Andreassen S, Benn JJ, Hovorka R, Olesen KG, Carson ER, 1994: A probabilistic approach to glucose prediction and insulin dose adjustment: description of a metabolic model and pilot evaluation study. Computer Methods Programs Biomedical 41:153-163.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....the optimal policies can change over the time, suggesting only suboptimal solutions that, sometimes, can differ widely from the optimal ones. So, in literature, some other modellization instruments are proposed in order to cope with dynamic decision problems as dynamic influence diagrams used in [3,4] or the framework described in [5] In this context, my work presents an innovative approach, based on Markov Decision Processes (MDPs) 6] to describe and solve decision problems in which the optimal choice has to be revised periodically in accordance to the evolution of the patient s ....

Andreassen S., Benn J., Hovorka R., Olesen K. G., and Carson E. R. A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study. Computer Methods and Prog. in Biomedicine 1994; 41:153-165.

....the major drawback of Dynamic In fluence Diagrams is related to the unnecessarily large number of state variables that they must manage. As a matter of fact, in the majority of the published systems that exploits Bayesian Belief Networks or Dynamic Influence Diagrams in Biomedical applic ations [12, 13, 14, 15], intermediate variables are used to simplify knowledge acquisition or speed up computations. In these applications, different solutions, often related the peculiarity of the problem at hand, have been implemented. In this paper we will cope with this problem in a more general fashion, by ....

S. Andreassen, J. Benn, R. Hovorka, K.G. Olesen, E.R. Carson, A probabilistic approach to glucose pre- diction and insulin dose adjustment: description of metabolic model and pilot evaluation study, Computer Methods and Programs in Biomedicine, 41 (1994) 153-165.

.... rate of change of the blood glucose as a result of the insulin dependent glucose production (term three in equation (12) the insulin dependent utilization (term four) the insulin independent glucose removal (term five) and renal clearance (term six) The small circles indicate published data [27] and the solid lines show the fitted parameterization used in equation (12) The plots for insulin dependent glucose production and utilization, respectively are shown for three (from top to down: 0 U ml, 10 U ml, 20 U ml) respectively four (from top to down: 80 U ml, 40 U ml, 20 U ml, 10 U ml) ....

Andreassen S., Benn J., Hovorka R., Olesen K., Carson E., "A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study", Computer Methods and Programs in Biomedicine, Vol. 41, pp. 153--165, Elsevier Sc. Publ. Ireland, 1994.

....analysis of the collected data are potentially of great help for the therapy revision as well as for the overall assessment of the patient s behavior. Since the early 1980 s, several systems have been proposed to assist patients and physicians, through a wide spectrum of different approaches [2] [6]. Nevertheless, the analysis of data coming from home monitoring of IDDM patients still remains a rather complex task. The main problems related to the data collection process are as follows. BGL sampling is still invasive, although progressively less painful. Such problem hampers the ....

S. Andreassen, J. Benn, R. Hovorka, K. G. Olesen, and E. R. Carson, "A probabilistic approach to glucose prediction and insulin dose adjustment: Description of metabolic model and pilot evaluation study," Comput. Meth. Programs Biomed., vol. 41, pp. 153--165, 1994.

....collection, data analysis, decision support, and, more recently, in a telematic management of the disease [8] Nevertheless, the analysis of data coming from home monitoring of IDDM patients still remains a rather complex task. A wide spectrum of approaches have been proposed in the literature [6, 9, 11 13]. One of the main difficulties is related to the problem that, in real clinical practice, often the only available data are the BGL measurements, that may be automatically down loaded from blood glucose reflectometers. This practical limitation has led to the definition of decision support tools ....

Andreassen, S., Benn, J., Hovorka, R., Olesen, K.G., Carson, E.R.: A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study. Computer Methods and Programs in Biomedicine. 41 (1994) 153-165.

....collection, data analysis, decision support, and, more recently, in a telematic manage ment of the disease [5] Nevertheless, the analysis of data coming from home monitoring of IDDM patients still remains a rather complex task. A wide spectrum of approaches have been proposed in the literature [24, 12, 20, 2]. One of the main difficulties is related to the problem that, in real clinical practice, often the only available data are the BGL measurements, that may be automatically down loaded from blood glucose reflectometers. This practical limita tion has led to the definition of decision support tools ....

....the patient behavior. 5 Comparison with related approaches As mentioned in the introduction, a high number of approaches have been presented for the analysis of data coming from Diabetic patients home monitoring. A number of such approaches have been devoted to the prediction of BGL time series [24, 22, 2], while few of them were oriented to an overall interpretation of the patient behavior [20, 31, 28] including 18 some commercial products, like Camit Pro TM or Eurotouch TM. While the difference of our approach with respect to the former class of such system is quite clear, a brief comment ....

