Lucas, P.J.F. (1998). Computer-based Decision Support in the Management of Primary Gastric non-Hodgkin Lymphoma, Methods of Information in Medicine, 37, pp. 206-219.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... 1980s by Pearl [39] Since then, an increasing num ber of successful applications of such networks in di#erent problem domains have been developed, which demonstrates that they have established their position in Artificial Intelligence as valuable representations of reasoning with uncertainty [1,25,28,12,31] . A belief network consists of a qualitative and a quantitative part. The qualitative part is a directed graph, where the nodes represent the domain s variables (in a medical diagnostic application the variables could for example be a laryngitis and its symptoms such as a sore throat and fever) ....

P.J.F. Lucas, H. Boot, and B.G. Taal. Computer-based decision-support in the management of primary gastric non-hodgkin lymphoma. Methods of Information in Medicine, 37:206 -- 219, 1998.

....that is able to assist clinicians in the clinical management of ventilator associated pneumonia. A number of other decision support systems have been developed in the past, also using the framework of probabilistic networks (e.g. 2,22,33] sometimes in combination with decision theory (e.g. [18,19,28]) The probabilisticnetwork formalism is one of few formalisms for knowledge representation in which qualitative and quantitative medical knowledge can be easily integrated. It is also a formalism that matches decision theory. Hence, the formalism is eminently suitable for building ....

Lucas PJF, Boot H, Taal BG. Computer-based decision-support in the management of primary gastric non-Hodgkin lymphoma. Methods Inf Med 1998;37:206 -- 19.

.... late 1980s by Pearl [1] Since then, an increasing number of successful applications of such networks in different problem domains have been developed, which demonstrates that they have established their position in Artificial Intelligence as valuable representations of reasoning with uncertainty [2 6]. A belief network consists of a qualitative and a quantitative part. The qualitative part is a directed graph, where the nodes represent the domain s variables (in a medical diagnostic application the variables could for example be a laryngitis and its symptoms such as a sore throat and fever) ....

Lucas, P.J.F., Boot, H., and Taal, B.G., Computer-based decision-support in the management of primary gastric non-Hodgkin lymphoma, Methods of Information in Medicine 37, 206-219, 1998.

....that is able to assist clinicians in the clinical management of ventilator associated pneumonia. A number of other decision support systems have been developed in the past, also using the framework of probabilistic networks (e.g. 2, 22, 33] sometimes in combination with decision theory (e.g. [18, 19, 28]) The probabilistic network formalism is one of few formalisms for knowledge representation in which qualitative and quantitative medical knowledge can be easily integrated. It is also a formalism that matches decision theory. Hence, the formalism is eminently suitable for building ....

Lucas PJF, Boot H, Taal BG. Computer-based decision-support in the management of primary gastric non-Hodgkin lymphoma. Methods Inf Med 1998; 37: 206-219.

....satisfy various conditions. To allow for automated construction of a meaningful probabilistic network, the data 2 must have been collected very carefully. Biases that are introduced in the data as a result of the data collection strategies used will usually have an e#ect on the resulting network [Lucas et al. 1998]. This e#ect may not be desirable, however, for the purpose for which the network is being developed. Unfortunately, selection biases are not easily detected in a network, once it has been constructed. Also, the variables and associated values that are recorded in the data collection should match ....

P.J.F. Lucas, H. Boot, and B.G. Taal. Computer-based decisionsupport in the management of primary gastric non-Hodgkin lymphoma. Methods of Information in Medicine, vol. 37, 1998, pp. 206 -- 219.

No context found.

Lucas, P.J.F. (1998). Computer-based Decision Support in the Management of Primary Gastric non-Hodgkin Lymphoma, Methods of Information in Medicine, 37, pp. 206-219.

No context found.

P.J.F. Lucas, H. Boot, B.G. Taal. Computer-based decision-support in the management of primary gastric non-Hodgkin lymphoma. Methods of Information in Medicine, 1998; 37: 206-219.

....discussed here is still in prototype stage; further development needs to take place in order to introduce it in actual clinical practice. The gastric NHL Bayesian belief network only incorporates variables that are widely used by clinicians in choosing the appropriate therapy for patients [3]. The relevance of most of these variables is supported by literature on prognostic factors in gastric NHL. First, the information used in the clinical management of primary gastric NHL was subdivided in pretreatment information, i.e. information that is required for treatment selection, ....

P.J.F. Lucas, H. Boot, B.G. Taal. Computer-based decision-support in the management of primary gastric non-Hodgkin lymphoma. Methods of Information in Medicine, 1998; 37: 206-219.

.... networks by the principal investigator, will be extended [15, 17, 34, 28, 31] When used to predict the likelihood of future events, we call these models prognostic [6, 44] Within the research team there is significant experience in building Bayesian networks for the medical domain (e.g. [36, 42, 41, 51, 2, 4]) Experience built up in the ICEA project (see Section 6.6) is particularly useful in dealing with the problem domain of infectious disease management in the ICU. 6.4.3 Learning temporal Bayesian models Learning a Bayesian network can be separated into two tasks, structure learning, ....

P.J.F. Lucas, H. Boot and B.G. Taal. Computer-based decision-support in the management of primary gastric non-Hodgkin lymphoma, Methods of Information in Medicine 1998c; 37: 206--219.

