Predicted Survival Probabilities Research Articles

Tensor Landmark Analysis With Application to ADNI data

ABSTRACTRecent advancements in data collection have facilitated the use of multidimensional arrays, also known as tensors, in prediction of health outcomes. In this article, we introduce a tensor landmark model for predicting survival outcomes using multiple longitudinal biomarkers as tensor covariates through CANDECOMP/PARAFAC (CP) decomposition. An iteratively reweighted least squares estimation is adopted for components of the tensor coefficients in the landmark model. We also present empirical results of the Akaike information criterion (AIC) and the Bayesian information criterion (BIC) for right‐censored data to select the CP rank. Simulations and Alzheimer's disease neuroimaging initiative (ADNI) data analysis demonstrate that our proposed model accurately estimates survival coefficients and predicts survival probabilities. The implementation code can be found online (https://github.com/sparkqkr/TensorCoxReg).

PRIORiTize-TAVI might predict more than transcatheter aortic valve replacement waiting list priorities

Abstract Background PRIORiTize-TAVI (Predicting moRtalIty Or uRgent TAVI on waiting list) is a novel scoring system targeting identification of patients at risk of urgent hospitalization or death whilst waiting for transcatheter aortic valve replacement (TAVR) (1). Patients are divided into those with low (<2 points), intermediate (3-4 points) and high (>5) waiting risk. Given the novelty of the score, additional research is required to further validate it. However, being compiled of parameters applied in everyday practice that are readily available, the application of the score might have the potential to go beyond its original purpose. Purpose We aimed to investigate whether PRIORiTize-TAVI score can predict postprocedural complications and survival following TAVR. Methods We conducted a registry-based study including patients who underwent TAVR at our institution from 2018 to 2023. PRIORiTize-TAVI was calculated summarizing points according to NYHA status, left ventricular ejection fraction (LVEF), NTproBNP, mean pressure gradient (meanPG) and Charlson comorbidity index (CCI) at the time of patient placement onto TAVR waiting list. Results This registry-based study included 252 patients with median age 80 (interquartile range (IQR) 76-84) years, 50% female. Median PRIORiTize score was 5 (IQR 3-7). There were 41 (16%) low, 51 (20%) intermediate and 160 (64%) high risk patients. High risk patients had more frequently atrial fibrillation and ischaemic stroke, had lower LVEF and estimated glomerular filtration rate (eGFR), higher NTproBNP levels and CCI (Table 1). MeanPG also statistically differed between intermediate risk group and low risk group, but not with high risk group (Table 1). TAVR waiting time was significantly shorter in high-risk patients in comparison to intermediate risk patients. There was no statistical difference regarding baseline characteristics between the three groups (Table 1). There was no statistical difference in post-TAVR complications between the three groups. Survival significantly differed depending on PRIORiTize risk stratification (P=.023), as shown in Figure 1. Median survival time for high risk patients was 49 months (IQR 42-49), whereas the median survival time for intermediate and low risk patients could not be reached. High risk patients had lower survival probability compared to both intermediate (HR 4.089; CI 1.91-8.73) and low risk (HR 2.078; CI 0.77-5.59) patients. Intermediate risk patients had a higher survival probability compared to low risk patients (HR 0.508; CI 1.67-1.54). Length of TAVR waiting time had no significant association with overall survival (P=.891). Conclusions Our data show that PRIORiTize-TAVI has a potential to be used as a prognostic tool to predict survival probability post-TAVR. Our results suggest that prolonged time on TAVR waiting list did not negatively impact patient postprocedural complications nor survival.Table 1Figure 1

Integrated immunogenomic analyses of single-cell and bulk profiling construct a T cell-related signature for predicting prognosis and treatment response in osteosarcoma

