Early detection is critical for effective management of Alzheimer's disease (AD) and screening for mild cognitive impairment (MCI) is common practice. Among several deeplearning techniques that have been applied to assessing structural brain changes on magnetic resonance imaging (MRI), convolutional neural network (CNN) has gained popularity due to its superb efficiency in automated feature learning with the use of a variety of multilayer perceptrons. Meanwhile, ensemble learning (EL) has shown to be beneficial in the robustness of learning-system performance via integrating multiple models. Here, we proposed a classifier ensemble developed by combining CNN and EL, i.e., the CNN-EL approach, to identify subjects with MCI or AD using MRI: i.e., classification between (1) AD and healthy cognition (HC), (2) MCIc (MCI patients who will convert to AD) and HC, and (3) MCIc and MCInc (MCI patients who will not convert to AD). For each binary classification task, a large number of CNN models were trained applying a set of sagittal, coronal, or transverse MRI slices; these CNN models were then integrated into a single ensemble. Performance of the ensemble was evaluated using stratified fivefold cross-validation method for 10 times. The number of the intersection points determined by the most discriminable slices separating two classes in a binary classification task among the sagittal, coronal, and transverse slice sets, transformed into the standard Montreal Neurological Institute (MNI) space, acted as an indicator to assess the ability of a brain region in which the points were located to classify AD. Thus, the brain regions with most intersection points were considered as those mostly contributing to the early diagnosis of AD. The result revealed an accuracy rate of 0.84 ± 0.05, 0.79 ± 0.04, and 0.62 ± 0.06, respectively, for classifying AD vs. HC, MCIc vs. HC, and MCIc vs. MCInc, comparable to previous reports and a 3D deep learning approach (3D-SENet) based on a more state-of-the-art and popular Squeeze-and-Excitation Networks model using channel attention mechanism. Notably, the intersection points accurately located the medial temporal lobe and several other structures of the

Background.In older adults admitted to intensive care units (ICUs), frailty influences prognosis. We examined the relationship between the frailty index (FI) based on deficit accumulation and early and late survival.Methods.Older patients (≥65 years) admitted to a specialized geriatric ICU at the Liuhuaqiao Hospital, Guangzhou, China between July–December 2011 (n = 155; age 82.7±7.1 y; 87.1% men) were followed for 300 days. The FI was calculated as the proportion present of 52 health deficits. FI performance was compared with that of several prognostic scores.Results.The 90-day death rate was 38.7% (n = 60; 27 died within 30 days). The FI score was correlated with the Glasgow Coma Scale, Karnofsky Scale, Palliative Performance Scale, Acute Physiology Score—APACHE II and APACHE IV (r 2 = 0.52 to 0.72, p < 0.001). Patients who died within 30 days had higher mean FI scores (0.41±0.11) than those who survived to 300 days (0.22±0.11; F = 38.91, p < 0.001). Each 1% increase in the FI from the previous level was associated with an 11% increase in the 30-day mortality risk (95% CI: 7%–15%) adjusting for age, sex, and the prognostic scores. The FI discriminated patients who died in 30 days from those who survived with moderately high accuracy (AUC = 0.89±0.03). No one with an FI score >0.46 survived past 90 days.Conclusion.ICU survival was strongly associated with the level of frailty at admission. An FI based on health deficit accumulation may help improve critical care outcome prediction in older adults.

No abstract

T he artificial intelligence community began advocating knowledge engineering early in the 1980s, and developers have since built thousands of knowledge-based systems in diverse domains. However, knowledge acquisition-the critical first stage of the knowledge engineering process-remains a bottleneck in KBS construction.Knowledge acquisition methods are usually manual or semiautomatic, limited with regard to speed and accuracy, and expensive. As rapid information technology development increases our ability to generate and store data, it's important to develop efficient methods of automatically extracting knowledge from it with little or no reliance on a knowledge engineer or domain expert. Furthermore, automatic methods can assist experts by clarifying different factors involved in making a decision. 1 This is especially useful in large problem domains, where experts can find it difficult to articulate processes, even when they can readily provide examples and solutions that inductive systems can use.Zdzislaw Pawlak first advocated the rough set theory as an approach to automatic knowledge acquisition in 1982. 2 RS offers a way to find attribute dependencies in database-like information systems, such as a decision table. 3 The basic idea is to derive decision or classification rules through data attribute and attribute-value reductions. RS constrains the data reductions by keeping the discernibility relations among data objects in the table unchanged. However, most RS-based reduction algorithms derive heuristic knowledge only from dependencies between condition and decision attributes. Thus, the obtained rules tend to incorporate noise and other peculiarities in the decision table.We've developed a novel approach to knowledge acquisition based on RS and principal component analysis. A PCA-based quantitative index measures the relative importance of different condition attributes among the state space constructed by all condition attributes. The index strengthens the attribute and attribute-value reductions while maintaining the decision table's discernibility relations. Our KA-RSPCA algorithm outperformed four other RS algorithms on two test data sets. Data reductionIn real life, condition attributes are not equally important to the classification knowledge that a decision table contains. Some attributes are even redundant and so mislead the decision-making process as well as waste storage and processing resources. An attribute reduct defines an attribute set for generating an efficient rule set. 4 It helps the decision maker concentrate on the most essential factors by reducing the decision table. The quality and number of attributes in an attribute reduct influence the performance and complexity of decision rules.A decision table can often have more than one attribute reduct without changing the original dependency relation between condition attributes and decision attributes. The problem of obtaining all attribute reducts has proved unsolvable, so the focus instead is on how to determine the best reduct and ho...