Survival Risk Prediction of Esophageal Cancer Based on Self-Organizing Maps Clustering and Support Vector Machine Ensembles

ObjectiveTo compare the performance of a deep learning survival network with the tumor, node, and metastasis (TNM) staging system in survival prediction and test the reliability of individual treatment recommendations provided by the network.MethodsIn this population-based cohort study, we developed and validated a deep learning survival model using consecutive cases of newly diagnosed stage I to IV esophageal cancer between January 2004 and December 2015 in a Surveillance, Epidemiology, and End Results (SEER) database. The model was externally validated in an independent cohort from Fujian Provincial Hospital. The C statistic was used to compare the performance of the deep learning survival model and TNM staging system. Two other deep learning risk prediction models were trained for treatment recommendations. A Kaplan–Meier survival curve was used to compare survival between the population that followed the recommended therapy and those who did not.ResultsA total of 9069 patients were included in this study. The deep learning network showed more promising results in predicting esophageal cancer-specific survival than the TNM stage in the internal test dataset (C-index=0.753 vs. 0.638) and external validation dataset (C-index=0.687 vs. 0.643). The population who received the recommended treatments had superior survival compared to those who did not, based on the internal test dataset (hazard ratio, 0.753; 95% CI, 0.556-0.987; P=0.042) and the external validation dataset (hazard ratio, 0.633; 95% CI, 0.459-0.834; P=0.0003).ConclusionDeep learning neural networks have potential advantages over traditional linear models in prognostic assessment and treatment recommendations. This novel analytical approach may provide reliable information on individual survival and treatment recommendations for patients with esophageal cancer.

show abstract

“…Sun et al. developed a survival risk prediction model for EC based on nine blood indices ( 27 ). Lin et al.…”

Section: Discussionmentioning

confidence: 99%

Development and validation of a deep learning model to predict survival of patients with esophageal cancer

et al. 2022

View full text Add to dashboard Cite

show abstract

“…e method of principal component analysis (PCA) to reducing linear dimensions might be helpful in data analysis and compression [37]. Using this method, which involves finding orthogonal linear combinations of the attributes of the initial data set, it is possible to combine qualities that are not connected with one another.…”

Section: Methodsmentioning

confidence: 99%

Machine Learning-Based Multimodel Computing for Medical Imaging for Classification and Detection of Alzheimer Disease

Alghamedy

Shafiq

Liu

et al. 2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

Alzheimer is a disease that causes the brain to deteriorate over time. It starts off mild, but over the course of time, it becomes increasingly more severe. Alzheimer’s disease causes damage to brain cells as well as the death of those cells. Memory in humans is especially susceptible to this. Memory loss is the first indication of Alzheimer’s disease, but as the disease progresses and more brain cells die, additional symptoms arise. Medical image processing entails developing a visual portrayal of the inside of a body using a range of imaging technologies in order to discover and cure problems. This paper presents machine learning-based multimodel computing for medical imaging for classification and detection of Alzheimer disease. Images are acquired first. MRI images contain noise and contrast problem. Images are preprocessed using CLAHE algorithm. It improves image quality. CLAHE is better to other methods in its capacity to enhance the look of mammography in minute places. A white background makes the lesions more obvious to the naked eye. In spite of the fact that this method makes it simpler to differentiate between signal and noise, the images still include a significant amount of graininess. Images are segmented using the k-means algorithm. This results in the segmentation of images and identification of region of interest. Useful features are extracted using PCA algorithm. Finally, images are classified using machine learning algorithms.

show abstract

“…PSO algorithms can also determine and identify the optimum solutions by sharing information with individual particles groups. In PSO analysis, a group of n particles within a swarm S (Equation (3)) is used, while each particle of S i is represented by a vector [ 36 ]: S = {S 1 , S 2 , … S n }, S is a vector represented by = { x i , v i , p i } where x i represents the current position, v i represents the current velocity, and p i shows the well-known best position within the swarm. After the position and velocity for each particle were identified, the current position and recorded position were evaluated and analysed with the performance score.…”

Section: Methodsmentioning

confidence: 99%

Production of Aluminum AA6061 Hybrid Nanocomposite from Waste Metal Using Hot Extrusion Process: Strength Performance and Prediction by RSM and Random Forest

et al. 2021

View full text Add to dashboard Cite

To date, various studies have analysed the effects of reinforced ceramic on the properties of AA6061 recycled aluminum alloy chips, such as the tensile strength and fractography. However, a comprehensive analysis of the properties of hybrid composite with the addition of nano-silica oxide and nano-copper oxide reinforcements is still very limited. Therefore, this study aimed to optimise the factors comprising the preheating temperature (PHT), preheating time (PHti), and volume fraction (VF) of reinforcements then determine their impacts on the physical and mechanical properties of the recycled solid-state extruded composite aluminum chips. A total of 45 specimens were fabricated through the hot extrusion technique. The response surface methodology (RSM) was employed to study the optimisation at a PHT range of 450–550 °C with PHti of 1–3 h and VF of 1–3 vol% for both reinforcements (SiO2 and CuO). Moreover, a random forest (RF) model was developed to optimize the model based on a metaheuristic method to improve the model performance. Based on the experimental results the RF model achieve better results than response surface methodology (RSM). The functional quadratic regression is curvature and the tested variable shows stable close data of the mean 0 and α2. Based on the Pareto analysis, the PHT and VF were key variables that significantly affected the UTS, microhardness, and density of the product. The maximum properties were achieved at an optimum PHT, PHti, and VF of 541 °C, 2.25 h, 1 vol% SiO2 and 2.13 vol% CuO, respectively. Furthermore, the morphological results of the tensile fractured surface revealed the homogenous distribution of nano-reinforced CuO and SiO2 particles in the specimens’ structure.

show abstract

Survival Risk Prediction of Esophageal Cancer Based on Self-Organizing Maps Clustering and Support Vector Machine Ensembles

Cited by 28 publications

References 48 publications

Development and validation of a deep learning model to predict survival of patients with esophageal cancer

Development and validation of a deep learning model to predict survival of patients with esophageal cancer

Machine Learning-Based Multimodel Computing for Medical Imaging for Classification and Detection of Alzheimer Disease

Production of Aluminum AA6061 Hybrid Nanocomposite from Waste Metal Using Hot Extrusion Process: Strength Performance and Prediction by RSM and Random Forest

Contact Info

Product

Resources

About