score in terms of predicting the recurrence of PTC.ConclusionThe diagnostic accuracy of a combination of fT3, fT4, fT3/fT4, and TSH can help improve our clinical estimate of the risk of recurrent PTC, thus allowing the development of a more effective treatment plan for patients.

Applying Machine Learning Approach to Start-up Success Prediction

Predicting the success of a new venture has always been a topical issue for both investors and researchers. Nowadays, it has become even more relevant concerning start-ups-young innovative and technology enterprises aimed at scaling their businesses. The purpose of this study is to create a model for predicting start-ups’ success based on their descriptive characteristics. A model that connects such start-up features as the period from foundation to the first financing, the area of activity, type, and amount of the first financing round, business model, and applied technologies, with the start-up investment success, which refers to re-investment, has been developed using data from the Dealroom platform on statistics of start-ups activity and their description. The final sample included 123 start-ups that are founded or operate in Ukraine. Three machine learning algorithms are compared: Logistic Regression, Decision Tree, and Random Forest. Acceptable results were obtained in terms of Accuracy, Sensitivity, and F-score, despite the limited data. The best model concerning start-up success prediction is determined by a Decision Tree, with an average effectiveness of 61%, 55%, and 52%, respectively. The AUC level for the Decision Tree achieved 58%, which is lower than the Logistic Regression and Random Forest scores (65%). But the last models had done so well by better predicting start-up failures, while more practical is the ability to predict their success. All models showed an acceptable level of AUC to confirm with confidence their effectiveness. The decision support system for the investment object can be helpful for entrepreneurs, venture analysts, or politicians who can use the built models to predict the success of a start-up. This forecast, in turn, can be used to drive better investment decisions and develop relevant economic policies to improve the overall start-up ecosystem

Prediction of lymph node metastasis in early colorectal cancer based on histologic images by artificial intelligence

Risk evaluation of lymph node metastasis (LNM) for endoscopically resected submucosal invasive (T1) colorectal cancers (CRC) is critical for determining therapeutic strategies, but interobserver variability for histologic evaluation remains a major problem. To address this issue, we developed a machine-learning model for predicting LNM of T1 CRC without histologic assessment. A total of 783 consecutive T1 CRC cases were randomly split into 548 training and 235 validation cases. First, we trained convolutional neural networks (CNN) to extract cancer tile images from whole-slide images, then re-labeled these cancer tiles with LNM status for re-training. Statistical parameters of the tile images based on the probability of primary endpoints were assembled to predict LNM in cases with a random forest algorithm, and defined its predictive value as random forest score. We evaluated the performance of case-based prediction models for both training and validation datasets with area under the receiver operating characteristic curves (AUC). The accuracy for classifying cancer tiles was 0.980. Among cancer tiles, the accuracy for classifying tiles that were LNM-positive or LNM-negative was 0.740. The AUCs of the prediction models in the training and validation sets were 0.971 and 0.760, respectively. CNN judged the LNM probability by considering histologic tumor grade.

Crack damage monitoring for compressor blades based on acoustic emission with novel feature and hybridized feature selection

Nowadays, acoustic emission (AE) has been widely used to the nondestructive evaluation (NDE) and structural health monitoring (SHM) for compressor blades. However, traditional AE features and feature selection methods are generally difficult to identify the cracked blades of compressor due to its complex structure and background noise. To solve this problem, the crack damage monitoring method based on novel feature and hybridized feature selection is proposed to identify crack of compressor blades, which is aimed at improving the crack identification accuracy with optimal features. First, the novel feature of spectral centroid with energy shift (SCES) is established. Besides, the hybridized feature selection method is proposed based on Laplacian random forest scores (LRFS), which can evaluate and select features adaptively. By fusing information of selected features from AE sensors, the long short-term memory (LSTM) network is used to classify cracked blades. The proposed method is applied experimentally to identify cracks at different speeds and locations of AE sensors, which has the average accuracy of 98.93%. The comparative results demonstrate the effectiveness and superiority of the proposed method in AE-based SHM for compressor blades under different working conditions.

