umu.sePublications
Change search
Refine search result
1 - 25 of 25
CiteExportLink to result list
Permanent link
Cite
Citation style
• apa
• ieee
• modern-language-association-8th-edition
• vancouver
• Other style
More styles
Language
• de-DE
• en-GB
• en-US
• fi-FI
• nn-NO
• nn-NB
• sv-SE
• Other locale
More languages
Output format
• html
• text
• asciidoc
• rtf
Rows per page
• 5
• 10
• 20
• 50
• 100
• 250
Sort
• Standard (Relevance)
• Author A-Ö
• Author Ö-A
• Title A-Ö
• Title Ö-A
• Publication type A-Ö
• Publication type Ö-A
• Issued (Oldest first)
• Issued (Newest first)
• Created (Oldest first)
• Created (Newest first)
• Last updated (Oldest first)
• Last updated (Newest first)
• Disputation date (earliest first)
• Disputation date (latest first)
• Standard (Relevance)
• Author A-Ö
• Author Ö-A
• Title A-Ö
• Title Ö-A
• Publication type A-Ö
• Publication type Ö-A
• Issued (Oldest first)
• Issued (Newest first)
• Created (Oldest first)
• Created (Newest first)
• Last updated (Oldest first)
• Last updated (Newest first)
• Disputation date (earliest first)
• Disputation date (latest first)
Select
The maximal number of hits you can export is 250. When you want to export more records please use the Create feeds function.
• 1. Babb, Penny
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Statistical challenges of administrative and transaction data: Discussion on the paper by Hand2018In: Journal of the Royal Statistical Society: Series A (Statistics in Society), ISSN 0964-1998, E-ISSN 1467-985X, Vol. 181, no 3, p. 578-605Article in journal (Other academic)
• 2.
Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine, Reumatology.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics. Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine, Reumatology.
Rheumatoid Factor Isotypes in Relation to Antibodies Against Citrullinated Peptides and Anti-Carbamylated Antibodies in Individuals Before the Onset of Rheumatoid Arthritis2016In: Arthritis Research & Therapy, ISSN 1478-6354, E-ISSN 1478-6362, Vol. 18, article id 43Article in journal (Refereed)

Background: The presence of rheumatoid factor (RF), anti-carbamylated protein antibodies (anti-CarP) and antibodies against citrullinated protein and peptides (ACPA) precedes the onset of symptoms of rheumatoid arthritis (RA) by several years. Relationships between the development of these antibodies are not obvious.

Methods: Three isotypes [immunoglobulin A (IgA), IgG and IgM) of RF were analysed in 321 pre-symptomatic individuals who provided 598 samples collected a median of 6.2 (interquartile range 7.2) years before the onset of symptoms, and in 492 population control subjects. All samples were donated to the Biobank of Northern Sweden. RF isotypes were analysed using the EliA system (Phadia GmbH, Freiburg, Germany) with 96 % specificity according to receiver operating characteristic curves. Ten ACPA specificities were analysed using the ImmunoCAP ISAC system, and anti-CCP2 and anti-CarP antibodies were evaluated using enzyme-linked immunosorbent assays.

Results: The frequencies of RF isotypes in pre-symptomatic individuals were significantly increased compared with control subjects (p < 0.0001). In samples collected >= 15 years before the onset of symptoms, the IgA-RF isotype was significantly more prevalent than the most frequent ACPAs. Combinations of IgM- and IgA-RF isotypes with ACPA specificities [a-enolase (CEP-1/Eno(5-21))], fibrinogen (Fib)beta(36-52), Fiba(580-600), filaggrin (CCP-1/Fil(307-324)) and anti-CCP2 antibodies were associated with a significantly shorter time to onset of symptoms (p < 0.001-0.05). Using conditional inference tree analysis, anti-CCP2 in combination with anti-filaggrin antibodies gave the highest probability, 97.5 %, for disease development.

Conclusions: RF isotypes predicted the development of RA, particularly in combination with ACPA, anti-CCP2 or anti-CarP antibodies. The highest probability for disease development was the presence of anti-CCP2 and anti-filaggrin antibodies.

• 3. Dawid, Philip
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Beyond subjective and objective in statistics: Discussion on the paper by Gelman and Hennig2017In: Journal of the Royal Statistical Society: Series A (Statistics in Society), ISSN 0964-1998, E-ISSN 1467-985X, Vol. 180, no 4, p. 997-1033Article in journal (Other academic)

Subjectivity and objectivity in statistical modelling is discussed.