....a mismatch between common task speci c computer based models and the complexity of the eld of medicine. In contrast to task speci c models, Bayesian networks that re ect the nature of a problem domain have the virtue of being declarative models; they can, therefore, be reused for di erent tasks [13]. They can also be employed to look at particular problems from di erent angles, just by varying the supplied evidence and the questions posed to the model. However, developing such models is challenging, both in terms of the amount of probabilistic information required and modelling e ort. When ....

P.J.F. Lucas, H. Boot, B.G. Taal. Computer-based decision-support in the management of primary gastric non-Hodgkin lymphoma. Meth Inf Med 37 (1998) 206-219.

....as the rst, most probable one. In these 16 patients, the average probability of colonisation with that speci c organism was 48 and the average probability of pneumonia with that speci c organism was 18 . R N Avgc SDc Avgp SDp 1 16 48 8.02 18 12.53 2 6 37 4.02 18 5.61 3 17 24 1.88 12 2. 14 4 8 17 4.27 6 4.78 5 8 16 4.50 6 4.83 6 3 19 0.58 10 0.58 7 5 10 2.19 4 2.19 Table 2: R: rank of the causative organism according to PTA in comparison to sputum culture (sputum culture unknown to PTA) N : number of times the rank occurs; Avg c : average of probabilities of colonisation for speci c ....

....probable one. In these 16 patients, the average probability of colonisation with that speci c organism was 48 and the average probability of pneumonia with that speci c organism was 18 . R N Avgc SDc Avgp SDp 1 16 48 8.02 18 12.53 2 6 37 4.02 18 5.61 3 17 24 1.88 12 2.14 4 8 17 4.27 6 4. 78 5 8 16 4.50 6 4.83 6 3 19 0.58 10 0.58 7 5 10 2.19 4 2.19 Table 2: R: rank of the causative organism according to PTA in comparison to sputum culture (sputum culture unknown to PTA) N : number of times the rank occurs; Avg c : average of probabilities of colonisation for speci c rank; SD c : ....

P.J.F. Lucas, H. Boot and B.G. Taal, Computerbased decision-support in the management of primary gastric non-Hodgkin lymphoma, Methods of Information in Medicine 37 (1998) 206-219.

....as a running example [5] 2 Modelling Developing a model of a realistic medical problem is usually far from easy, and using Bayesian networks yields no exception in this respect. As is the case with other representation formalisms, there are particular guidelines which facilitate developing a BN [4]. We start by summarising some facts concerning the running example of this paper. colonisation pneumonia colonisation PA colonisation HI colonisation SP pneumonia PA pneumonia HI pneumonia SP pneumonia symptoms signs lab hospitalisation aspiration mechanical ventilation ....

P.J.F. Lucas, H. Boot and B.G. Taal, Computer-based decision-support in the management of primary gastric non-Hodgkin lymphoma, Methods of Information in Medicine 37 (1998) 206--219.

.... model of gastric NHL can indeed be used in the context of di erent tasks, such as prediction of prognosis, treatment selection by comparing the likely outcomes of alternative treatment choices, possibly using preference information expressed as utilities and generation of patient pro les [8]. Two independent form Bayesian networks were derived from the network shown in Figure 2. In one of those, the variable early result was taken as a class variable; the other includes 5 year result as class variable. The networks are depicted in Figure 3. Their probability distributions were also ....

P.J.F. Lucas, H. Boot, B.G. Taal. Computer-based decision-support in the management of primary gastric non-Hodgkin lymphoma. Methods of Information in Medicine 1998; 37: 206-219.

.... lymphoma of the stomach can indeed be used in the context of di erent tasks, such as prediction of prognosis, treatment selection by comparing the likely outcomes of alternative treatment choices, possibly using preference information expressed as utilities and generation of patient pro les [7]. The independent form Bayesian network corresponding to the network shown in Figure 2 is depicted in Figure 3. Note that this model includes a single class variable: the posttreatment variable 5 year result; in addition, all pretreatment and treatment variables are incorporated. The remainder of ....

P.J.F. Lucas, H. Boot, B.G. Taal. Computer-based decisionsupport in the management of primary gastric non-Hodgkin lymphoma. Methods of Information in Medicine 1998; 37: 206-219.

....and retrieve patient data to the clinician is not enough. Clinical information systems should also o#er facilities to assist clinicians in dealing with hard clinical problems, such as facilities for decision support. In this paper, we describe the development of a decision theoretic expert system [4], i.e. a system based on a combination of the theory of probabilistic networks (Bayesian belief networks) and decision theory [6, 7] called PTA (Pneumonia Therapy Advisor) that is capable of providing advice about the administration of appropriate antibiotic therapy to patients with pneumonia. ....

P.J.F. Lucas, H. Boot and B.G. Taal, Computer-based decision-support in the management of primary gastric non-Hodgkin lymphoma, Methods of Information in Medicine 37 (1998) 206--219.

No context found.

P.J.F. Lucas, H. Boot, and B.G. Taal. Computer-based decision-support in the management of primary gastric non-Hodgkin lymphoma. Methods of Information in Medicine, vol. 37, 1998, pp. 206 - 219.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC