A Tutorial on Learning with Bayesian Networks

Heckerman, David

doi:10.1007/978-3-540-85066-3_3

Cited by 987 publications

(1,013 citation statements)

References 53 publications

Supporting

Mentioning

980

Contrasting

Unclassified

Order By: Relevance

“…Algorithms to perform this structure discovery have been developed relatively recently 23 and have become computationally tractable even more recently. 24 It is now possible to discover a Bayesian network of several hundred nodes from a relatively small dataset in a matter of hours on a desktop computer.…”

Section: Background and Significancementioning

confidence: 99%

Patient-tailored prioritization for a pediatric care decision support system through machine learning

Klann

Anand

Downs

2013

J Am Med Inform Assoc

View full text Add to dashboard Cite

Objective Over 8 years, we have developed an innovative computer decision support system that improves appropriate delivery of pediatric screening and care. This system employs a guidelines evaluation engine using data from the electronic health record (EHR) and input from patients and caregivers. Because guideline recommendations typically exceed the scope of one visit, the engine uses a static prioritization scheme to select recommendations. Here we extend an earlier idea to create patient-tailored prioritization. Materials and methods We used Bayesian structure learning to build networks of association among previously collected data from our decision support system. Using area under the receiver-operating characteristic curve (AUC) as a measure of discriminability (a sine qua non for expected value calculations needed for prioritization), we performed a structural analysis of variables with high AUC on a test set. Our source data included 177 variables for 29 402 patients. Results The method produced a network model containing 78 screening questions and anticipatory guidance (107 variables total). Average AUC was 0.65, which is sufficient for prioritization depending on factors such as population prevalence. Structure analysis of seven highly predictive variables reveals both face-validity (related nodes are connected) and non-intuitive relationships. Discussion We demonstrate the ability of a Bayesian structure learning method to 'phenotype the population' seen in our primary care pediatric clinics. The resulting network can be used to produce patient-tailored posterior probabilities that can be used to prioritize content based on the patient's current circumstances. Conclusions This study demonstrates the feasibility of EHR-driven population phenotyping for patient-tailored prioritization of pediatric preventive care services. BACKGROUND AND SIGNIFICANCE

show abstract

Section: Background and Significancementioning

confidence: 99%

Patient-tailored prioritization for a pediatric care decision support system through machine learning

Klann

Anand

Downs

2013

J Am Med Inform Assoc

View full text Add to dashboard Cite

show abstract

“…This, of course, makes it useless with respect to structure learning, and therefore, a penalty factor for network complexity should be introduced to the scoring equation (9.23). According to Heckerman (1995), one obtains:…”

Section: Learning Bayesian Networkmentioning

confidence: 99%

Data Integration, Metabolic Networks and Systems Biology

Redestig

Szymański

Hirai

et al. 2011

Annual Plant Reviews Volume 43

View full text Add to dashboard Cite

As analytical techniques and data pre-processing methods continue to improve, the bottleneck of metabolomics is shifting towards later stages of data analysis and biological interpretation. High-coverage metabolomics is only possible when combining data from multiple platforms necessitating efficient methods for data integration. Metabolomic data sets with high coverage provide a unique opportunity to estimate and study metabolic networks. Once established, these networks can provide a backbone for systems biology approaches where the aim is to construct fundamental models of metabolic regulation. In this chapter, we provide an overview of status of these topics and describe current methods and tools, their drawbacks and advantages for integrative plant metabolomics.

show abstract

“…environmental policy studies (Wolfson et al 1996). A BBN is a graphical model for probabilistic relationships among a set of variables (Pearl 1993, Heckerman 1999 and gives a compact representation of reasoning under uncertainty by making reference to Bayes' rule for computing probabilistic inference (Smid et al 2010). BBNs offer many advantages.…”

Section: Introductionmentioning

confidence: 99%

“…BBNs offer many advantages. They readily handle incomplete data sets (Heckerman 1999), they concisely represent probabilistic relationships (Cooper 1990, Pearl 1993, and their graphical user interface makes the approach simple to use for non-experts (Smid et al 2010). Their drawback is that they do not allow for inclusion of direct feedbacks in the analysis, which limits their use in vulnerability assessments.…”

Section: Introductionmentioning

confidence: 99%

Application of Fuzzy Cognitive Mapping in Livelihood Vulnerability Analysis

Murungweni¹,

Wijk²,

Andersson³

et al. 2011

E&S

View full text Add to dashboard Cite

ABSTRACT. Feedback mechanisms are important in the analysis of vulnerability and resilience of social-ecological systems, as well as in the analysis of livelihoods, but how to evaluate systems with direct feedbacks has been a great challenge. We applied fuzzy cognitive mapping, a tool that allows analysis of both direct and indirect feedbacks and can be used to explore the vulnerabilities of livelihoods to identified hazards. We studied characteristics and drivers of rural livelihoods in the Great Limpopo Transfrontier Conservation Area in southern Africa to assess the vulnerability of inhabitants to the different hazards they face. The process involved four steps: (1) surveys and interviews to identify the major livelihood types; (2) description of specific livelihood types in a system format using fuzzy cognitive maps (FCMs), a semi-quantitative tool that models systems based on people's knowledge; (3) linking variables and drivers in FCMs by attaching weights; and (4) defining and applying scenarios to visualize the effects of drought and changing park boundaries on cash and household food security. FCMs successfully gave information concerning the nature (increase or decrease) and magnitude by which a livelihood system changed under different scenarios. However, they did not explain the recovery path in relation to time and pattern (e.g., how long it takes for cattle to return to desired numbers after a drought). Using FCMs revealed that issues of policy, such as changing situations at borders, can strongly aggravate effects of climate change such as drought. FCMs revealed hidden knowledge and gave insights that improved the understanding of the complexity of livelihood systems in a way that is better appreciated by stakeholders.

show abstract

A Tutorial on Learning with Bayesian Networks

Cited by 987 publications

References 53 publications

Patient-tailored prioritization for a pediatric care decision support system through machine learning

Patient-tailored prioritization for a pediatric care decision support system through machine learning

Data Integration, Metabolic Networks and Systems Biology

Application of Fuzzy Cognitive Mapping in Livelihood Vulnerability Analysis

Contact Info

Product

Resources

About