PurposesT cells play a crucial role as regulators of anti-tumor activity within the tumor microenvironment (TME) and are closely associated with the progression of osteosarcoma (OS). Nevertheless, the specific role of T cell-related genes (TCRGs) in the pathogenesis of OS remains unclear.MethodsFirst, we processed single-cell RNA sequencing (scRNA-seq) data of OS from the public databases and performed cell annotation. We identified highly variable genes in each cell type using the "FindAllMarkers" function, explored the distribution of different clusters, and investigated inter-cellular communication patterns via the "CellChat" framework. Then, we used multivariate Cox analysis to construct a TCRG and developed a nomogram to predict survival probabilities for OS patients. Finally, we validated the aforementioned results using various cell lines and investigated the immune cell infiltration, expression of immune checkpoints, chemotherapy sensitivity, and the efficacy of targeted therapies across different risk groups.ResultsFrom the scRNA-seq data, we identified 3,000 highly variable genes, presented the top 10 genes, and validated the expression of core genes across different cell lines.Moreover, our analysis delved into interactions between T cells and other cell types. Our analyses constructed a predictive T cell-related signature (TCRS) that incorporated these prognostic TCRGs, showing a clear prognostic separation between the high-risk and low-risk OS patient groups in multiple cohorts. Survival analysis indicated better outcomes for patients classified in the high-risk group. The low-risk group exhibited elevated levels of CD4 memory resting T cells, contrasting with the higher levels of macrophage M0 observed in the high-risk group via the CIBERSORT algorithm. Furthermore, we observed that the low-risk group exhibitedAQ1 significant up-regulation of immune checkpoint genes (ICGs) and lower Tumour Immune Dysfunction and Exclusion (TIDE) scores, suggesting that they may be suitable for immunotherapy. Conversely, the high-risk group appeared more responsive to chemotherapy and targeted therapies, according to our drug sensitivity analysis.ConclusionIn conclusion, our study identified TCRGs, constructed and validated a TCRS for OS, and assessed immune response and drug sensitivity in different risk groups of OS patients. These findings provide novel insights into personalized treatment strategies for OS, potentially guiding more effective therapeutic interventions.

Developing a prognostic model for hepatocellular carcinoma based on MED19 and clinical stage and determining MED19 as a therapeutic target

BackgroudMediator complex subunit 19 (MED19), a member of the mediator complex, has been demonstrated to involve in tumorigenesis of hepatocellular carcinoma (HCC). However, the regulation mechanisms of MED19, the immune landscape linking MED19 to HCC and its predictive value of immunotherapy treatment in HCC are so far unknown.MethodsHere, we analyzed data from The Cancer Genome Atlas and other databases to assess the expression of MED19 and its prognosis and therapeutical-targets impact in HCC.ResultsMED19 expression was upregulated in HCC tissues compared to non-tumorous liver tissues and that its upregulation was positively associated with advanced clinicopathology features. The multivariate analysis showed that MED19 was an independent predictor of outcome in HCC. In vitro experiments revealed that MED19 knockdown suppressed hepG2 cells proliferation, colony forming and invasion and induced apoptosis. Furthermore, MED19 inhibition resulted in G0/G1 phase arrest in hepG2 cells. We screened differentially expressed genes between low and high MED19 expression groups. Enrichment analyses showed that these genes were mainly linked to nuclear division and cell cycle. The pattern of tumor-infiltrating immune was demonstrated to be related with MED19 expression in HCC. TIDE analyses showed that patients in the low-expression group presented significantly better immunotherapy. Moreover, we developed a predicted model for HCC patient’s prognosis. Receiver operating characteristic analyses revealed that this model processed a favorable performance in predicting the prognosis of HCC patients. Finally, a nomogram was built for predicting survival probability of individual HCC patient.ConclusionThese findings suggest that MED19 as a novel biomarker that has significant association with immune landscape and immunotherapy response in HCC. The proposed prediction model composed of MED19 and pathological stage has a better role in determining prognosis and stratifying of HCC.

Integrated analysis of N6-methyladenosine- and 5-methylcytosine-related long non-coding RNAs for predicting prognosis in cervical cancer

BackgroundN6-methyladenosine (m6A) and 5-methylcytosine (m5C) play a role in modifying long non-coding RNAs (lncRNAs) implicated in tumorigenesis and progression. This study was performed to evaluate prognostic value of m6A- and m5C-related lncRNAs and develop an efficient model for prognosis prediction in cervical cancer (CC).MethodsUsing gene expression data of TCGA set, we identified m6A- and m5C-related lncRNAs. Consensus Clustering Analysis was performed for samples subtyping based on survival-related lncRNAs, followed by analyzing tumor infiltrating immune cells (TIICs). Optimal signature lncRNAs were obtained using lasso Cox regression analysis for constructing a prognostic model and a nomogram to predict prognosis.ResultsWe built a co-expression network of 23 m6A-related genes, 15 m5C-related genes, and 62 lncRNAs. Based on 9 m6A- and m5C-related lncRNAs significantly associated with overall survival (OS) time, two molecular subtypes were obtained, which had significantly different OS time and fractions of TIICs. A prognostic model based on six m6A- and m5C-related signature lncRNAs was constructed, which could dichotomize patients into two risk subgroups with significantly different OS time. Prognostic power of the model was successfully validated in an independent dataset. We subsequently constructed a nomogram which could accurately predict survival probabilities. Drug sensitivity analysis found preferred chemotherapeutic agents for high and low-risk patients, respectively.ConclusionOur study reveals that m6A- and m5C-related lncRNAs are associated with prognosis and immune microenvironment of CC. The m6A- and m5C-related six-lncRNA signature may be a useful tool for survival stratification in CC and open new avenues for individualized therapies.