• 4. Ericzon, Bo-Goran
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics. Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine, Medicine.
To transplant or not to transplant - Lessons learned from 20 years global collaboration in liver transplantation for hereditary transthyretin amyloidosis2013In: Hepatology, ISSN 0270-9139, E-ISSN 1527-3350, Vol. 58, p. 1011A-1011AArticle in journal (Other academic)
• 5. Ericzon, Bo-Göran
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics. Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine, Medicine.
Liver transplantation for hereditary transthyretin amyloidosis: after 20 years still the best therapeutic alternative?2015In: Transplantation, ISSN 0041-1337, E-ISSN 1534-6080, Vol. 99, no 9, p. 1847-1854Article in journal (Refereed)

BACKGROUND: Until recently, liver transplantation (Ltx) was the only available treatment for hereditary transthyretin (TTR) amyloidosis; today, however, several pharmacotherapies are tested. Herein, we present survival data from the largest available database on transplanted hereditary TTR patients to serve as a base for comparison.

METHODS: Liver transplantation was evaluated in a 20-year retrospective analysis of the Familial Amyloidosis Polyneuropathy World Transplant Registry.

RESULTS: From April 1990 until December 2010, data were accumulated from 77 liver transplant centers. The Registry contains 1940 patients, and 1379 are alive. Eighty-eight Ltx were performed in combination with a heart and/or kidney transplantation. Overall, 20-year survival after Ltx was 55.3%. Multivariate analysis revealed modified body mass index, early onset of disease (<50 years of age), disease duration before Ltx, and TTR Val30Met versus non-TTR Val30Met mutations as independent significant survival factors. Early-onset patients had an expected mortality rate of 38% that of the late-onset group (P < 0.001). Furthermore, Val30Met patients had an expected mortality rate of 61% that of non-TTR Val30Met patients (P < 0.001). With each year of duration of disease before Ltx, expected mortality increased by 11% (P < 0.001). With each 100-unit increase in modified body mass index at Ltx, the expected mortality decreased to 89% of the expected mortality (P < 0.001). Cardiovascular death was markedly more common than that observed in patients undergoing Ltx for end-stage liver disease.

CONCLUSIONS: Long-term survival after Ltx, especially for early-onset TTR Val30Met patients, is excellent. The risk of delaying Ltx by testing alternative treatments, especially in early-onset TTR Val30Met patients, requires consideration.

• 6.
Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine, Medicine.
Umeå University, Faculty of Social Sciences, Department of Statistics. Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine, Medicine.
Continuous development of arrhythmia is observed in Swedish transplant patients with familial amyloidotic polyneuropathy (amyloidogenic transthyretin Val30Met variant)2011In: Liver transplantation, ISSN 1527-6465, E-ISSN 1527-6473, Vol. 17, no 2, p. 122-128Article in journal (Refereed)

In patients with familial amyloidotic polyneuropathy (FAP), heart complications are prognostic factors for mortality and morbidity after liver transplantation (LT). However, only a few studies have analyzed the development of arrhythmia in transplant patients with FAP. We investigated the development of arrhythmia requiring pacemaker insertion (PMI) in Swedish transplant patients with FAP, and we related the findings to gender, age at disease onset, and survival. One hundred four transplant patients with the amyloidogenic transthyretin Val30Met mutation were included in the study. Twenty-six (25%) received a pacemaker during the observation period (a median of 11 years after disease onset). This frequency was comparable to that noted in a previous study describing the natural course of FAP. No significant differences in PMI between early-onset cases (<50 years old) and late-onset cases (≥ 50 years old) or between genders were observed. PMI was not significantly related to patient survival. Our study confirms our previously reported short-time observation: LT does not prevent the development of heart arrhythmia necessitating PMI. The development of arrhythmia is unrelated to gender or age at disease onset, and the yearly risk does not appear to decrease with time after LT.

• 7.
Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine, Medicine.
Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine, Medicine. Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine, Medicine. Umeå University, Faculty of Medicine, Department of Surgical and Perioperative Sciences, Clinical Physiology. Center for Surgical Sciences, Karolinska Institute, Karolinska University Hospital, Huddinge, Sweden. Umeå University, Faculty of Science and Technology, Department of Mathematics and Mathematical Statistics. Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine, Medicine. Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine, Medicine.
Development of cardiomyopathy after liver transplantation in Swedish hereditary transthyretin amyloidosis (ATTR) patients2011In: Amyloid: Journal of Protein Folding Disorders, ISSN 1350-6129, E-ISSN 1744-2818, Vol. 18, no 4, p. 200-205Article in journal (Refereed)

Background: Recent studies of liver transplanted (LTx) familial amyloidotic polyneuropathy (FAP) patients have shown a progression of cardiomyopathy in some patients after LTx, but knowledge of the underlying factors remains limited.

Methods: Seventy-five patients, who had undergone LTx from 1996 to 2008, were included. They had all been examined by echocardiography 1-16 months before LTx. Fifty-four had been re-examined 7-34 months, and forty-two 36-137 months after LTx.

Results: A significant increase in interventricular septum (IVS) thickness occurred after LTx (p < 0.01), particularly in males (p = 0.002) and late onset patients (p = 0.003). The development of post-LTx cardiomyopathy was related to patient's age at onset of the disease, male gender and pre-LTx IVS thickness. On multivariate regression analysis, however, age at onset was the only significant predictor for the development of cardiomyopathy (odds ratio = 1.14, 95% confident interval 1.01-1.30, p = 0.04).

Conclusion: An increase of IVS thickness can be observed in FAP patients after LTx. Age at onset of the disease is the main predictor for increased IVS thickness and for the development of cardiomyopathy after liver transplantation.

• 8.
Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine.
Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine. Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine. Umeå University, Faculty of Medicine, Department of Public Health and Clinical Medicine. Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Amyloid fibril composition within hereditary Val30Met (p. Val50Met) transthyretin amyloidosis families2019In: PLoS ONE, ISSN 1932-6203, E-ISSN 1932-6203, Vol. 14, no 2, article id e0211983Article in journal (Refereed)

Background: The amyloid fibril in hereditary transthyretin (TTR) Val30Met (pVal50Met) amyloid (ATTR Val30Met) amyloidosis is composed of either a mixture of full-length and TTR fragments (Type A) or of only full-length TTR (Type B). The type of amyloid fibril exerts an impact on the phenotype of the disease, and on the outcome of diagnostic procedures and therapy. The aim of the present study was to investigate if the type of amyloid fibril remains the same within ATTR Val30Met amyloidosis families. Methods: Fifteen families were identified in whom at least two first-degree relatives had their amyloid fibril composition determined. The type of ATTR was determined by Western blot in all but two patients. For these two patients a positive 99mTc-3,3-diphosphono-1,2-propanodicarboxylic acid scintigraphy indicated ATTR Type A. Results: In 14 of the 15 families, the same amyloid fibril composition was noted irrespective of differences in age at onset. In the one family, different ATTR fibril types was found in two brothers with similar ages at onset. Conclusions: Family predisposition appears to have an impact on amyloid fibril composition in members of the family irrespective of their age at onset of disease, but if genetically determined, the gene/genes are likely to be situated at another location than the TTR gene in the genome.

• 9.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
A geometric view on Pearson's correlation coefficient and a generalization of it to non-linear dependencies2016In: Ratio Mathematica, ISSN 1592-7415, Vol. 30, p. 3-21, article id 1Article in journal (Refereed)

Measuring strength or degree of statistical dependence between two random variables is a common problem in many domains. Pearson's correlation coefficient $\rho$ is an accurate measure of linear dependence. We show that $\rho$ is a normalized, Euclidean type distance between joint probability distribution of the two random variables and that when their independence is assumed while keeping their marginal distributions. And the normalizing constant is the geometric mean of two maximal  distances; each between the joint probability distribution when the full linear dependence is assumed while preserving respective marginal distribution and that when the independence is assumed. Usage of it  is  restricted to linear dependence because it is based on  Euclidean type distances that are generally not metrics and considered full dependence is linear. Therefore, we argue that if a suitable distance metric is used while considering all possible maximal dependences then it can measure any non-linear dependence.  But then, one must define all the full dependences.  Hellinger distance that is a metric can be used as the distance measure between probability distributions and obtain a generalization of $\rho$ for the discrete case.