S. Andreassen, J. Benn, R. Hovorka, K.G. Olesen, E.R. Carson, A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study. Computer Methods and Programs in Biomedicine 41 (1994) 153- 165.

.... far, systems which provide therapy recommendations are either based on expert systems or are based on physiological models [2] 3] 4] 5] 6] 7] 8] 9] 10] 11] 12] A promising approach DRAFT May 6, 1999 TRESP, BRIEGEL AND MOODY 3 is pursued by Andreassen, Hejlesen and co workers [13], 14] who use a causal probabilistic network to model the glucose metabolism and to derive therapy recommendations. So far, none of the systems have gained widespread acceptance in therapy. Hejlesen, Andreassen, Hovorka, and Cavan [14] attribute this fact to the major problems associated with the ....

....t Gamma1 ) 3) Gammac 5 (v t;1 v t;2 ) y t Gamma1 c 6 ) Gamma c 7 p y t Gamma1 Gamma c 8 y 3 t Gamma1 Gamma c 9 (v t;6 v t;7 ) where y t is the blood glucose at time t. 1 This nonlinear difference equation was derived from parameterizing published data describing the dependencies [13], 18] 15] 16] 19] The second term on the right side of the difference equation describes the increase in blood glucose due to carbohydrates in the food, the third term approximates the insulin 1 As mentioned above, the time resolution of our system is 15 minutes. May 6, 1999 DRAFT ....

[Article contains additional citation context not shown here]

Andreassen S., Benn J., Hovorka R., Olesen K., and Carson E., "A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study," Computer Methods and Programs in Biomedicine, vol. 41, pp. 153--165, 1994, Elsevier Sc. Publ. Ireland.

....not exactly the same (neither top down modeling nor repetitive structures appear in the Koller and Pfeffer framework where the main focus is on modularisation and probability functions) 1. 1 Limitations of conventional Bayesian network construction Suppose we want to build the BN described in [Andreassen et al. 1994] for evaluating a diabetes patients insulin dose. A natural way to think about this domain conveys a model for each hour with some variables describing the state of the subject during the previous hour together with some static characteristics of the subject (see Figure 1) The whole BN is then ....

....utilisation. Glucuria: Amount of glucose in the urine of the diabetes during this hour. Met Irr: Metabolic irregularities (trap for random fluctuations) BG: Amount of glucose is in the diabetes blood during this hour. Figure 1: The pattern for each hour of the BN for insulin prediction (see [Andreassen et al. 1994]) 1.2 A new framework The BN designer should be able to implement the BN fragment by fragment (in any order) and to use each fragment any number of times in any step of the construction process. We call these fragments templates 1 . Each change made inside such a template should update each ....

[Article contains additional citation context not shown here]

Andreassen, S., Benn, J., Hovorka, R., Olesen, K. G., and Carson, E. R. (1994). A probabilistic approach to glucose prediction and insulin dose adjustment: Description of a metabolic model and pilot evaluation study. Computer Methods and Programs in Biomedicine, 41:153--165.

....automated reasoning tools. They can handle the automatic transmission of the data from the patient s house to the clinic, thus increasing the frequency and the reliability of the communication. Many examples of the application of computer systems to diabetes therapy exist in the literature [2, 7], although few of them were really successful. We believe that, in order to be effective, a computerbased system for diabetes therapy must be designed according to the constraints imposed by the characteristics of the disease and of the current therapeutic schemes. We have therefore defined a ....

S. Andreassen, J. Benn, R. Hovorka, K. G. Olesen, and E. R. Carson. A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study. Computer Methods and Programs in Biomedicine, (41):153--165, 1994.

....introduce a compact representation of BNs with repetitive structures. Compact representation of repetitive structures In this section we will introduce a compact way of representing repetitive structures, illustrated through a model used for adjusting the insuline dose of diabetics developed by (Andreassen et al. 1994). In their model there is a BN fragment modeling an hour for a diabetic. This fragment is represented as a class in Figure 6. Twenty four interconnected instantiations of this class then represents a day for a diabetic. CHO old Meal BG old Ins Sens Ins Release Bound Ins Dep Ra Renal Gl ....

Andreassen, S., Benn, J., Hovorka, R., Olesen, K. G., and Carson, E. R. (1994). A probabilistic approach to glucose prediction and insulin dose adjustment: Description of a metabolic model and pilot evaluation study. Computer Methods and Programs in Biomedicine, 41:153-165.

