Qiuchong Chen,1,* Yixue Zhang,1,* Mengjun Zhang,1 Ziying Li,1 Jindong Liu1,2 1Department of Anesthesiology, The Affiliated Hospital of Xuzhou Medical University, Xuzhou, Jiangsu, People's Republic of China; 2Jiangsu Province Key Laboratory of Anesthesiology, Xuzhou Medical University, Xuzhou, Jiangsu, People's Republic of China *These authors contributed equally to this work Correspondence: Jindong Liu, Department of Anesthesiology, The Affiliated Hospital of Xuzhou Medical University, 99 Huaihai Road West, Quanshan District, Xuzhou, Jiangsu, 221000, People's Republic of China, Email [email protected] Objective: There has been a worldwide increment in acute kidney injury (AKI) incidence among elderly orthopedic operative patients. The AKI prediction model provides patients' early detection a possibility at risk of AKI; most of the AKI prediction models derive, however, from the cardiothoracic operation. The purpose of this study is to predict the risk of AKI in elderly patients after orthopedic surgery based on machine learning algorithm models. Methods: We organized a retrospective study being comprised of 1000 patients with postoperative AKI undergoing orthopedic surgery from September 2016, to June, 2021. They were divided into training (80%;n 799) and test (20%;n 201) sets.We utilized nine machine learning (ML) algorithms and used intraoperative information and preoperative clinical features to acquire models to predict AKI. The performance of the model was evaluated according to the area under the receiver operating characteristic (AUC), sensitivity, specificity and accuracy. Select the optimal model and establish the nomogram to make the prediction model visualization. The concordance statistic (C-statistic) and calibration curve were used to discriminate and calibrate the nomogram respectively. Results: In predicting AKI, nine ML algorithms posted AUC of 0.656– 1.000 in the training cohort, with the randomforest standing out and AUC of 0.674– 0.821 in the test cohort, with the logistic regression model standing out.

This is part numero tres of the Explaining Machine Learning Algorithms to a 10-Year Old series. If you read the two previous ones about XGBoost Regression and K-Means Clustering, then you know the drill. We have a scary-sounding algorithm, so let's strip it of its scary bits and understand the simple intuition behind it. In the same vein as K-Means Clustering, today we are going to talk about another popular clustering algorithm -- Hierarchical Clustering. Let's say a clothing store has collected the ages of 9 of its customers, labeled C1-C9, and the amount each of them spent at the store in the last month.

Originally published on Towards AI the World's Leading AI and Technology News and Media Company. If you are building an AI-related product or service, we invite you to consider becoming an AI sponsor. At Towards AI, we help scale AI and technology startups. Let us help you unleash your technology to the masses. Time series analysis is a way of analyzing the data which is sequenced in a data-time format.

For a given predictor, the t-statistic (and its associated p-value) tests whether or not there is a statistically significant relationship between a given predictor and the outcome variable, that is whether or not the beta coefficient of the predictor is significantly different from zero. Null hypothesis (H0): the coefficients are equal to zero (i.e., no relationship between x and y) Alternative Hypothesis (Ha): the coefficients are not equal to zero (i.e., there is some relationship between x and y) Another aspect to pay attention to your linear models is the p-value of the coefficients. A p-value indicates whether or not you can reject or accept a hypothesis. A very small p value means that the predictor is probably an excellent addition to your model. A standard way to test if the predictors are not meaningful is looking if the p-values smaller than 0.05.

Which machine learning algorithm should I use? The ML Algorithm cheat sheet helps you choose the best machine learning algorithm for your predictive analytics solution. Your decision is driven by both the nature of your data and the goal you want to achieve with your data. What do you want to do with your data? I must state here that we need to have a solid understanding of the iterative system of methods that guide Data Scientists on the ideal approach to solving problems using the Data Science Methodology.

Measurement of uncertainty of predictions from machine learning methods is important across scientific domains and applications. We present, to our knowledge, the first such technique that quantifies the uncertainty of predictions from a classifier and accounts for both the classifier's belief and performance. We prove that our method provides an accurate estimate of the probability that the outputs of two neural networks are correct by showing an expected calibration error of less than 0.2% on a binary classifier, and less than 3% on a semantic segmentation network with extreme class imbalance. We empirically show that the uncertainty returned by our method is an accurate measurement of the probability that the classifier's prediction is correct and, therefore has broad utility in uncertainty propagation.

This paper investigates probability density functions (PDFs) that are continuous everywhere, nearly uniform around the mode of distribution, and adaptable to a variety of distribution shapes ranging from bell-shaped to rectangular. From the viewpoint of computational tractability, the PDF based on the Fermi-Dirac or logistic function is advantageous in estimating its shape parameters. The most appropriate PDF for $n$-variate distribution is of the form: $p\left(\mathbf{x}\right)\propto\left[\cosh\left(\left[\left(\mathbf{x}-\mathbf{m}\right)^{\mathsf{T}}\boldsymbol{\Sigma}^{-1}\left(\mathbf{x}-\mathbf{m}\right)\right]^{n/2}\right)+\cosh\left(r^{n}\right)\right]^{-1}$ where $\mathbf{x},\mathbf{m}\in\mathbb{R}^{n}$, $\boldsymbol{\Sigma}$ is an $n\times n$ positive definite matrix, and $r>0$ is a shape parameter. The flat-topped PDFs can be used as a component of mixture models in machine learning to improve goodness of fit and make a model as simple as possible.

Adversarial robustness is one of the essential safety criteria for guaranteeing the reliability of machine learning models. While various adversarial robustness testing approaches were introduced in the last decade, we note that most of them are incompatible with non-differentiable models such as tree ensembles. Since tree ensembles are widely used in industry, this reveals a crucial gap between adversarial robustness research and practical applications. This paper proposes a novel whitebox adversarial robustness testing approach for tree ensemble models. Concretely, the proposed approach smooths the tree ensembles through temperature controlled sigmoid functions, which enables gradient descent-based adversarial attacks. By leveraging sampling and the log-derivative trick, the proposed approach can scale up to testing tasks that were previously unmanageable. We compare the approach against both random perturbations and blackbox approaches on multiple public datasets (and corresponding models). Our results show that the proposed method can 1) successfully reveal the adversarial vulnerability of tree ensemble models without causing computational pressure for testing and 2) flexibly balance the search performance and time complexity to meet various testing criteria.

Graph convolutional network (GCN) is an emerging neural network approach. It learns new representation of a node by aggregating feature vectors of all neighbors in the aggregation process without considering whether the neighbors or features are useful or not. Recent methods have improved solutions by sampling a fixed size set of neighbors, or assigning different weights to different neighbors in the aggregation process, but features within a feature vector are still treated equally in the aggregation process. In this paper, we introduce a new convolution operation on regular size feature maps constructed from features of a fixed node bandwidth via sampling to get the first-level node representation, which is then passed to a standard GCN to learn the second-level node representation. Experiments show that our method outperforms competing methods in semi-supervised node classification tasks. Furthermore, our method opens new doors for exploring new GCN architectures, particularly deeper GCN models.