Bayesian inference of hemodynamic changes in functional arterial spin labeling data.

The study of brain function using MRI relies on acquisition techniques that are sensitive to different aspects of the hemodynamic response contiguous to areas of neuronal activity. For this purpose different contrasts such as arterial spin labeling (ASL) and blood oxygenation level dependent (BOLD) functional MRI techniques have been developed to investigate cerebral blood flow (CBF) and blood oxygenation, respectively. Analysis of such data typically proceeds by separate, linear modeling of the appropriate CBF or BOLD time courses. In this work an approach is developed that provides simultaneous inference on hemodynamic changes via a nonlinear physiological model of ASL data acquired at multiple echo times. Importantly, this includes a significant contribution by changes in the static magnetization, M, to the ASL signal. Inference is carried out in a Bayesian framework. This is able to extract, from dual-echo ASL data, probabilistic estimates of percentage changes of CBF, R(2) (*), and the static magnetization, M. This approach provides increased sensitivity in inferring CBF changes and reduced contamination in inferring BOLD changes when compared with general linear model approaches on single-echo ASL data. We also consider how the static magnetization, M, might be related to changes in CBV by assuming the same mechanism for water exchange as in vascular space occupancy.     

Correspondence analysis as a screening method for indicants for clinical diagnosis

In clinical diagnosis, a patient's symptoms are observed and the probabilities of various diseases are assessed. A widely used method of formalizing this approach is independent Bayes in which symptoms are assumed to be independent conditional on the disease category. Correspondence analysis provides a method for examining the dependence between symptoms and assists in the selection of a reduced set of symptoms for the application of the independent Bayes method. This approach is illustrated on two data sets concerned with patients attending Accident and Emergency departments with chest pain and acute abdominal pain, respectively.     

Bayesian estimation of a random effects heteroscedastic probit model

Summary Bayesian analysis is given of a random effects binary probit model that allows for heteroscedasticity. Real and simulated examples illustrate the approach and show that ignoring heteroscedasticity when it exists may lead to biased estimates and poor prediction. The computation is carried out by an efficient Markov chain Monte Carlo sampling scheme that generates the parameters in blocks. We use the Bayes factor, cross‐validation of the predictive density, the deviance information criterion and Receiver Operating Characteristic (ROC) curves for model comparison.     

Development of an open case-based decision-support system for diagnosis in oral pathology

BACKGROUND: Making diagnoses in oral pathology are often difficult and confusing in dental practice, especially for the less-experienced dental student. One of the most promising areas in bioinformatics is computer-aided diagnosis, where a computer system is capable of imitating human reasoning ability and provides diagnoses with an accuracy approaching that of expert professionals. This type of system could be an alternative tool for assisting dental students to overcome the difficulties of the oral pathology learning process. This could allow students to define variables and information, important to improving the decision-making performance. However, no current open data management system has been integrated with an artificial intelligence system in a user-friendly environment. Such a system could also be used as an education tool to help students perform diagnoses. The aim of the present study was to develop and test an open case-based decision-support system. METHODS: An open decision-support system based on Bayes' theorem connected to a relational database was developed using the C++ programming language. The software was tested in the computerisation of a surgical pathology service and in simulating the diagnosis of 43 known cases of oral bone disease. The simulation was performed after the system was initially filled with data from 401 cases of oral bone disease. RESULTS: The system allowed the authors to construct and to manage a pathology database, and to simulate diagnoses using the variables from the database. CONCLUSION: Combining a relational database and an open decision-support system in the same user-friendly environment proved effective in simulating diagnoses based on information from an updated database.     

An epidemiologic approach to computerized medical diagnosis—AEDMI program

A program called “An Epidemiological Approach to Computerized Medical Diagnosis” (AEDMI) is presented. Using an interactive questionnaire, physician-patient interviews are conducted and a summary of the relevant clinical data is provided. Standard items, obained on a multi-centre basis, form a large-scale data base. Simultaneously, the reasoning of clinical experts in each real case is analyzed to obtain a knowledge-rules data base. The methodology of the program combines Bayesian systems, expert systems, and other new lines of researcg such as neural networks or case-based reasoning. The general concepts of clinical decision making aid systems are reviewed. This publication is aimed at obtaining international cooperation.     

Latent variable discovery in classification models

The naive Bayes model makes the often unrealistic assumption that the feature variables are mutually independent given the class variable. We interpret a violation of this assumption as an indication of the presence of latent variables, and we show how latent variables can be detected. Latent variable discovery is interesting, especially for medical applications, because it can lead to a better understanding of application domains. It can also improve classification accuracy and boost user confidence in classification models.     

基于数据挖掘的低增生性骨髓增生异常综合征与再生障碍性贫血分类模型研究

目的 构建低增生性骨髓增生异常综合征（hypo-MDS）与再生障碍性贫血（AA）鉴别诊断的决策树、贝叶斯、卷积神经网络、改进的支持向量机四种模型并选择出最优模型。方法 收集2010—2019年华北理工大学附属医院的AA与hypo-MDS患者的病例资料,使用统计学方法筛选指标,将处理后的样本以4[DK]∶1随机分为训练集和测试集,构建决策树、贝叶斯、卷积神经网络、改进的支持向量机四种模型,采用五折交叉验证法多次重复验证,通过灵敏度、AUC等指标评价鉴别诊断效果。结果 hypo-MDS患者红细胞、血红蛋白含量等指标低于AA患者,成熟单核细胞比例等指标高于AA患者,年龄和职业分布也存在差异(P＜0.05）;最终选出21个特异性指标。四种模型的分类效果比较:灵敏度分别为82.56%、65.12%、87.21%、79.07%;AUC分别为0.81、0.68、0.82、0.83;准确率分别为75.32%、69.48%、77.27%、74.03%。对卷积神经网络的误判病例分析得出年龄、血成熟淋巴细胞等7个指标均存在差异（P＜0.05）。结论 在决策树、贝叶斯、卷积神经网络、改进的支持向量机四种诊断模型中,卷积神经网络具有最佳分类效果。     

Combining clinical judgement and statistical data in expert systems: Over-the-telephone management decisions for critical congenital heart disease in the first month of life

When a new-born baby with congenital heart disease is referred to a regional specialist centre, the transportation management is crucial but must be decided on the basis of clinical information obtained over the telephone. We consider algorithmic and naive statistical approaches to helping in this decision, and on the basis of preliminary results the relative strengths and weaknesses are discussed. A synthesised logical and probabilistic approach appears to have the best potential and could be implemented on hand-held computers.Paper presented at symposium on Computer-assisted Decision Support and Database Management in Anaesthesia, Intensive Care and Cardio-pulmonary Medicine, Rotterdam, 1988.     

基于控制周期特征式的线性混合自动化验证

提出了一种基于归纳法思想的验证方法,通过控制周期上特性的描述,发现了基于控制周期特征式的线性混合自动机验证方法,这一方法采用定理证明过程来得出归纳证明的结构;采用模型检查方法来得出归纳证明的奠基和迭代步,这一方法同时兼顾了模型检查和定理证明的特点用此定理证明更高的自动化程度解决了单用模型检查不能解决的问题,得出了对著名案例ＧａｓＢｕｒｎｅｒ问题中的参数３ｎｏｎ＿ｌｅａｋｋｉｎｇ≥７６的最优范围。