Novel machine-learning prediction tools for overall survival of patients with chondrosarcoma: Based on recursive partitioning analysis.

Chondrosarcoma (CHS), a bone malignancy, poses a significant challenge due to its heterogeneous nature and resistance to conventional treatments. There is a clear need for advanced prognostic instruments that can integrate multiple prognostic factors to deliver personalized survival predictions for individual patients. This study aimed to develop a novel prediction tool based on recursive partitioning analysis (RPA) to improve the estimation of overall survival for patients with CHS. Data from the Surveillance, Epidemiology, and End Results (SEER) database were analyzed, including demographic, clinical, and treatment details of patients diagnosed between 2000 and 2018. Using C5.0 algorithm, decision trees were created to predict survival probabilities at 12, 24, 60, and 120 months. The performance of the models was assessed through confusion scatter plot, accuracy rate, receiver operator characteristic (ROC) curve, and area under ROC curve (AUC). The study identified tumor histology, surgery, age, visceral (brain/liver/lung) metastasis, chemotherapy, tumor grade, and sex as critical predictors. Decision trees revealed distinct patterns for survival prediction at each time point. The models showed high accuracy (82.40%-89.09% in training group, and 82.16%-88.74% in test group) and discriminatory power (AUC: 0.806-0.894 in training group, and 0.808-0.882 in test group) in both training and testing datasets. An interactive web-based shiny APP (URL: https://yangxg1209.shinyapps.io/chondrosarcoma_survival_prediction/) was developed, simplifying the survival prediction process for clinicians. This study successfully employed RPA to develop a user-friendly tool for personalized survival predictions in CHS. The decision tree models demonstrated robust predictive capabilities, with the interactive application facilitating clinical decision-making. Future prospective studies are recommended to validate these findings and further refine the predictive model.

Benefits of liver transplant in critically ill patients with acute-on-chronic liver failure: Implementation of an urgent living-donor program

We evaluated the liver transplantation (LT) criteria in acute-on-chronic liver failure (ACLF), incorporating an urgent living-donor LT (LDLT) program. Critically ill patients with a Chronic Liver Failure Consortium (CLIF-C) ACLF score (CLIF-C_ACLF_score) ≥65, previously considered unsuitable for LT, were included to explore the excess mortality threshold of the CLIF-C_ACLF_score (CLIF-C_ACLF_score_threshold). We followed 854 consecutive patients with ACLF (276 ACLF grade 2 and 215 ACLF grade 3) over 10 years among 4432 LT recipients between 2008 and 2019. For advanced ACLF patients without immediate deceased-donor (DD) allocation, an urgent LDLT program was expedited. The CLIF-C_ACLF_score_threshold was determined by the metrics of transplant survival benefit: >60% 1-year and >50% 5-year survival rate. In predicting post-LT mortality, the CLIF-C_ACLF_score outperformed the (model for end-stage liver disease-sodium) MELD-Na and (model for end-stage liver disease) MELD-3.0 scores but was comparable to the Sundaram ACLF-LT-mortality score. A CLIF-C_ACLF_score ≥65 (n = 54) demonstrated posttransplant survival benefits, with 1-year and 5-year survival rates of 66.7% and 50.4% (P < .001), respectively. Novel CLIF-C_ACLF_score_threshold for 1-year and 5-year mortalities was 70 and 69, respectively. A CLIF-C_ACLF_score-based nomogram for predicting survival probabilities, integrating cardiovascular disease, diabetes, and donor type (LDLT vs DDLT), was generated. This study suggests reconsidering the criteria for unsuitable LT with a CLIF-C_ACLF_score ≥65. Implementing a timely salvage LT strategy, and incorporating urgent LDLT, can enhance survival rates.

