Biomedical subjects

T Y Leong

Publications and source records attributed to T Y Leong.

18 recordsLinked to original sources

PubMed2001

Constructing influence views from data to support dynamic decision making in medicine.

A dynamic decision model can facilitate the complicated decision-making process in medicine, in which both time and uncertainty are explicitly considered. In this paper, we address the problem of automatic construction of a dynamic decision model from a large medical database. Within the DynaMoL (a dynamic decision modeling language) framework, a model can be represented in influence view. Thus, our proposed approach first learns the structures of the influence view based on the minimal description length (MDL) principle, and then obtains the conditional probabilities of the model by Bayesian method. The experiment results demonstrate that our system can efficiently construct the influence views from data with high fidelity.

Algorithms↗

PubMed2001

Nonparametric techniques to extract fuzzy rules for breast cancer diagnosis problem.

This paper addresses breast cancer diagnosis problem as a pattern classification problem. Specifically, the problem is studied using Wisconsin-Madison breast cancer data set. Fuzzy rules are generated from the input-output relationship so that the diagnosis becomes easier and transparent for both patients and physicians. For each class, at least one training pattern is chosen as the prototype, provided (a) the maximum membership of the training pattern is in the given class, and (b) among all the training patterns, the neighborhood of this training pattern has the least fuzzy-rough uncertainty in the given class. Using the fuzzy-rough uncertainty, a cluster is constructed around each prototype. Finally, these clusters are interpreted as the fuzzy rules that relate the prognostic factors and the diagnosis results. The advantages of the proposed algorithm are, (a) there is no need to know the structure of the training data, (b) the number of fuzzy rules does not increase with the increase of the number of input dimensions, and (c) small number of fuzzy rules is generated. With the three generated fuzzy rules, 96.20% classification efficiency is achieved, which is comparable to other rule generation techniques.

Algorithms↗

PubMed2001

PDL: a definition language for trend pattern representation and detection in medicine.

This paper proposes a pattern definition language, PDL, to effectively represent and manipulate trend patterns to support medical decision making in time-critical domains. Based on a modified version of SDL, a shape definitional language introduced by Agrawal, PDL extends the expressive power of SDL in the temporal domains. PDL also permits irregular length of elementary patterns to be matched in the query. This paper describes the syntax and the semantics of PDL, as well as illustrating how it can be applied in a time-critical medical domain.

Critical Care↗

PubMed2001

Fuzzy K-means clustering with missing values.

Fuzzy K-means clustering algorithm is a popular approach for exploring the structure of a set of patterns, especially when the clusters are overlapping or fuzzy. However, the fuzzy K-means clustering algorithm cannot be applied when the real-life data contain missing values. In many cases, the number of patterns with missing values is so large that if these patterns are removed, then sufficient number of patterns is not available to characterize the data set. This paper proposes a technique to exploit the information provided by the patterns with the missing values so that the clustering results are enhanced. There are various preprocessing methods to substitute the missing values before clustering the data. However, instead of repairing the data set at the beginning, the repairing can be carried out incrementally in each iteration based on the context. In that case, it is more likely that less uncertainty is added while incorporating the repair work. This scheme is further consolidated in this paper by fine-tuning the missing values using the information from other attributes. The applications of the proposed method in medical domain have produced good performance.

Algorithms↗

PubMed2000

Using linear regression functions to abstract high-frequency data in medicine.

This paper investigates the problem of representing medical time series in linear piece-wise functions and proposes a novel algorithm to transform time-stamped numeric data into simple linear regression functions. We apply methods that involve the hat matrix leverage value and the studentized deleted residual to identify outliers, and a heuristic approach to remove them from the data sets. By distinguishing the breaking points from true outliers, we can efficiently break the data set with respect to the underlying patterns. Using a rough segmentation step, our approach avoids using the whole data set as input, and reduces space requirement. The experimental results indicate our method can achieve more accurate representation of the underlying patterns in data sets collected in the intensive care units efficiently.

Algorithms↗

PubMed2000

Application of K-nearest neighbors algorithm on breast cancer diagnosis problem.

This paper addresses the Breast Cancer diagnosis problem as a pattern classification problem. Specifically, this problem is studied using the Wisconsin-Madison Breast Cancer data set. The K-nearest neighbors algorithm is employed as the classifier. Conceptually and implementation-wise, the K-nearest neighbors algorithm is simpler than the other techniques that have been applied to this problem. In addition, the Knearest neighbors algorithm produces the overall classification result 1.17% better than the best result known for this problem.

Algorithms↗

PubMed2000

A data preprocessing framework for supporting probability-learning in dynamic decision modeling in medicine.

Data preprocessing is needed when real-life clinical databases are used as the data sources to learn the probabilities for dynamic decision models. Data preprocessing is challenging as it involves extensive manual effort and time in developing the data operation scripts. This paper presents a framework to facilitate automated and interactive generation of the problem-specific data preprocessing scripts. The framework has three major components: 1) A model parser that parses the decision model definition, 2) A graphical user interface that facilitates the interaction between the user and the system, and 3) A script generator that automatically generates the specific database scripts for the data preprocessing. We have implemented a prototype system of the framework and evaluated its effectiveness via a case study in the clinical domain. Preliminary results demonstrate the practical promise of the framework.

Artificial Intelligence↗

PubMed2000

The rapid diagnosis of smear-negative pulmonary tuberculosis: a cost-effectiveness analysis.

OBJECTIVE: The prompt diagnosis of smear-negative pulmonary tuberculosis (PTB) is a clinical challenge. It may be achieved by a number of tests which have varying accuracies, costs and degrees of invasiveness. The objective of this study was to compare the cost-effectiveness of clinical judgement (empirical), the Roche Cobas amplicor assay for Mycobacterium tuberculosis (amplicor), acid-fast staining of bronchoalveolar lavage specimens (BAL), nucleic acid amplification tests of bronchoalveloar lavage specimens for M. tuberculosis (BAL + NAA), computed tomography (CT) and amplicor assay followed by BAL. METHODOLOGY: The range of predictive values of the various strategies were derived from published data and a new study of 441 consecutive adult patients with suspected smear-negative PTB prospectively stratified into three pretest risk groups: low, intermediate and high. The cost-effectiveness was evaluated with a decision tree model (DATA software). RESULTS: The incidence of PTB was 5.7% (4% culture positive) for the whole group, 95% in the high-risk group, 0.9% in the low-risk group and 3.4% in the intermediate-risk group. The sensitivity of the empirical approach was 49% and of the amplicor assay was 44%. Patient outcomes were expressed as life expectancy for the base case of a 58-year-old man with a pretest probability of 5.7%. At this low pretest risk the differences in life expectancies between tests was < 0.1 years and the empirical approach incurred the lowest cost. Sensitivity analysis at increasing pretest risks showed better life expectancies (approximately 1 years) for CT scan and test combinations than empirical and amplicor for additional costs of US$243-US$309. Bronchoalveolar lavage had the worst overall cost-effectiveness. CONCLUSIONS: We conclude that the pretest risk of active PTB was a key determinant of test utility; that the AMPLICOR assay was comparable to clinical judgement; that BAL was the least useful test; and that with increasing risks, CT scan and test combinations performed better. Further studies are needed to better define patients with intermediate risk for PTB and to directly compare the cost-effectiveness of more sensitive nucleic acid amplification tests such as the enhanced Gen Probe, CT scan and test combinations/sequences in these patients.

Bacteriological Techniques↗

PubMed1999

Supporting multi-level multi-perspective dynamic decision making in medicine.

Most medical decision problems are exceedingly complex and contain a large number of variables. Abstraction facilitates the process of building a decision model by allowing a model builder to work at a level of detail that he is most comfortable with; it is also useful in time-critical situations or when there is insufficient data to support complete specification of probabilities of the uncertain events. In this paper, we identify and formalize abstraction and refinement operations commonly used in model construction. We illustrate the use of these mechanisms with an example on the follow-up management of colorectal cancer patients after surgery.

Colorectal Neoplasms↗

PubMed1999

Decision analytic approach to severe head injury management.

Severe head injury management in the intensive care unit is extremely challenging due to the complex domain, the uncertain intervention efficacies, and the time-critical setting. We adopt a decision analytic approach to automate the management process. We document our experience in building a simplified influence diagram that involves about 3000 numerical parameters. We identify the inherent problems in structuring a model with unclear domain relationships, numerous interacting variables, and real-time multiple inputs. We analyze the effectiveness and limitations of the decision analytic approach and present a set of desiderata for effective knowledge acquisition in this setting. We also propose a semi-qualitative approach to parameter elicitation.

Craniocerebral Trauma↗

PubMed1999

PROBES: a framework for probability elicitation from experts.

A decision analytic model represents uncertainties as probability distributions. These distributions are hard to assess especially for large and dynamic models. We propose an integrated framework that facilitates elicitation of the relevant probability distributions for dynamic decision models from the domain experts. The experts usually use some judgmental heuristics to aid probability assessments; the resulting distributions may be proned to cognitive biases. Our framework aims to minimize the effects of these biases and to improve the quality of decisions made. We have implemented a prototype system of the framework and evaluated its effectiveness via a case study in the follow-up management of colorectal cancer patients after curative surgery. Preliminary results demonstrate the practical promise of the framework.

Colorectal Neoplasms↗

PubMed1998

Dynamic decision analysis in medicine: a data-driven approach.

