Likelihood inferences with interval-censored data

En l’anàlisi de la supervivència el problema de les dades censurades en un interval es tracta, usualment,via l’estimació per màxima versemblança. Amb l’objectiu d’utilitzar una expressió simplificada de la funció de versemblança, els mètodes estàndards suposen que les condicions que produeixen la censura no afecten el temps de fallada. En aquest article formalitzem les condicions que asseguren la validesa d’aquesta versemblança simplificada. Així, precisem diferents condicions de censura no informativa i definim una condició de suma constant anàloga a la derivada en el context de censura per la dreta. També demostrem que les inferències obtingudes amb la versemblançaa simplificada són correctes quan aquestes condicions són certes. Finalment, tractem la identificabilitat de la funció distribució del temps de fallada a partir de la informació observada i estudiem la possibilitat de contrastar el compliment de la condició de suma constant.

The Analysis of interval censoring and double censoring via Markov chain Monte Carlo methods

L'Anàlisi de la supervivència s'utilitza en diferents camps per analitzar el temps transcorregut entre dos esdeveniments. El que distingeix l'anàlisi de la supervivència d'altres àrees de l'estadística és que les dades normalment estan censurades. La censura en un interval apareix quan l'esdeveniment final d'interès no és directament observable i només se sap que el temps de fallada està en un interval concret. Un esquema de censura més complex encara apareix quan tant el temps inicial com el temps final estan censurats en un interval. Aquesta situació s'anomena doble censura. En aquest article donem una descripció formal d'un mètode bayesà paramètric per a l'anàlisi de dades censurades en un interval i dades doblement censurades així com unes indicacions clares de la seva utilització o pràctica. La metodologia proposada s'ilustra amb dades d'una cohort de pacients hemofílics que es varen infectar amb el virus VIH a principis dels anys 1980's.

Efficient knowledge retrieval to calibrate input variables in forest fire prediction

Forest fires are a serious threat to humans and nature from an ecological, social and economic point of view. Predicting their behaviour by simulation still delivers unreliable results and remains a challenging task. Latest approaches try to calibrate input variables, often tainted with imprecision, using optimisation techniques like Genetic Algorithms. To converge faster towards fitter solutions, the GA is guided with knowledge obtained from historical or synthetical fires. We developed a robust and efficient knowledge storage and retrieval method. Nearest neighbour search is applied to find the fire configuration from knowledge base most similar to the current configuration. Therefore, a distance measure was elaborated and implemented in several ways. Experiments show the performance of the different implementations regarding occupied storage and retrieval time with overly satisfactory results.

Prediction of the economic cost of individual long-term care in the Spanish population

Pensions together with savings and investments during active life are key elements of retirement planning. Motivation for personal choices about the standard of living, bequest and the replacement ratio of pension with respect to last salary income must be considered. This research contributes to the financial planning by helping to quantify long-term care economic needs. We estimate life expectancy from retirement age onwards. The economic cost of care per unit of service is linked to the expected time of needed care and the intensity of required services. The expected individual cost of long-term care from an onset of dependence is estimated separately for men and women. Assumptions on the mortality of the dependent people compared to the general population are introduced. Parameters defining eligibility for various forms of coverage by the universal public social care of the welfare system are addressed. The impact of the intensity of social services on individual predictions is assessed, and a partial coverage by standard private insurance products is also explored. Data were collected by the Spanish Institute of Statistics in two surveys conducted on the general Spanish population in 1999 and in 2008. Official mortality records and life table trends were used to create realistic scenarios for longevity. We find empirical evidence that the public long-term care system in Spain effectively mitigates the risk of incurring huge lifetime costs. We also find that the most vulnerable categories are citizens with moderate disabilities that do not qualify to obtain public social care support. In the Spanish case, the trends between 1999 and 2008 need to be further explored.

