Predictor Of Survival In Patients Research Articles

A modified GLIM criteria-based nomogram for the survival prediction of gastric cancer patients undergoing surgical resection

BackgroundThis study aimed to develop a comprehensive model based on five GLIM variables to predict the individual survival and provide more appropriate patient counseling.MethodsThis retrospective cohort study included 301 gastric cancer (GC) patients undergoing radical resection. C-reactive protein (CRP) as an inflammatory marker was included in GLIM criteria and a nomogram for predicting 5-year overall survival (OS) in GC patients was established. The Bootstrap repeated sampling for 1000 times was used for internal validation.ResultsOf the total 301 patients, 20 (6.64%) died within 5 years. CRP improved the sensitivity and accuracy of the survival prediction model (AUC = 0.782, 0.694 to 0.869 for the model without CRP; AUC = 0.880, 0.809 to 0.950 for the model adding CRP). Besides, a GLIM-based nomogram was established with an AUC of 0.889. The C-index for predicting OS was 0.878 (95% CI: 0.823 to 0.934), and the calibration curve fitted well. Decision curve analysis (DCA) showed the clinical utility of the nomogram based on GLIM.ConclusionThe addition of CRP improved the sensitivity and accuracy of the survival prediction model. The 5-year survival probability of GC patients undergoing radical resection can be reliably predicted by the nomogram presented in this study.

Integrated MRI radiomics, tumor microenvironment, and clinical risk factors for improving survival prediction in patients with glioblastomas.

To construct acomprehensive model for predicting the prognosis of patients with glioblastoma (GB) using aradiomics method and integrating clinical risk factors, tumor microenvironment (TME), and imaging characteristics. In this retrospective study, we included 148 patients (85males and 63females; median age 53years) with isocitrate dehydrogenase-wildtype GB between January 2016 and April 2022. Patients were randomly divided into the training (n = 104) and test (n = 44) sets. The best feature combination related to GB overall survival (OS) was selected using LASSO Cox regression analyses. Clinical, radiomics, clinical-radiomics, clinical-TME, and clinical-radiomics-TME models were established. The models' concordance index (C-index) was evaluated. The survival curve was drawn using the Kaplan-Meier method, and the prognostic stratification ability of the model was tested. LASSO Cox analyses were used to screen the factors related to OS in patients with GB, including MGMT (hazard ratio [HR] = 0.642; 95% CI 0.414-0.997; P = 0.046), TERT (HR = 1.755; 95% CI 1.095-2.813; P = 0.019), peritumoral edema (HR = 1.013; 95% CI 0.999-1.027; P = 0.049), tumor purity (TP; HR = 0.982; 95% CI 0.964-1.000; P = 0.054), CD163 + tumor-associated macrophages (TAMs; HR = 1.049; 95% CI 1.021-1.078; P < 0.001), CD68 + TAMs (HR = 1.055; 95% CI 1.018-1.093; P = 0.004), and the six radiomics features. The clinical-radiomics-TME model had the best survival prediction ability, the C‑index was 0.768 (0.717-0.819). The AUC of 1‑, 2‑, and 3‑year OS prediction in the test set was 0.842, 0.844, and 0.795, respectively. The clinical-radiomics-TME model is the most effective for predicting the survival of patients with GB. Radiomics features, TP, and TAMs play important roles in the prognostic model.

Developing survival prediction models in colorectal cancer using epigenome-wide DNA methylation data from whole blood

While genome-wide association studies are valuable in identifying CRC survival predictors, the benefit of adding blood DNA methylation (blood-DNAm) to clinical features, including the TNM system, remains unclear. In a multi-site population-based patient cohort study of 2116 CRC patients with baseline blood-DNAm, we analyzed survival predictions using eXtreme Gradient Boosting with a 5-fold nested leave-sites-out cross-validation across four groups: traditional and comprehensive clinical features, blood-DNAm, and their combination. Model performance was assessed using time-dependent ROC curves and calibrations. During a median follow-up of 10.3 years, 1166 patients died. Although blood-DNAm-based predictive signatures achieved moderate performances, predictive signatures based on clinical features outperformed blood-DNAm signatures. The inclusion of blood-DNAm did not improve survival prediction over clinical features. M1 stage, age at blood collection, and N2 stage were the top contributors. Despite some prognostic value, incorporating blood DNA methylation did not enhance survival prediction of CRC patients beyond clinical features.

Radiogenomics based survival prediction of small-sample glioblastoma patients by multi-task DFFSP model.

In this paper, the multi-task dense-feature-fusion survival prediction (DFFSP) model is proposed to predict the three-year survival for glioblastoma (GBM) patients based on radiogenomics data. The contrast-enhanced T1-weighted (T1w) image, T2-weighted (T2w) image and copy number variation (CNV) is used as the input of the three branches of the DFFSP model. This model uses two image extraction modules consisting of residual blocks and one dense feature fusion module to make multi-scale fusion of T1w and T2w image features as backbone. Also, a gene feature extraction module is used to adaptively weight CNV fragments. Besides, a transfer learning module is introduced to solve the small sample problem and an image reconstruction module is adopted to make the model anatomy-aware under a multi-task framework. 256 sample pairs (T1w and corresponding T2w MRI slices) and 187 CNVs of 74 patients were used. The experimental results show that the proposed model can predict the three-year survival of GBM patients with the accuracy of 89.1 %, which is improved by 3.2 and 4.7 % compared with the model without genes and the model using last fusion strategy, respectively. This model could also classify the patients into high-risk and low-risk groups, which will effectively assist doctors in diagnosing GBM patients.

Prognostic nomogram combining 18F-FDG PET/CT radiomics and clinical data for stage III NSCLC survival prediction

The aim of this study was to establish and validate the precision of a novel radiomics approach that integrates 18Fluorine-fluorodeoxyglucose (18F-FDG) positron emission tomography (PET)-computed tomography (CT) scan data with clinical information to improve the prognostication of survival rates in patients diagnosed with stage III Non-Small Cell Lung Cancer (NSCLC) who are not candidates for surgery. We evaluated pretreatment 18F-FDG PET-CT scans from 156 individuals diagnosed with stage III inoperable NSCLC at Shandong Cancer Hospital. These individuals were divided into two groups: a training set comprising 110 patients and an internal validation set consisting of 46 patients. By employing random forest classifier and cox proportional hazards model , we identified and utilized relevant features to create predictive models and a nomogram. The effectiveness of these models was assessed through the use of the receiver operating characteristics(ROC) curves, Kaplan–Meier (KM) curves, and the application of the nomogram. Our findings showed that the combined model, which integrates both clinical and radiomic data, outperformed those based solely on clinical or radiomic features in predicting 3-year overall survival(OS). Furthermore, calibration plots revealed a high level of agreement between predicted and actual survival times. The research successfully established a predictive radiomics model that integrates 18F-FDG PET/CT imaging with clinical indicators to enhance survival predictions for patients with stage III inoperable NSCLC.

933P Prediction of survival in patients with head and neck merkel cell carcinoma: Statistical and machine-learning approaches

933P Prediction of survival in patients with head and neck merkel cell carcinoma: Statistical and machine-learning approaches

Primary lung signet-ring cell carcinoma: a national analysis.