....rst two points are addressed. The last issue will be addressed in a later report. 1. 1 Limits of simple Bayesian network framework Suppose we want to build a BN for the modelisation of the 24 hour glucose prediction and insulin dose adjustment for insulin dependent diabetic subjects (see [1] and [2]) The natural way of thinking about this kind of BN is to model the computation of the blood glucose each hour given some variables describing the state of the subject during the previous hour and some static characteristics of the subject (see Table 1) The whole BN then consists of the ....

Steen Andreassen, Jonathan Benn, Roman Hovorka, Kristian G. Olesen, and Ewart R. Carson. A probabilistic approach to glucose prediction and insulin dose adjustment: Description of a metabolic model and pilot evaluation study. Computer Methods and Programs in Biomedicine, 41:153-165, 1994.

....uncertainty is of central concern. In this paper, such systems will be called decision theoretic expert systems. The main applications of the formalisms are in classification, e.g. diagnosis (cf. 4] and in decision making under uncertainty, e.g. optimal treatment management of a patient (cf. [1]) The decision theoretic network formalisms originate from two different fields: knowledgebased systems [11] and statistical decision theory [16] which is also reflected in the various issues that arise when building decision theoretic expert systems. As with any expert system, extracting ....

Andreassen, S, Benn, J, Hovorka, R, Olesen, KG, Carson, ER, 1994. A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study, Computer Methods and Programs in Biomedicine 41, 153--165.

....uncertainty is of central concern. In this paper, such systems will be called decision theoretic expert systems. The main applications of the formalisms are in classi cation, e.g. diagnosis (cf. 4] and in decision making under uncertainty, e.g. optimal treatment management of a patient (cf. [1]) The decision theoretic network formalisms originate from two di erent elds: knowledgebased systems [11] and statistical decision theory [16] which is also re ected in the various issues that arise when building decision theoretic expert systems. As with any expert system, extracting ....

Andreassen, S, Benn, J, Hovorka, R, Olesen, KG, Carson, ER, 1994. A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study, Computer Methods and Programs in Biomedicine 41, 153-165.

....the major drawback of Dynamic Influence Diagrams is related to the unnecessarily large number of state variables that they must manage. As a matter of fact, in the majority of the published systems that exploits Bayesian Belief Networks or Dynamic Influence Diagrams in Biomedical applications [12, 13, 14, 15], intermediate variables are used to simplify knowledge acquisition or speed up computations. In these applications, different solutions, often related the peculiarity of the problem at hand, have been implemented. In this paper we will cope with this problem in a more general fashion, by ....

S. Andreassen, J. Benn, R. Hovorka, K.G. Olesen, E.R. Carson, A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study, Computer Methods and Programs in Biomedicine, 41 (1994) 153-165.

....600 days. This example demonstrates the high theoretical and practical importance of considering, when necessary, dynamic models instead of static ones. 7. Conclusion Decision Theoretic planning can play a crucial role in assessing optimal therapies in a large number of medical applications [13], 14] In this paper we have shown the applicability of MDPs to a complex medical problem, using a novel approach, based on graphical models. Our proposed framework is particularly appealing in simplifying the knowledge acquisition process, but presents the same computational complexity as the ....

S. Andreassen, J. Benn, R. Hovorka, K. G. Olesen, and E. R. Carson. A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study. Computer Methods and Programs in Biomedicine, 41:153--165, 1994.

....tools have been proposed in the field of data interpretation. The methodologies used to this purpose range from simple statistical analysis and graphical representation of the raw data, to more complex techniques, like time series analysis [ Deutsch et al. 1994 ] causal probabilistic networks [ Andreassen et al. 1994 ] and temporal abstractions [ Shahar and Musen, 1996 ] In this paper we explicitly cope with the problem of defining an effective system for helping data interpretation even when the information is very poor. We propose a novel approach based on the combination of ta methods with statistical ....

S. Andreassen, J. Benn, R. Hovorka, K.G. Olesen, and E.R. Carson. A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study. Computer Methods and Programs in Biomedicine, 41 (1994) 153-165.

.... researchers have used belief networks in dynamic domains such as the fall diagnosis problem, where the world changes and the focus is reasoning over time [7, 11, 15, 10] Such dynamic applications include robot navigation and map learning based on temporal belief networks [7] monitoring diabetes [1], monitoring robot vehicles [13] oil forecasting [5] 17] forecasting sleep apnea [4] automated vehicle control [8] and traffic plan recognition [18] For such applications the network grows over time, as the state of each domain variable at different times is represented by a series of ....