• 10.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Book Review: Bayesian Theory and Applications2013In: Qvintensen, ISSN 2000-1819, no 4, p. 18-19Article, book review (Other academic)
• 11.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Causal Effect Estimation Methods2014In: Journal of Statistical and Econometric Methods, ISSN 2241-0376, Vol. 3, no 2, p. 153-170Article in journal (Refereed)

Relationship between two popular modeling frameworks of causalinference from observational data, namely, causal graphical model andpotential outcome causal model is discussed. How some popular causaleﬀect estimators found in applications of the potential outcome causalmodel, such as inverse probability of treatment weighted estimator anddoubly robust estimator can be obtained by using the causal graphicalmodel is shown. We conﬁne to the simple case of binary outcome andtreatment variables with discrete confounders and it is shown how togeneralize results to cases of continuous variables.

• 12.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Correlation and causation explained2013Other (Other (popular science, discussion, etc.))

Though the word correlation means usually how two quantities vary together, perhaps it may be due to extensive use of Pearson’s correlation coefficient. It is often associated with a linear relationship between the two. Even many scientific people misinterpret zero correlation as independence of the two quantities forgetting about what it really means; there is no linear association between the two quantities concerned.

Furthermore many confuse correlation with causation, i.e., many interpret non–zero correlation as an implied causal relationship. There are many examples of it even in scientific literature. Here our point is not to discuss all these misinterpretations, but to look at another thing on correlation, that is also related to causation in some sense.

• 13.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Data Deluge and Its Analysis Issues2016In: Proceedings of the 2nd International Conference in Accounting Researchers and Educators (ICARE 2016), Kelaniya: Department of Accountancy, University of Kelaniya , 2016, p. 21-21, article id 21Conference paper (Refereed)

Current availability of enormous amount of data is mainly due to technological advances. They are useful drawing inferences for creating new businesses, formulation of new policies or revising existing ones, etc. However, much of analyses are performed either by subject domain experts implementing mathematical and computational models incorrectly or by mathematical and computational professionals, purely on data driven basis without paying required attention to the subject domain knowledge. Both of these exercises often result in incorrect inferences and therefore they may harm the society, especially when their inferences are used in practice. We argue that, in order to get valid inferences these two parties should work together. Here we briefly discuss some of the issues that the large-scale data analyses should take into account, especially in open data and big data. We also briefly discuss our solutions that are rather simple to implement.

• 14.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Discussion on the paper by Caron and Fox2017In: Journal of The Royal Statistical Society Series B-statistical Methodology, ISSN 1369-7412, E-ISSN 1467-9868, Vol. 79, no 5, p. 1359-1359Article in journal (Other academic)

A measure of dependence for graph models

• 15.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
On Associative Confounder Bias2015In: Thirteenth Scandinavian Conference on Artificial Intelligence / [ed] Sławomir Nowaczyk, 2015, Vol. 278, p. 157-166Conference paper (Refereed)

Conditioning on some set of confounders that causally affect both treatmentand outcome variables can be sufficient for eliminating bias introduced by allsuch confounders when estimating causal effect of the treatment on the outcomefrom observational data. It is done by including them in propensity score modelin so-called potential outcome framework for causal inference whereas in causalgraphical modeling framework usual conditioning on them is done. However inthe former framework, it is confusing when modeler finds a variable that is noncausallyassociated with both the treatment and the outcome. Some argue that suchvariables should also be included in the analysis for removing bias. But others arguethat they introduce no bias so they should be excluded and conditioning onthem introduces spurious dependence between the treatment and the outcome, thusresulting extra bias in the estimation. We show that there may be errors in boththe arguments in different contexts. When such a variable is found neither of theactions may give the correct causal effect estimate. Selecting one action over theother is needed in order to be less wrong.We discuss how to select the better action.

• 16.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Probabilistic Analysis of Balancing Scores for Causal Inference2015In: Journal of Mathematics Research, ISSN 1916-9795, E-ISSN 1916-9809, Vol. 7, no 2, p. 90-100Article in journal (Refereed)

Propensity scores are often used for stratification of treatment and control groups of subjects in observational data to remove confounding bias when estimating of  causal effect of the treatment on an outcome in so-called potential outcome causal modeling framework. In this article, we try to get some insights into basic behavior of  the propensity scores in a probabilistic sense. We do a simple analysis of their usage confining to the case of discrete confounding covariates and outcomes. While making clear about behavior of the propensity score our analysis shows how the so-called prognostic score can be derived simultaneously. However the prognostic score is derived in a limited sense in the current literature whereas our derivation is more general and shows all possibilities of having the score. And we call it outcome score. We argue that application of both the propensity score and the outcome score is the most efficient way for  reduction of dimension in the confounding covariates as opposed to current belief that the propensity score alone is the most efficient way.