Investigating distinct clinical features and constructing a nomogram model for survival probability in adults with cerebellar high-grade gliomas

BackgroundThe clinical features of cerebellar high-grade gliomas (cHGGs) in adults have not been thoroughly explored. This large-scale, population-based study aimed to comprehensively outline these traits and construct a predictive model.MethodsPatient records diagnosed with gliomas were collected from various cohorts and analyzed to compare the features of cHGGs and supratentorial HGGs (sHGGs). Cox regression analyses were employed to identify prognostic factors for overall survival and to develop a nomogram for predicting survival probabilities in patients with cHGGs. Multiple machine learning methods were applied to evaluate the efficacy of the predictive model.ResultsThere were significant differences in prognosis, with SEER-cHGGs showing a median survival of 7.5 months and sHGGs 14.9 months (p < 0.001). Multivariate Cox regression analyses revealed that race, WHO grade, surgical procedures, radiotherapy, and chemotherapy were independent prognostic factors for cHGGs. Based on these factors, a nomogram was developed to predict 1-, 3-, and 5-year survival probabilities, with AUC of 0.860, 0.837, and 0.810, respectively. The model’s accuracy was validated by machine learning approaches, demonstrating consistent predictive effectiveness.ConclusionsAdult cHGGs are distinguished by distinctive clinical features different from those of sHGGs and are associated with an inferior prognosis. Based on these risk factors affecting cHGGs prognosis, the nomogram prediction model serves as a crucial tool for clinical decision-making in patient care.

Novel decision tree models predict the overall survival of patients with submandibular gland cancer.

While the accurate prediction of the overall survival (OS) in patients with submandibular gland cancer (SGC) is paramount for informed therapeutic planning, the development of reliable survival prediction models has been hindered by the rarity of SGC cases. The purpose of this study is to identify key prognostic factors for OS in SGC patients using a large database and construct decision tree models to aid the prediction of survival probabilities in 12, 24, 60 and 120 months. We performed a retrospective cohort study using the Surveillance, Epidemiology and End Result (SEER) program. Demographic and peri-operative predictor variables were identified. The outcome variables overall survival at 12-, 24-, 60, and 120 months. The C5.0 algorithm was utilized to establish the dichotomous decision tree models, with the depth of tree limited within 4 layers. To evaluate the performances of the novel models, the receiver operator characteristic (ROC) curves were generated, and the metrics such as accuracy rate, and area under ROC curve (AUC) were calculated. A total of 1,705, 1,666, 1,543, and 1,413 SGC patients with a follow up of 12, 24, 60 and 120 months and exact survival status were identified from the SEER database. Predictor variables of age, sex, surgery, radiation, chemotherapy, tumor histology, summary stage, metastasis to distant lymph node, and marital status exerted substantial influence on overall survival. Decision tree models were then developed, incorporating these vital prognostic indicators. Favorable consistency was presented between the predicted and actual survival statuses. For the training dataset, the accuracy rates for the 12-, 24-, 60- and 120-month survival models were 0.866, 0.767, 0.737 and 0.797. Correspondingly, the AUC values were 0.841, 0.756, 0.725, and 0.774 for the same time points. Based on the most important predictor variables identified using the large, SEER database, decision tree models were established that predict OS of SGC patients. The models offer a more exhaustive evaluation of mortality risk and may lead to more personalized treatment strategies.

Prognostic factors for patients with pathologic T1-T2N+ esophageal squamous cell carcinoma: A retrospective study with external validation

BackgroundLymph node metastasis is significantly associated with a worse prognosis in patients with localized early-stage esophageal squamous cell carcinoma. This study aimed to explore the prognostic factors and develop a nomogram for predicting survival in patients with pathologic T1-2N+ esophageal squamous cell carcinoma. MethodsBetween 2014 and 2022, patients with pT1-2N+ esophageal squamous cell carcinoma who underwent esophagectomy with lymphadenectomy at 2 institutes were reviewed and assigned to training and external validation cohorts. Independent prognostic factors were identified via univariate and multivariate Cox regression analyses. The nomogram model was developed and evaluated by the area under the receiver operating characteristic curve and calibration curve. ResultsIn total, 268 patients with a median age of 65 years (range, 40–82) were included and assigned to training (n = 190) and external validation (n = 78) cohorts. The Cox proportional hazards model demonstrated that body mass index (P = .031), surgical approach (P < .001), T stage (P = .015), and Clavien-Dindo classification (P < .001) were independent prognostic factors in the training cohort. The nomogram showed good discrimination, with an area under the receiver operating characteristic curve for 1-year, 3-year, and 5-year of 0.810, 0.789, and 0.809 in the training cohort and 0.782, 0.679, and 0.698 in the validation cohort. The calibration curve showed that the predicted survival probability was in good agreement with the actual survival probability. ConclusionLower body mass index, left surgical approach, T2 stage, and Clavien–Dindo classification grade III to V were related to worse prognosis in patients with pT1-T2N+ esophageal squamous cell carcinoma. The developed nomogram may predict individual survival accurately.