S.A. Andreassen, J.J. Benn, R. Hovorks, K. G. Olesen, and R. E Carson. A probabilistic approach to glucose prediction and insulin dose adjustment: Description of metabolic model and pilot evaluation study. Unpublished draft, 1991.

.... work on the dynamic construction of belief networks [6, 7] More recently, dynamic belief networks (DBNs) also called temporal probabilistic networks [12, 13] and dynamic causal probabilistic networks [23] have been of interest as modeling tools for environments that change over time [23, 10, 14, 3]. For such applications, the network expands over time, as the state of each domain variable at different times is represented by a series of nodes. The monitoring system in this paper is based on the construction of such a dynamic belief network to represent the world and the events which occur ....

....networks in dynamic domains, where the world changes and the focus is reasoning over time. Such dynamic applications include Dean et al. s work on robot navigation, planning and map learning, based on temporal belief networks, 14] and Andreassen et al. s work on monitoring diabetes over time [3]. The DBN monitoring system described in this paper may be described by Kjaerulff s formal computational scheme for reducing and expanding dynamic probabilistic networks [23] Agogino et al. 1] use real time influence diagrams for diagnostic reasoning, monitoring and controlling mechanical ....

S. Andreassen, J. Benn, R. Hovorks, K. G. Olesen, and R. E. Carson, "A probabilistic approach to glucose prediction and insulin dose adjustment: Description of metabolic model and pilot evaluation study." Unpublished draft, 1991.

.... (also called Temporal Probabilistic Networks [Dean and Kanazawa, 1989, Dean et al. 1990] and Dynamic Causal Probabilistic Networks [Kjaerulff, 1992] have been of interest as modelling tools for environments that change over time [Kjaerulff, 1992, Dagum et al. 1992, Dean and Wellman, 1991, Andreassen et al. 1991, Nicholson, 1992] For such applications the network expands over time, as the state of each domain variable at different times is represented by a series of nodes. Dynamic Belief networks have the following general characteristics. The nodes can be divided into three general categories: world ....

Andreassen, S.A.; Benn, J.J.; Hovorks, R.; Olesen, K. G.; and Carson, R. E 1991. A probabilistic approach to glucose prediction and insulin dose adjustment: Description of metabolic model and pilot evaluation study. Unpublished draft.

.... dominant in order to detect and describe functional dependencies between input and output variables (e.g. 3] These methods, however, assume a certain kind of functional dependency (usually a linear polynomial in the input variables) and then adapt available parameters appropriately (see also [2]) Methods based on the k NNR are not restricted to a certain a priori chosen type of functional dependency. They essentially can approximate arbitrary (continuous) functions (cp. 7, 9, 5, 19] Most interpolation methods (e.g. regression, genetic programming, neural networks) generate an ....

Andreassen, S.; Benn, J.; Hovorka, R.; Olesen, K.; Carson, E.: A Probabilistic Approach to Glucose Prediction and Insulin Dose Adjustment: Description of Metabolic Model and Pilot Evaluation Study, Comp. Methods Programs Biomed. 41:153--165, 1994.

....discretization level, In this paper, we have studied the problem with 3 time slices (H 3) i,e, Before Breakfast (BR) Before Lunch (LU) and Before Dinner (D) 5 discretiza tion values (L = 5) for BGL, 20 discretization levels for , k 0. 125 hours and the cost function defined in DIAS [14]. It is finally crucial to remark that such new model may present two basic problems in its application: i) the iden tification procedure may lead to a very fiat probability distribution for S. This may be due to an insufficient change in insulin regimen related to the data exploited for ....

Andreassen, S., Benn, J., Hovorka, R., Olesen, K.G., Carson, E.R.: A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study. Computer Methods and Programs in Biomedicine 41 (1994) 153-165

No context found.

Andreassen S, Benn JJ, Hovorka R, Olesen KG, Carson ER, 1994: A probabilistic approach to glucose prediction and insulin dose adjustment: description of a metabolic model and pilot evaluation study. Computer Methods Programs Biomedical 41:153-163.

No context found.

Andreassen S, Benn JJ, Hovorka R, Olesen KG, Carson ER, 1994: A probabilistic approach to glucose prediction and insulin dose adjustment: description of a metabolic model and pilot evaluation study. Computer Methods Programs Biomedical 41:153-163.

No context found.

Andreassen S, Benn J, Hovorka R, Olesen KG, Carson ER. A probabilistic approach to glucose prediction and insulin dose adjustment: description of metabolic model and pilot evaluation study. Computer Methods and Programs in Biomedicine. 1994; 41:153-165.

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