• 17.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Probabilistic Graphical Models and Their Inferences (Tutorial)2019In: 2019 IEEE 4th International Workshops on Foundations and Applications of Self* Systems (FAS*W) / [ed] Robert Birke and Ingrid Nunes, 2019, p. 251-252Conference paper (Other academic)

Probabilistic graphical models are useful for mod- elling stochastic phenomena for doing inferences and reasoning under uncertainty. Especially, chain graph models and Bayesian networks can be used as probabilistic expert systems where inferences can be done with junction tree algorithm, etc. And they can be extended to capture multi-stage decision contexts. Fundamentally these models capture (in)dependence structure of the context, but model learning is hard in practice. There are methods to do this, from simple independence test-based ones to more advanced score-based methods. When these models are used as classifiers, model learning can be done discriminatively, thus resulting higher classification accuracies in them.

• 18.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Resolution to four probability paradoxes: Two-envelope, Wallet-game, Sleeping Beauty and Newcomb’s2019In: Proceedings of The 34th International Workshop on Statistical Modelling 2019Guimarães, Portugal, Volume II / [ed] Luís Meira-Machado and Gustavo Soutinho, Portugal, 2019, Vol. 2, p. 252-257Conference paper (Refereed)

So-called two-envelope, wallet-game, Sleeping Beauty and Newcomb’s paradoxes are resolved through simple logical and analytical arguments. We stress the need of such simple solutions to them, due to current controversies around their solutions. Unnecessarily complicated or misleading solutions can be avoided if contexts of these problems are analyzed critically and perhaps with common- sense arguments. Our simple solutions are important for applied probability and statistical methods, especially for practitioners of them.

• 19.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Resolving the Lord's Paradox2017In: Proceedings of the 32nd International Workshop on Statistical Modelling: Volume II / [ed] Marco Grzegorczyk and Giacomo Ceoldo, Groningen, Netherlands: International Workshop on Statistical Modelling (IWSM) , 2017, Vol. II, p. 223-226Conference paper (Refereed)

An explanation to Lord’s paradox using ordinary least square regres- sion models is given. It is not a paradox at all, if the regression parameters are interpreted as predictive or as causal with stricter conditions and be aware of laws of averages. We use derivation of a super-model from a given sub-model, when its residuals can be modelled with other potential predictors as a solution.

• 20.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Viewing Simpson’s Paradox2014In: Statistica & Applicazioni, ISSN 1824-6672, Vol. XII, no 2, p. 225-235Article in journal (Refereed)

Well known Simpson’s paradox is puzzling and surprising for many, especially for the empirical researchersand users of statistics. However there is no surprise as far as mathematical details areconcerned. A lot more is written about the paradox but most of them are beyond the grasp of suchusers. This short article is about explaining the phenomenon in an easy way to grasp using simplealgebra and geometry. The mathematical conditions under which the paradox can occur are madeexplicit and a simple geometrical illustration is used to describe it. We consider the reversal of theassociation between two binary variables, say, X and Y by a third binary variable, say, Z. We showthat it is always possible to define Z algebraically for non-extreme dependence between X and Y,therefore occurrence of the paradox depends on identifying it with a practical meaning for it in agiven context of interest, that is up to the subject domain expert. And finally we discuss the paradoxin predictive contexts since in literature it is argued that the paradox is resolved using causal reasoning.

• 21.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
A Consistency Result for Bayes Classifiers with Censored Response Data2013In: Theoretical mathematics and applications, ISSN 1792-9709, Vol. 3, no 4, p. 47-54Article in journal (Refereed)

Naive Bayes classiﬁers have proven to be useful in many prediction problems with complete training data. Here we consider the situation where a naive Bayes classiﬁer is trained with data where the response is right censored. Such prediction problems are for instance encountered in proﬁling systems used at National Employment Agencies. In this paper we propose the maximum collective conditional likelihood estimator for the prediction and show that it is strongly consistent under the usual identiﬁability condition.