Novel prognostic nomograms for postoperative patients with oral cavity squamous cell carcinoma in the central region of China

BackgroundOral cavity squamous cell carcinoma (OCSCC) is the most common pathological type in oral tumors. This study intends to construct a novel prognostic nomogram model based on China populations for these resectable OCSCC patients, and then validate these nomograms.MethodsA total of 607 postoperative patients with OCSCC diagnosed between June 2012 and June 2018 were obtained from two tertiary medical institutions in Xinxiang and Zhengzhou. Then, 70% of all the cases were randomly assigned to the training group and the rest to the validation group. The endpoint time was defined as overall survival (OS) and disease-free survival (DFS). The nomograms for predicting the 3-, and 5-year OS and DFS in postoperative OCSCC patients were established based on the independent prognostic factors, which were identified by the univariate analysis and multivariate analysis. A series of indexes were utilized to assess the performance and net benefit of these two newly constructed nomograms. Finally, the discrimination capability of OS and DFS was compared between the new risk stratification and the American Joint Committee on Cancer (AJCC) stage by Kaplan-Meier curves.Results607 postoperative patients with OCSCC were selected and randomly assigned to the training cohort (n = 425) and validation cohort (n = 182). The nomograms for predicting OS and DFS in postoperative OCSCC patients had been established based on the independent prognostic factors. Moreover, dynamic nomograms were also established for more convenient clinical application. The C-index for predicting OS and DFS were 0.691, 0.674 in the training group, and 0.722, 0.680 in the validation group, respectively. Besides, the calibration curve displayed good consistency between the predicted survival probability and actual observations. Finally, the excellent performance of these two nomograms was verified by the NRI, IDI, and DCA curves in comparison to the AJCC stage system.ConclusionThe newly established and validated nomograms for predicting OS and DFS in postoperative patients with OCSCC perform well, which can be helpful for clinicians and contribute to clinical decision-making.

New joint model for the dynamic prediction of heart failure in phospholamban (PLN) p.(Arg14del) positive individuals

Abstract Background PLN p.(Arg14del) positive individuals are at risk for developing arrhythmogenic cardiomyopathy which can cause severe heart failure. Future gene therapy could be lifesaving and is rapidly advancing. It is therefore of importance to identify eligible patients for this evolving treatment, with individuals risk prediction playing an imperative role in this process. Purpose To develop a dynamic prediction model for the individual risk prediction of heart failure in PLN p.(Arg14del) positive individuals. Method Data were collected of 564 PLN p.(Arg14del) positive individuals with no baseline history of heart failure or myocardial infarction. During a median follow-up of 6.3 years (interquartile range 2.9-10.0), 74 individuals experienced heart failure, defined as a composite endpoint of heart failure hospitalization, left ventricle assist device implantation, heart transplantation and heart failure-related death. For the individual prediction of time to heart failure we used a joint model which simultaneously estimates a longitudinal model for repeatedly measured left ventricle ejection fraction (LVEF) during follow-up and a Cox regression model for the hazards of the heart failure endpoint. Using the longitudinal model we predicted the LVEF-values of all patients at future time points using the estimated random-effects and these predicted LVEF values were used as a time-dependent covariate in the Cox model which also contained age at start of follow-up as predictor variable. Results The joint model trained in this study had a very good performance in discriminating the individuals at risk for heart failure, with the 5-year mean AUC of 0.86 and 10-year mean AUC of 0.84. The LVEF had a hazard ratio of 0.89 (95% confidence interval (CI) 0.86-0.93, p&amp;lt;0.001), and age at baseline 1.04 (95% CI 1.03-1.06], p&amp;lt;0.001), meaning a lower LVEF and higher age at presentation were associated with heart failure. Figure 1 shows an example of an individual risk prediction with on the x-axis the follow-up time in years, on the y-axis (left) the LVEF in % and on the y-axis (right) the predicted survival probability. Conclusion Individual risk prediction with a simple model including only LVEF and age is of great importance for identifying PLN p.(Arg14del) positive individuals who are at risk of developing heart failure. The introduction of this novel joint prediction model is the first step in identifying individuals at risk for heart failure and may aid patient selection for future gene therapy interventions.