Inflammation Markers Have Important Value in Predicting Relapse in Patients with papillary thyroid carcinoma: A Long-Term Follow-Up Retrospective Study.

PurposeMany markers of inflammation are increasingly found to have prognostic significance in some cancers. This study investigated the prognostic value of albumin/globulin (AGR), lymphocyte/monocyte ratio (LMR), and other inflammatory markers, including neutrophil/lymphocyte ratio (NLR) and platelet/lymphocyte ratio (PLR), in patients with papillary thyroid carcinoma (PTC).MethodsWe retrospectively analyzed the data of 764 patients newly diagnosed with PTC (608 women, 156 men) aged 10-83 years. Univariate and multivariate analyses were used to analyze recurrence rates and assess potential prognostic factors. Furthermore, we used random survival forests to construct a random survival forest score (RSFscore). The correlations between various inflammatory factors and traditional prognostic factors were analyzed. We also compared the areas under the curve (AUCs) of the RSFscore and 4 inflammation-based markers.ResultsAGR, NLR, PLR, and LMR were strongly associated with invasive clinicopathological features (tumor size, lesions, lymph node metastasis, and lymph node metastasis rate) and postoperative recurrence. In the multivariate analysis, AGR and LMR were independent prognostic markers for recurrent PTC. Higher NLR and PLR values indicated a higher risk of recurrence, while higher LMR and AGR values suggested a lower recurrence risk. The predictive power of the combined indicators was stronger than that of single indicators alone.ConclusionCompared to the analysis of a single indicator, the combination of inflammatory markers was more helpful in determining the risk of PTC recurrence, which has an important impact on predicting patients’ cancer-free survival and quality of life.

Identification of NSP3 (SH2D3C) as a Prognostic Biomarker of Tumor Progression and Immune Evasion for Lung Cancer and Evaluation of Organosulfur Compounds from Allium sativum L. as Therapeutic Candidates.

The multi-domain non-structural protein 3 (NSP3) is an oncogenic molecule that has been concomitantly implicated in the progression of coronavirus infection. However, its oncological role in lung cancer and whether it plays a role in modulating the tumor immune microenvironment is not properly understood. In the present in silico study, we demonstrated that NSP3 (SH2D3C) is associated with advanced stage and poor prognoses of lung cancer cohorts. Genetic alterations of NSP3 (SH2D3C) co-occurred inversely with Epidermal Growth Factor Receptor (EGFR) alterations and elicited its pathological role via modulation of various components of the immune and inflammatory pathways in lung cancer. Our correlation analysis suggested that NSP3 (SH2D3C) promotes tumor immune evasion via dysfunctional T-cell phenotypes and T-cell exclusion mechanisms in lung cancer patients. NSP3 (SH2D3C) demonstrated a high predictive value and association with therapy resistance in lung cancer, hence serving as an attractive target for therapy exploration. We evaluated the in silico drug-likeness and NSP3 (SH2D3C) target efficacy of six organosulfur small molecules from Allium sativum using a molecular docking study. We found that the six organosulfur compounds demonstrated selective cytotoxic potential against cancer cell lines and good predictions for ADMET properties, drug-likeness, and safety profile. E-ajoene, alliin, diallyl sulfide, 2-vinyl-4H-1,3-dithiin, allicin, and S-allyl-cysteine docked well into the NSP3 (SH2D3C)-binding cavity with binding affinities ranging from –4.3~–6.70 Ă and random forest (RF) scores ranging from 4.31~5.26 pKd. However, S-allyl-cysteine interaction with NSP3 (SH2D3C) is unfavorable and hence less susceptible to NSP3 ligandability. In conclusion, our study revealed that NSP3 is an important onco-immunological biomarker encompassing the tumor microenvironment, disease staging and prognosis in lung cancer and could serve as an attractive target for cancer therapy. The organosulfur compounds from A. sativum have molecular properties to efficiently interact with the binding site of NSP3 and are currently under vigorous preclinical study in our laboratory.