• 22.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Discussion on the meeting on 'Data visualization'2019In: Journal of the Royal Statistical Society: Series A (Statistics in Society), ISSN 0964-1998, E-ISSN 1467-985X, Vol. 182, no 2, p. 433-441Article in journal (Other academic)

Visualizing both conditional and marginal associations in contingency tables by using simple diagrams is discussed

• 23.
Department of Mathematical and Computing Sciences, Tokyo Institute of Technology, Tokyo, Japan.
Department of Mathematical and Computing Sciences, Tokyo Institute of Technology, Tokyo, Japan.
Asymptotic Properties of Maximum Collective Conditional Likelihood Estimators for Naïve Bayes Classifiers2006In: International Journal of Statistics and Systems, ISSN 0973-2675Article in journal (Refereed)

Bayesian networks that are probabilistic expert systems can be used as classifiers. Special type of Bayesian networks called naive Bayes classifiers are popular in practice due to their good performance although they are relatively simple.

Enhancement of the performance of the naïve Bayes classifier is often done through various parameter learning methods where the usual method is the method of maximum likelihood estimation. Nevertheless, since the true target of interest of Bayes classifiers is estimation of the conditional probabilities, it is natural to learn their parameters by maximization of the collective conditional likelihoods. Therefore, recently there has been a growing interest in learning the parameters of the naïve Bayes classifiers through maximizing collective conditional likelihoods.

Strong consistency and asymptotic normality are two basic statistical properties which any decent estimator should have although they are primarily of theoretical nature. In this research, we prove the strong consistency and asymptotic normality of the maximum collective conditional estimators for naïve Bayes classifiers. Essentially our proof follows the classical ideas well-developed for the theory of maximum likelihood estimation.

• 24.
Department of Mathematical and Computing Sciences, Tokyo Institute of Technology, Tokyo, Japan.
Department of Mathematical and Computing Sciences, Tokyo Institute of Technology, Tokyo, Japan. Department of Mathematical and Computing Sciences, Tokyo Institute of Technology, Tokyo, Japan.
Appraisal of companies with Bayesian networks2006In: International Journal of Business Intelligence and Data Mining, ISSN 1743-8187, Vol. 1, no 3, p. 329-346Article in journal (Refereed)

Appraisal of companies is an important business activity. We mainly apply Bayesian networks for this classification task for Japanese electric company data. Firstly, few standard statistical techniques are performed. Then Bayesian networks are applied in four steps: (1) for implementing a current procedure of economical experts, where economical variables are clustered and then summarised for computing a score for deciding the economical state of the company, (2) the same is done but with clustering of economical variables based on data, (3) the naive Bayes classifier and (4) an improved naive Bayes classifier through adjusting its conditional density of each feature variable given the class variable, which are initially obtained by maximum likelihood estimation. Adjustments are done by using the simulated annealing optimisation. Finally, a sensible way for appraisal of companies is discussed.

• 25.
Umeå University, Faculty of Social Sciences, Umeå School of Business and Economics (USBE), Statistics.
Umeå University, Faculty of Medicine, Department of Radiation Sciences.
Discriminative Prediction of Adverse Events for Optimized Therapies Following Traumatic Brain Injury2019Conference paper (Refereed)

Traumatic brain injury (TBI) causes temporary or perma- nent alteration in brain functions. At intensive care units, TBI patients are usually multimodally monitored, thus rendering large volumes of data on many physiological variables. For the physician, these data are difficult to interpret due to their complexity, speed and volume. Thus, computa- tional aids are recommended, e.g., for predicting patient’s clinical status in near future. In this article, we describe a probabilistic model that can be used for aiding physician’s decision making process in TBI patient care in real time. Our model tries to capture time varying patterns of patient’s clinical information. The model is built by using a discrimi- native model learning framework so that it can predict adverse clinical events with a higher level of accuracy. That is, our model is built so that prediction of certain desired events are given more attention than that of the other less important ones. This can be achieved by estimating model parameters in such a way, for e.g. using a cost function, when a suitable model structure has been selected, that again can be done dis- criminatively. However, such estimation procedures have no closed form solutions, so numerical optimization methods are used.

1 - 25 of 25
CiteExportLink to result list
Permanent link
Cite
Citation style
• apa
• ieee
• modern-language-association-8th-edition
• vancouver
• Other style
More styles
Language
• de-DE
• en-GB
• en-US
• fi-FI
• nn-NO
• nn-NB
• sv-SE
• Other locale
More languages
Output format
• html
• text
• asciidoc
• rtf