#426 “Predict-PD”: a machine learning tool to predict patient survival in peritoneal dialysis patients

Abstract Background and Aims With the rising prevalence of chronic kidney disease (CKD), particularly among the elderly, the need for effective renal replacement therapy has become increasingly crucial. Peritoneal dialysis (PD), recognized as a gentler alternative to haemodialysis, offers advantages in cognitive preservation and overall quality of life. Patient survival on PD, a tangible outcome measure, gains particular importance, especially in an aging demographic confronting end-stage kidney disease. Predicting survival thus becomes vital in facilitating individualized treatment planning and has implications on resource allocation and patient counseling. Method Our study analysed data from the UK Renal Registry and included 22 711 incident dialysis patients choosing PD between 2007 and 2022. Our objective was to employ Artificial Intelligence (AI) algorithms in developing an advanced risk stratification indicator with high predictive capabilities, specifically tailored for use in the United Kingdom's transplant selection procedure. We conducted experiments using three distinct machine learning models and evaluated their performance in terms of calibration and discrimination, utilizing metrics such as the integrated Brier score (IBS) and Harrell's concordance index. We assessed the potential clinical utility using decision curve analysis. Results The XGBoost model demonstrated excellent predictive efficacy for patient survival, with a concordance of 0.78. Further scrutiny revealed AUC values at 1 year (0.81), 3 years (0.78), and 5 years (0.77). The integrated Brier score, a comprehensive measure of predictive accuracy, was 0.09. There was no statistical difference between the actual and predicted survival probabilities (P = 0.72). Decision curve analysis accentuated the model's clinical applicability. Conclusion This study illustrates the potential of machine learning in prognosticating patient survival in PD, offering a pathway towards individualized patient-centric management in nephrology. The findings advocate for the integration of data-driven methodologies in clinical decision-making, paving the way for enhanced personalized care in CKD management and can serve as a guide for future research.

Identifying and Validating an Angiogenesis-related Signature for the Prognosis of Head and Neck Squamous Cell Carcinoma.

The present study aimed todevelop a prognostic model for HNSCC treatment on the basis of angiogenesis-related signatures. Head and Neck Squamous Cell Carcinoma (HNSCC) is the most frequent malignancy with poor prognostic outcomes in the head and neck. Angiogenesis plays a critical role in tumorigenesis and is expected to be an effective therapeutic target. The RNA-seq dataset TCGA-HNSCC and the hallmark gene set were used for angiogenesis-related RiskScore model construction. The RNA-seq data was downloaded from The Cancer Genome Atlas (TCGA), and the hallmark gene set was used to measure the angiogenesis score using the GSVA R package. Then, the optimal cutoff point for prognostic classification was calculated by the survminer package, and Weighted Gene Co-expression Network Analysis (WGCNA) was used to identify angiogenesis gene modules . Multi/univariable and Lasso Cox analyses were performed to develop the RiskScore model, and the classifier efficiency was evaluated by the Receiver Operating Characteristic curve (ROC). Furthermore, a nomogram was designed for survival probability prediction, and the immune infiltration and immunotherapy differences among different risk patients were assessed. After calculating the angiogenesis score, we found that this indicator and patients' prognosis were closely correlated, especially when patients with a high angiogenesis score had a poor prognosis. Then, WGCNA identified a blue gene module positively correlated with angiogenesis. Multivariate and Lasso Cox analysis further identified 9 risk model genes for developing a RiskScore, which was used to divide low- and high- -risk groups of patients. Those with a high risk tended to show poor prognosis, immune infiltration, and higher immune escape. Finally, a nomogram was developed to optimize the risk model, and it exhibited excellent short- and long-term survival prediction performance. We constructed a reliable RiskScore model for the prognostic prediction of HNSCC patients, contributing to precise therapeutic intervention of the cancer.

Nomogram Predicts Prognostic Factors for Head and Neck Cutaneous Melanoma: A Population-Based Analysis

BackgroundThe head and neck cutaneous melanoma (HNCM) accounts for 20% of newly diagnosed melanoma. Research on prognostic models for their survival yet remains largely unexplored. This study employed a nomogram approach to develop and validate a predictive model for both overall survival (OS) and disease-specific survival (DSS) in patients with HNCM. MethodsThis study analyzed the HNCM patients diagnosed between 2004 and 2014 from SEER database. To identify independent prognostic factors for HNCM, we integrated results from univariate Cox regression analysis, random survival forests, and LASSO regression with cross-validation. A nomogram was designed and validated based on the identified characteristics to predict the 3-, 5-, and 8-year OS and DSS of patients with HNCM. ResultsAge, Stage, Ulceration, Thickness, Chemotherapy, LN_Metastasis, Radiation were identified as independent prognostic factors. The nomogram achieved a satisfactory performance with C-indices of 0.824(DSS) and 0.757(OS) in the training cohort and 0.827(DSS) and 0.749(OS) in the validation cohort, respectively. The AUC for the OS at 3-, 5-, and 8- years were 0.789, 0.788, and 0.794 for the training cohort, and 0.778, 0.776, and 0.795 for the validation cohort, respectively. For DSS, the AUC at 3-, 5-, and 8- years were 0.859, 0.842, and 0.828 in the training cohort, and 0.864, 0.844, and 0.834 in the validation cohort, respectively. The calibration curve showed that there was a strong correlation between the observed outcomes and the predicted survival probability. ConclusionThis study established and validated predictive nomograms for HNCM patients with robust predictive performance.

Identification of prognostic signatures in remnant gastric cancer through an interpretable risk model based on machine learning: a multicenter cohort study

ObjectiveThe purpose of this study was to develop an individual survival prediction model based on multiple machine learning (ML) algorithms to predict survival probability for remnant gastric cancer (RGC).MethodsClinicopathologic data of 286 patients with RGC undergoing operation (radical resection and palliative resection) from a multi-institution database were enrolled and analyzed retrospectively. These individuals were split into training (80%) and test cohort (20%) by using random allocation. Nine commonly used ML methods were employed to construct survival prediction models. Algorithm performance was estimated by analyzing accuracy, precision, recall, F1-score, area under the receiver operating characteristic curve (AUC), confusion matrices, five-fold cross-validation, decision curve analysis (DCA), and calibration curve. The best model was selected through appropriate verification and validation and was suitably explained by the SHapley Additive exPlanations (SHAP) approach.ResultsCompared with the traditional methods, the RGC survival prediction models employing ML exhibited good performance. Except for the decision tree model, all other models performed well, with a mean ROC AUC above 0.7. The DCA findings suggest that the developed models have the potential to enhance clinical decision-making processes, thereby improving patient outcomes. The calibration curve reveals that all models except the decision tree model displayed commendable predictive performance. Through CatBoost-based modeling and SHAP analysis, the five-year survival probability is significantly influenced by several factors: the lymph node ratio (LNR), T stage, tumor size, resection margins, perineural invasion, and distant metastasis.ConclusionsThis study established predictive models for survival probability at five years in RGC patients based on ML algorithms which showed high accuracy and applicative value.

Geriatric nutritional risk index predicts the prognosis of gastric cancer patients treated with immune checkpoint inhibitors.

The nutritional status is closely linked to the immune function of patients. Previous studies have demonstrated the utility of the Geriatric Nutritional Risk Index (GNRI) in assessing nutritional status. The aim of this study is to investigate the prognostic significance of GNRI in patients with gastric cancer who received immune checkpoint inhibitor (ICI) therapy. The study enrolled 89 gastric cancer patients who received different types of immune checkpoint inhibitors (ICIs) between August 2016 and December 2020, along with 57 patients who underwent chemotherapy during the same period as a control group. The GNRI cutoff point was established based on prior research. Differences in clinical and pathological features were analyzed using the Chi-square test or independent samples t-test. Univariate and multivariate analyses were used to identify prognostic factors for both progression-free survival (PFS) and overall survival (OS). Furthermore, nomograms were created to predict the likelihood of patient survival. There were 31 cases (21.2%) with GNRI < 92.00 and 115 cases (78.8%) with GNRI ≥ 92.00. Patients with low GNRI had significantly shorter PFS (21.33 months vs 28.37 months, P = .001) and OS (33.06 months vs 41.63 months, P = .001) than those with high GNRI, among all patients. Similar results were also found in patients treated with ICIs. Additionally, GNRI was identified as an independent prognostic factor. The C-index and 95% CI of the nomograms for predicting survival probabilities were 0.667 (0.600-0.735) and 0.685 (0.622-0.749), respectively. GNRI was significantly associated with survival time in patients with gastric cancer who received ICIs, patients with low GNRI had shorter PFS and OS. GNRI might be able to identify patients who might benefit from ICIs.

Cardiovascular mortality risk in patients with ovarian cancer: a population-based study