Primary lung signet-ring cell carcinoma (LSRCC) is a rare form of aggressive lung cancer whose clinical features remain inadequately discerned. The objective of this study was to evaluate the clinicopathological characteristics and independent prognostic factors of primary LSRCC. Overall survival (OS) of patients with LSRCC, lung adenocarcinoma (LAC), and lung mucinous adenocarcinoma (LMAC) in the National Cancer Database from 2004 to 2018 was evaluated using Kaplan-Meier and multivariable Cox proportional hazards modeling. Independent prognostic indicators for patients with LSRCC were determined using multivariable Cox proportional hazards analysis. A total of 1,705 LSRCC, 504,006 LAC, and 15,883 LMAC patients were included in our analysis. LSRCC histology was significantly associated with younger age, male sex, larger and more poorly differentiated tumors, later American Joint Committee on Cancer (AJCC) stage disease, higher clinical T, N, and M status, more use of chemotherapy, and less use of surgery when compared to LAC and LMAC patients. In unadjusted analysis, patients with LSRCC had significantly worse OS when compared to patients with LAC and LMAC. In multivariable analysis, patients with LSRCC experienced significantly worse OS when compared to only patients with LAC. Independent predictors of survival for patients with LSRCC were younger age, later year of diagnosis, lower Charlson/Deyo comorbidity condition scores, lower AJCC stage, higher income, smaller tumors, treatment with surgery, and receipt of chemotherapy. In this national analysis, LSRCC was found to be associated with distinct clinicopathological characteristics from those of LAC.

Telomere-related prognostic signature for survival assessments in lung adenocarcinoma.

Telomere-related genes (TRGs) are important in many different types of cancers. However, there is a lack of research on the relationship between their expression and prognosis in lung adenocarcinoma (LUAD) patients. This study is to investigate the prognostic value of TRGs in LUAD and to develop a TRG signature that can predict patient survival. A total of 2,086 TRGs were obtained from a database of genes involved in telomere maintenance (TelNet), while the clinical information and tumor RNA expression profiles of 513 LUAD patients were acquired from The Cancer Genome Atlas (TCGA) database. Statistical methodologies, such as least absolute shrinkage and selection operator (LASSO)-Cox, were employed to construct a prognostic model with predictive capabilities. We analyzed 1,339 telomere-associated differentially expressed genes and identified a ten-gene predictive signature for LUAD. This signature exhibited effective prognostic classification capabilities across multiple datasets, including GSE3141 (58 samples), GSE8894 (63 samples), GSE50081 (127 samples), and GSE72094 (398 samples). Furthermore, we screened tumor-sensitive drugs targeting this signature. High telomere levels were associated with reduced survival in lung cancer patients who underwent surgery. Compared to the traditional TNM (tumor node metastasis classification) grading method, our telomere-associated gene panel demonstrated superior prediction accuracy. Notably, patients in the high-risk group, defined by the telomere-associated signature, exhibited improved responses to immunotherapy, suggesting potential benefits for this subgroup of patients. This study presents a comprehensive molecular signature comprising TRGs, which holds potential for functional and therapeutic investigations. Additionally, it serves as an integrated tool to identify crucial molecules for immunotherapy in lung cancer.

Tumor burden with AFP improves survival prediction for TACE-treated patients with HCC: An international observational study☆

Background & aimsCurrent prognostic models for patients with HCC undergoing transarterial chemoembolization (TACE) are not extensively validated and widely accepted. We aimed to develop and validate a continuous model incorporating tumor burden and biology for individual survival prediction and risk stratification. MethodsOverall, 4,377 treatment-naïve recommended TACE candidates from 39 centres in 5 countries were enrolled and divided into a training, internal validation, and two external validation datasets. The novel model was developed using a Cox multivariable regression analysis and compared with our original 6-and-12 model (the largest tumor size [ts, centimetres] + tumor number [tn]) and other available models in terms of predictive accuracy. ResultsThe proposed model named the ‘6-and-12 model 2.0’ was generated as ‘ts + tn + 1.5 × log10 alpha-fetoprotein [AFP])’, showed good discrimination (C-index 0.674) and calibration (Hosmer-Lemeshow test p=0.147), and outperformed current existing models. An easy-to-use stratification was proposed according to the different AFP levels (≤100, 100–400, 400–2,000, 2,000–10,000, 10,000–40,000, and >40,000 ng/ml) along with the corresponding tumor burden cut-offs (8/14, 7/13, 6/12, 5/11, 4/10, and any tumor burden), i.e. if the AFP level was 400–2,000 ng/ml, the stratification should be low-(≤6)/intermediate-(6-12)/high-risk (>12) strata. Hence, it could divide the patients into three distinct risk categories with a median overall survival of 45.0 (95% confidence interval [CI], 40.1–49.9), 30.0 (95% CI, 26.1–33.9), and 15.4 (95% CI, 13.4–17.4) months (p<0.001) from low-risk to high-risk strata, respectively. These findings were confirmed in validation and subgroup analyses. ConclusionsThe 6-and-12 model 2.0 significantly improved individual outcome prediction and better stratified the recommended TACE candidates, which could be used in clinical practice, as well as trial design.

2006P Prognostic factors in metastatic urothelial cancer (mUC): Developing an accessible model for predicting patient survival

2006P Prognostic factors in metastatic urothelial cancer (mUC): Developing an accessible model for predicting patient survival

247.5: Mitochondrial flavin mononucleotide measured at beginning of hypothermic machine perfusion predicts patient survival after liver transplantation.

247.5: Mitochondrial flavin mononucleotide measured at beginning of hypothermic machine perfusion predicts patient survival after liver transplantation.

Multi-task deep latent spaces for cancer survival and drug sensitivity prediction.

Cancer is a very heterogeneous disease that can be difficult to treat without addressing the specific mechanisms driving tumour progression in a given patient. High-throughput screening and sequencing data from cancer cell-lines has driven many developments in drug development, however, there are important aspects crucial to precision medicine that are often overlooked, namely the inherent differences between tumours in patients and the cell-lines used to model them in vitro. Recent developments in transfer learning methods for patient and cell-line data have shown progress in translating results from cell-lines to individual patients in silico. However, transfer learning can be forceful and there is a risk that clinically relevant patterns in the omics profiles of patients are lost in the process. We present MODAE, a novel deep learning algorithm to integrate omics profiles from cell-lines and patients for the purposes of exploring precision medicine opportunities. MODAE implements patient survival prediction as an additional task in a drug-sensitivity transfer learning schema and aims to balance autoencoding, domain adaptation, drug-sensitivity prediction, and survival prediction objectives in order to better preserve the heterogeneity between patients that is relevant to survival. While burdened with these additional tasks, MODAE performed on par with baseline survival models, but struggled in the drug-sensitivity prediction task. Nevertheless, these preliminary results were promising and show that MODAE provides a novel AI-based method for prioritizing drug treatments for high-risk patients. https://github.com/UEFBiomedicalInformaticsLab/MODAE.

Delayed kernels for longitudinal survival analysis and dynamic prediction.

Predicting patient survival probabilities based on observed covariates is an important assessment in clinical practice. These patient-specific covariates are often measured over multiple follow-up appointments. It is then of interest to predict survival based on the history of these longitudinal measurements, and to update predictions as more observations become available. The standard approaches to these so-called 'dynamic prediction' assessments are joint models and landmark analysis. Joint models involve high-dimensional parameterizations, and their computational complexity often prohibits including multiple longitudinal covariates. Landmark analysis is simpler, but discards a proportion of the available data at each 'landmark time'. In this work, we propose a 'delayed kernel' approach to dynamic prediction that sits somewhere in between the two standard methods in terms of complexity. By conditioning hazard rates directly on the covariate measurements over the observation time frame, we define a model that takes into account the full history of covariate measurements but is more practical and parsimonious than joint modelling. Time-dependent association kernels describe the impact of covariate changes at earlier times on the patient's hazard rate at later times. Under the constraints that our model (a) reduces to the standard Cox model for time-independent covariates, and (b) contains the instantaneous Cox model as a special case, we derive two natural kernel parameterizations. Upon application to three clinical data sets, we find that the predictive accuracy of the delayed kernel approach is comparable to that of the two existing standard methods.

Organomics: A Concept Reflecting the Importance of PET/CT Healthy Organ Radiomics in Non-Small Cell Lung Cancer Prognosis Prediction Using Machine Learning.

Non-small cell lung cancer is the most common subtype of lung cancer. Patient survival prediction using machine learning (ML) and radiomics analysis proved to provide promising outcomes. However, most studies reported in the literature focused on information extracted from malignant lesions. This study aims to explore the relevance and additional value of information extracted from healthy organs in addition to tumoral tissue using ML algorithms. This study included PET/CT images of 154 patients collected from available online databases. The gross tumor volume and 33 volumes of interest defined on healthy organs were segmented using nnU-Net deep learning-based segmentation. Subsequently, 107 radiomic features were extracted from PET and CT images (Organomics). Clinical information was combined with PET and CT radiomics from organs and gross tumor volumes considering 19 different combinations of inputs. Finally, different feature selection (FS; 5 methods) and ML (6 algorithms) algorithms were tested in a 3-fold data split cross-validation scheme. The performance of the models was quantified in terms of the concordance index (C-index) metric. For an input combination of all radiomics information, most of the selected features belonged to PET Organomics and CT Organomics. The highest C-index (0.68) was achieved using univariate C-index FS method and random survival forest ML model using CT Organomics + PET Organomics as input as well as minimum depth FS method and CoxPH ML model using PET Organomics as input. Considering all 17 combinations with C-index higher than 0.65, Organomics from PET or CT images were used as input in 16 of them. The selected features and C-indices demonstrated that the additional information extracted from healthy organs of both PET and CT imaging modalities improved the ML performance. Organomics could be a step toward exploiting the whole information available from multimodality medical images, contributing to the emerging field of digital twins in health care.

Automated 3D-Body Composition Analysis as a Predictor of Survival in Patients With Idiopathic Pulmonary Fibrosis.

Idiopathic pulmonary fibrosis (IPF) is the most common interstitial lung disease, with a median survival time of 2 to 5 years. The focus of this study is to establish a novel imaging biomarker. In this study, 79 patients (19% female) with a median age of 70 years were studied retrospectively. Fully automated body composition analysis (BCA) features (bone, muscle, total adipose tissue, intermuscular, and intramuscular adipose tissue) were combined into Sarcopenia, Fat, and Myosteatosis indices and compared between patients with a survival of more or less than 2 years. In addition, we divided the cohort at the median (high=≥ median, low=<median) of the respective BCA index and tested the impact on the overall survival using the Kaplan-Meier methodology, a log-rank test, and adjusted multivariate Cox-regression analysis. A high Sarcopenia and Fat index and low Myosteatosis index were associated with longer median survival (35 vs. 16mo for high vs. low Sarcopenia index, P=0.066; 44 vs. 14mo for high vs. low Fat index, P<0.001; and 33 vs. 14mo for low vs. high Myosteatosis index, P=0.0056) and better 5-year survival rates (34.0% vs. 23.6% for high vs. low Sarcopenia index; 47.3% vs. 9.2% for high vs. low Fat index; and 11.2% vs. 42.7% for high vs. low Myosteatosis index). Adjusted multivariate Cox regression showed a significant impact of the Fat (HR=0.71, P=0.01) and Myosteatosis (HR=1.12, P=0.005) on overall survival. The fully automated BCA provides biomarkers with a predictive value for the overall survival in patients with IPF.

CT-based multimodal deep learning for non-invasive overall survival prediction in advanced hepatocellular carcinoma patients treated with immunotherapy

ObjectivesTo develop a deep learning model combining CT scans and clinical information to predict overall survival in advanced hepatocellular carcinoma (HCC).MethodsThis retrospective study included immunotherapy-treated advanced HCC patients from 52 multi-national in-house centers between 2018 and 2022. A multi-modal prognostic model using baseline and the first follow-up CT images and 7 clinical variables was proposed. A convolutional-recurrent neural network (CRNN) was developed to extract spatial-temporal information from automatically selected representative 2D CT slices to provide a radiological score, then fused with a Cox-based clinical score to provide the survival risk. The model’s effectiveness was assessed using a time-dependent area under the receiver operating curve (AUC), and risk group stratification using the log-rank test. Prognostic performances of multi-modal inputs were compared to models of missing modality, and the size-based RECIST criteria.ResultsTwo-hundred seven patients (mean age, 61 years ± 12 [SD], 180 men) were included. The multi-modal CRNN model reached the AUC of 0.777 and 0.704 of 1-year overall survival predictions in the validation and test sets. The model achieved significant risk stratification in validation (hazard ratio [HR] = 3.330, p = 0.008), and test sets (HR = 2.024, p = 0.047) based on the median risk score of the training set. Models with missing modalities (the single-modal imaging-based model and the model incorporating only baseline scans) can still achieve favorable risk stratification performance (all p < 0.05, except for one, p = 0.053). Moreover, results proved the superiority of the deep learning-based model to the RECIST criteria.ConclusionDeep learning analysis of CT scans and clinical data can offer significant prognostic insights for patients with advanced HCC.Critical relevance statementThe established model can help monitor patients’ disease statuses and identify those with poor prognosis at the time of first follow-up, helping clinicians make informed treatment decisions, as well as early and timely interventions.Key PointsAn AI-based prognostic model was developed for advanced HCC using multi-national patients.The model extracts spatial-temporal information from CT scans and integrates it with clinical variables to prognosticate.The model demonstrated superior prognostic ability compared to the conventional size-based RECIST method.Graphical

Association of changes in body composition with all-cause mortality in patients undergoing hemodialysis: A prospective cohort study

ObjectivesThis study aimed to explore the effect of longitudinal body composition changes on mortality risk in patients undergoing hemodialysis and identify whether changes in body composition can more accurately predict mortality than baseline status. MethodsA prospective cohort study was conducted on 340 patients undergoing hemodialysis. Lean mass and body fat were determined using a bioimpedance spectroscopy (BIS) device and expressed as the lean tissue index (LTI) or fat tissue index (FTI), respectively. The patients were subjected to BIS at baseline and after 1 year. The hazard ratio (HR) for death was calculated using Cox regression analysis. ResultsAmong 340 patients, 289 were tested with a repeat BIS. LTI loss and FTI gain were observed in 51.2% and 47.1% of the patients, respectively. Low baseline LTI was a significant predictor of all-cause mortality after adjusting for demographic and biochemical parameters (HR, 2.41; P = 0.047), but not when comorbidities were included in the multivariate analysis. However, after adjusting for various confounding factors, LTI loss (HR, 3.40; P = 0.039) and FTI gain (HR, 4.06; P = 0.024) were independent risk factors for all-cause mortality, and the adjusted HR for LTI loss and FTI gain vs. no LTI loss and no FTI gain was 5.34 (P = 0.016). ConclusionsLTI loss and FTI gain, particularly their combination, are important predictors of survival in patients undergoing hemodialysis. Our results emphasize that longitudinal changes in LTI and FTI are more strongly associated with all-cause mortality than single-point values. Therefore, it is important to dynamically assess the muscle and fat tissues and develop potential targeted treatment strategies for this population.

Spatial transcriptomic revealed intratumor heterogeneity and cancer stem cell enrichment in colorectal cancer metastasis

Metastasis is the main cause of mortality in colorectal cancer (CRC) patients. Exploring the mechanisms of metastasis is of great importance in both clinical and fundamental CRC research. CRC is a highly heterogeneous disease with variable therapeutic outcomes of treatment. In this study, we applied spatial transcriptomics (ST) to generate a tissue-wide transcriptome from two primary colorectal cancer tissues and their matched liver metastatic tissues. Spatial RNA information showed intratumoral heterogeneity (ITH) of both primary and metastatic tissues. The comparison of gene expressions across tissues revealed an apparent enrichment of cancer stem cells (CSCs) in metastatic tissues and identified FOXD1 as a novel metastatic CSC marker. Trajectory and pseudo-time analyses revealed distinct evolutionary trajectories and a dedifferentiation-differentiation process during metastasis. CellphoneDB analysis suggested a dominant interaction of CD74-MIF with tumor cells in metastatic tissues. Further analysis confirmed FOXD1 as a maker of CSCs and the predictor of patient survival, especially in metastatic diseases. Our study found ITH of primary and metastatic tissues and provides novel insights into the cellular mechanisms underlying liver metastasis of CRC and foundations for therapeutic strategies for CRC metastasis.

Multimodal MRI-based deep-radiomics model predicts response in cervical cancer treated with neoadjuvant chemoradiotherapy

Platinum-based neoadjuvant chemotherapy (NACT) followed by radical hysterectomy has been proposed as an alternative treatment approach for cervical cancer (CC) in stage Ib2-IIb, who had a strong desire to be treated with surgery. Our study aims to develop a model based on multimodal MRI by using radiomics and deep learning to predict the treatment response in CC patients treated with neoadjuvant chemoradiotherapy (NACRT). From August 2009 to June 2013, CC patients in stage Ib2-IIb (FIGO 2008) who received NACRT at Fujian Cancer Hospital were enrolled in our study. Clinical information, contrast-enhanced T1-weighted imaging (CE-T1WI), and T2-weighted imaging (T2WI) data were respectively collected. Radiomic features and deep abstract features were extracted from the images using radiomics and deep learning models, respectively. Then, ElasticNet and SVM-RFE were employed for feature selection to construct four single-sequence feature sets. Early fusion of two multi-sequence feature sets and one hybrid feature set were performed, followed by classification prediction using four machine learning classifiers. Subsequently, the performance of the models in predicting the response to NACRT was evaluated by separating patients into training and validation sets. Additionally, overall survival (OS) and disease-free survival (DFS) were assessed using Kaplan-Meier survival curves. Among the four machine learning models, SVM exhibited the best predictive performance (AUC=0.86). Among the seven feature sets, the hybrid feature set achieved the highest values for AUC (0.86), ACC (0.75), Recall (0.75), Precision (0.81), and F1-score (0.75) in the validation set, outperforming other feature sets. Furthermore, the predicted outcomes of the model were closely associated with patient OS and DFS (p = 0.0044; p = 0.0039). A model based on MRI images with features from multiple sequences and different methods could precisely predict the response to NACRT in CC patients. This model could assist clinicians in devising personalized treatment plans and predicting patient survival outcomes.

Deep Learning-Based Prediction of Post-treatment Survival in Hepatocellular Carcinoma Patients Using Pre-treatment CT Images and Clinical Data.

The objective of this study was to develop and evaluate a model for predicting post-treatment survival in hepatocellular carcinoma (HCC) patients using their CT images and clinical information, including various treatment information. We collected pre-treatment contrast-enhanced CT images and clinical information including patient-related factors, initial treatment options, and survival status from 692 patients. The patient cohort was divided into a training cohort (n = 507), a testing cohort (n = 146), and an external CT cohort (n = 39), which included patients who underwent CT scans at other institutions. After model training using fivefold cross-validation, model validation was performed on both the testing cohort and the external CT cohort. Our cascaded model employed a 3D convolutional neural network (CNN) to extract features from CT images and derive final survival probabilities. These probabilities were obtained by concatenating previously predicted probabilities for each interval with the patient-related factors and treatment options. We utilized two consecutive fully connected layers for this process, resulting in a number of final outputs corresponding to the number of time intervals, with values representing conditional survival probabilities for each interval. Performance was assessed using the concordance index (C-index), the mean cumulative/dynamic area under the receiver operating characteristics curve (mC/D AUC), and the mean Brier score (mBS), calculated every 3months. Through an ablation study, we found that using DenseNet-121 as the backbone network and setting the prediction interval to 6months optimized the model's performance. The integration of multimodal data resulted in superior predictive capabilities compared to models using only CT images or clinical information (C index 0.824 [95% CI 0.822-0.826], mC/D AUC 0.893 [95% CI 0.891-0.895], and mBS 0.121 [95% CI 0.120-0.123] for internal test cohort; C index 0.750 [95% CI 0.747-0.753], mC/D AUC 0.819 [95% CI 0.816-0.823], and mBS 0.159 [95% CI 0.158-0.161] for external CT cohort, respectively). Our CNN-based discrete-time survival prediction model with CT images and clinical information demonstrated promising results in predicting post-treatment survival of patients with HCC.