Dynamic decision analysis concerns decision problems in which both time and uncertainty are explicitly considered. Two major challenges in dynamic decision analysis are on proper formulation of a model for the problem and effective elicitation of the numerous time-dependent conditional probabilities for the model. Based on a new, general dynamic decision modeling framework called DynaMoL (Dynamic decision Modeling Language), we propose a data-driven approach to addressing these issues. Our approach uses available problem data from large medical databases, guides the decision modeling at a proper level of abstraction and establishes a Bayesian learning method for automatic extraction of the probabilistic parameters. We demonstrate the theoretical implications and practical promises of this new approach to dynamic decision analysis in medicine through a comprehensive case study in the optimal follow-up of patients after curative colorectal cancer surgery.

Bayes Theorem↗

PubMed1998

Induction of diagnostic test strategies with multi-level information measures.

This paper presents a method for inducing clinical diagnostic test protocols or strategies from data. We represent testing strategies as a strategy tree. To support induction of strategy tree, we define three information measures: K-level information, K-level information gain, K-level gain ratio, and K-level cost index, for test selection during strategy building. These measures generalize Quinlan's information measures used in decision tree induction. We present theoretical and experimental results to show that the K-level cost index can be used to induce strategy trees in a practical domain.

Algorithms↗

PubMed1998

Modelling medical decisions in DynaMoL: a new general framework of dynamic decision analysis.

Dynamic decision analysis concerns decision problems in which both time and uncertainty are explicitly considered. We present a new dynamic decision analysis framework, called DynamoL, that supports graphical presentation of the decision factors in multiple perspectives. To alleviate the difficulty in assessing conditional probabilities over time in dynamic decision models, DynaMoL incorporates a Bayesian learning system to automatically learn the probabilistic parameters from large medical databases. We describe the DynaMoL modeling and learning architecture through a medical decision problem on the optimal follow-up schedule for patients after curative colorectal cancer surgery. We also show that the modeling experience and results indicate practical promise for the framework.

Artificial Intelligence↗

PubMed1998

Consistency management in multiple perspective medical decision analysis.

Multiple perspective reasoning is often involved in real-world decision analysis. Different perspectives may be suitable for different stages of the decision modeling process. Multiple perspective modeling calls for consistency management which ensures that the different perspectives reflect the same information. This paper summarizes the graphical perspectives currently supported in DynaMoL, a new framework for dynamic decision analysis. We introduce a new perspective, the tree view, into the framework. We present the main ideas involved in consistency management in the framework. We also discuss the critical issues involved in multiple perspective modeling of a simplified case study in medicine.

Colorectal Neoplasms↗

PubMed1993

Dynamic decision modeling in medicine: a critique of existing formalisms.

Dynamic decision models are frameworks for modeling and solving decision problems that take into explicit account the effects of time. These formalisms are based on structural and semantical extensions of conventional decision models, e.g., decision trees and influence diagrams, with the mathematical definitions of finite-state semi-Markov processes. This paper identifies the common theoretical basis of existing dynamic decision modeling formalisms, and compares and contrasts their applicability and efficiency. It also argues that a subclass of such dynamic decision problems can be formulated and solved more effectively with non-graphical techniques. Some insights gained from this exercise on automating the dynamic decision making process are summarized.

Decision Support Techniques↗

PubMed1991

Representation requirements for supporting knowledge-based construction of decision models in medicine.

This paper analyzes the medical knowledge required for formulating decision models in the domain of pulmonary infectious diseases (PIDs) with acquired immunodeficiency syndrome (AIDS). Aiming to support dynamic decision-modeling, the knowledge characterization focuses on the ontology of the clinical decision problem. Relevant inference patterns and knowledge types are identified.

Acquired Immunodeficiency Syndrome↗

PubMed1979

The role of chloroplast-membrane-protein synthesis in the circadian clock. Purification and partial characterization of a polypeptide which is suggested to be involved in the clock.

A polypeptide (polypeptide P39), which is presumed to involved in the photosynthetic circadian rhythm in the green alga Acetabularia, was purified from the EDTA-insoluble chloroplast membrane fraction by means of preparative dodecylsulfate gel electrophoresis and then partially characterized. The purity of the isolated polypeptide P39 was confirmed by a further electrophoresis on an analytical dodecylsulfate gel and further elucidated by amino-terminal analysis which shows that glycine is the only amino-terminal amino acid of the purified polypeptide material. The molecular weight of the polypeptide P39 was found to be about 39,000 on analytical gel electrophoresis and the value was further supported by those obtained from amino acid composition and peptide mapping. The amino acid composition of polypeptide P39 showed that the proportion of intermediate amino acid groups is high while the proportion of hydrophilic amino acid groups is well balanced by that of hydrophobic amino acid groups, a property characteristic of membrane proteins.

Acetabularia↗