ObjectivesOvarian cancer (OC) can occur at different ages and is affected by a variety of factors. In order to evaluate the risk of cardiovascular mortality in patients with ovarian cancer, we included influencing factors including age, histological type, surgical method, chemotherapy, whether distant metastasis, race and developed a nomogram to evaluate the ability to predict occurrence. At present, we have not found any correlation studies on cardiovascular death events in patients with ovarian cancer. This study was designed to provide targeted measures for effective prevention of cardiovascular death in patients with ovarian cancer.MethodsKaplan–Meier analysis and multivariable Cox proportional model were performed to evaluate the effectiveness of cardiovascular diseases on overall survival (OS) and ovarian cancer‐specific survival (OCSS). We compared multiple groups including clinical, demographic, therapeutic characteristics and histological types. Cox risk regression analysis, Kaplan–Meier survival curves, and propensity score matching were employed for analyzing the data.ResultsA total of 88,653 ovarian cancer patients were collected, of which 2,282 (2.57%) patients died due to cardiovascular-related diseases. Age, chemotherapy and whether satisfactory cytoreduction surgery is still the most important factors affecting the prognosis of ovarian cancer patients, while different histological types, diagnosis time, and race also have a certain impact on the prognosis. The newly developed nomogram model showed excellent predictive performance, with a C-index of 0.759 (95%CI: 0.757–0.761) for the group. Elderly patients with ovarian cancer are still a high-risk group for cardiovascular death [HR: 21.07 (95%CI: 5.21–85.30), p < 0.001]. The calibration curve showed good agreement from predicted survival probabilities to actual observations.ConclusionThis study found that age, histology, surgery, race, chemotherapy, and tumor metastasis are independent prognostic factors for cardiovascular death in patients with ovarian cancer. The nomogram-based model can accurately predict the OS of ovarian cancer patients. It is expected to inform clinical decision-making and help develop targeted treatment strategies for this population.

A clinical trial termination prediction model based on denoising autoencoder and deep survival regression

AbstractEffective clinical trials are necessary for understanding medical advances but early termination of trials can result in unnecessary waste of resources. Survival models can be used to predict survival probabilities in such trials. However, survival data from clinical trials are sparse, and DeepSurv cannot accurately capture their effective features, making the models weak in generalization and decreasing their prediction accuracy. In this paper, we propose a survival prediction model for clinical trial completion based on the combination of denoising autoencoder (DAE) and DeepSurv models. The DAE is used to obtain a robust representation of features by breaking the loop of raw features after autoencoder training, and then the robust features are provided to DeepSurv as input for training. The clinical trial dataset for training the model was obtained from the ClinicalTrials.gov dataset. A study of clinical trial completion in pregnant women was conducted in response to the fact that many current clinical trials exclude pregnant women. The experimental results showed that the denoising autoencoder and deep survival regression (DAE‐DSR) model was able to extract meaningful and robust features for survival analysis; the C‐index of the training and test datasets were 0.74 and 0.75 respectively. Compared with the Cox proportional hazards model and DeepSurv model, the survival analysis curves obtained by using DAE‐DSR model had more prominent features, and the model was more robust and performed better in actual prediction.

The potential of machine learning models to identify malnutrition diagnosed by GLIM combined with NRS-2002 in colorectal cancer patients without weight loss information

Background & AimsThe key step of the Global Leadership Initiative on Malnutrition (GLIM) is nutritional risk screening, while the most appropriate screening tool for colorectal cancer (CRC) patients is yet unknown. The GLIM diagnosis relies on weight loss information, and bias or even failure to recall patients’ historical weight can cause misestimates of malnutrition. We aimed to compare the suitability of several screening tools in GLIM diagnosis, and establish machine learning (ML) models to predict malnutrition in CRC patients without weight loss information. MethodsThis multicenter cohort study enrolled 4487 CRC patients. The capability of GLIM diagnoses combined with four screening tools in predicting survival probability was compared by Kaplan-Meier curves, and the most accurate one was selected as the malnutrition reference standard. Participants were randomly assigned to a training cohort (n=3365) and a validation cohort (n=1122). Several ML approaches were adopted to establish models for predicting malnutrition without weight loss data. We estimated feature importance and reserved the top 30% of variables for retraining simplified models. The area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity were calculated to assess and compare model performance. ResultsNRS-2002 was the most suitable screening tool for GLIM diagnosis in CRC patients, with the highest hazard ratio (1.59; 95% CI, 1.43-1.77). A total of 2076 (46.3%) patients were malnourished diagnosed by GLIM combined with NRS-2002. The simplified random forest (RF) model outperformed other models with an AUC of 0.830 (95% CI, 0.805-0.854), and accuracy, sensitivity and specificity were 0.775, 0.835 and 0.742, respectively. We deployed an online application based on the simplified RF model to accurately estimate malnutrition probability in CRC patients without weight loss information (https://zzuwtt1998.shinyapps.io/dynnomapp/). ConclusionsNutrition Risk Screening 2002 was the optimal initial nutritional risk screening tool in the GLIM process. The RF model outperformed other models, and an online prediction tool was developed to properly identify patients at high risk of malnutrition.