An Interval Intelligent-based Approach for Fault Detection and Modelling

Not considered in the analytical model of the plant, uncertainties always dramatically decrease the performance of the fault detection task in the practice. To cope better with this prevalent problem, in this paper we develop a methodology using Modal Interval Analysis which takes into account those uncertainties in the plant model. A fault detection method is developed based on this model which is quite robust to uncertainty and results in no false alarm. As soon as a fault is detected, an ANFIS model is trained in online to capture the major behavior of the occurred fault which can be used for fault accommodation. The simulation results understandably demonstrate the capability of the proposed method for accomplishing both tasks appropriately

Dynamic diagnosis based on interval analytical redundancy relations and signs of the symptoms

A model-based approach for fault diagnosis is proposed, where the fault detection is based on checking the consistencyof the Analytical Redundancy Relations (ARRs) using an interval tool. The tool takes into account the uncertainty in theparameters and the measurements using intervals. Faults are explicitly included in the model, which allows for the exploitation of additional information. This information is obtained from partial derivatives computed from the ARRs. The signs in the residuals are used to prune the candidate space when performing the fault diagnosis task. The method is illustrated using a two-tank example, in which these aspects are shown to have an impact on the diagnosis and fault discrimination, since the proposed method goes beyond the structural methods

Developing methodologies to allow the robust prediction dynamic context dominated by perturbations

El principal objectiu del projecte era desenvolupar millores conceptuals i metodològiques que permetessin una millor predicció dels canvis en la distribució de les espècies (a una escala de paisatge) derivats de canvis ambientals en un context dominat per pertorbacions. En un primer estudi, vàrem comparar l'eficàcia de diferents models dinàmics per a predir la distribució de l'hortolà (Emberiza hortulana). Els nostres resultats indiquen que un model híbrid que combini canvis en la qualitat de l'hàbitat, derivats de canvis en el paisatge, amb un model poblacional espacialment explícit és una aproximació adequada per abordar canvis en la distribució d'espècies en contextos de dinàmica ambiental elevada i una capacitat de dispersió limitada de l'espècie objectiu. En un segon estudi abordarem la calibració mitjançant dades de seguiment de models de distribució dinàmics per a 12 espècies amb preferència per hàbitats oberts. Entre les conclusions extretes destaquem: (1) la necessitat de que les dades de seguiment abarquin aquelles àrees on es produeixen els canvis de qualitat; (2) el biaix que es produeix en la estimació dels paràmetres del model d'ocupació quan la hipòtesi de canvi de paisatge o el model de qualitat d'hàbitat són incorrectes. En el darrer treball estudiarem el possible impacte en 67 espècies d’ocells de diferents règims d’incendis, definits a partir de combinacions de nivells de canvi climàtic (portant a un augment esperat de la mida i freqüència d’incendis forestals), i eficiència d’extinció per part dels bombers. Segons els resultats dels nostres models, la combinació de factors antropogènics del regim d’incendis, tals com l’abandonament rural i l’extinció, poden ser més determinants per als canvis de distribució que els efectes derivats del canvi climàtic. Els productes generats inclouen tres publicacions científiques, una pàgina web amb resultats del projecte i una llibreria per a l'entorn estadístic R.

SGP-1: prediction and validation of homologous genes based on sequence alignments

Conventional methods of gene prediction rely on the recognition of DNA-sequence signals, the coding potential or the comparison of a genomic sequence with a cDNA, EST, or protein database. Reasons for limited accuracy in many circumstances are species-specific training and the incompleteness of reference databases. Lately, comparative genome analysis has attracted increasing attention. Several analysis tools that are based on human/mouse comparisons are already available. Here, we present a program for the prediction of protein-coding genes, termed SGP-1 (Syntenic Gene Prediction), which is based on the similarity of homologous genomic sequences. In contrast to most existing tools, the accuracy of SGP-1 depends little on species-specific properties such as codon usage or the nucleotide distribution. SGP-1 may therefore be applied to nonstandard model organisms in vertebrates as well as in plants, without the need for extensive parameter training. In addition to predicting genes in large-scale genomic sequences, the program may be useful to validate gene structure annotations from databases. To this end, SGP-1 output also contains comparisons between predicted and annotated gene structures in HTML format. The program can be accessed via a Web server at http://soft.ice.mpg.de/sgp-1. The source code, written in ANSI C, is available on request from the authors.

An assessment of gene prediction accuracy in large DNA sequences

One of the first useful products from the human genome will be a set of predicted genes. Besides its intrinsic scientific interest, the accuracy and completeness of this data set is of considerable importance for human health and medicine. Though progress has been made on computational gene identification in terms of both methods and accuracy evaluation measures, most of the sequence sets in which the programs are tested are short genomic sequences, and there is concern that these accuracy measures may not extrapolate well to larger, more challenging data sets. Given the absence of experimentally verified large genomic data sets, we constructed a semiartificial test set comprising a number of short single-gene genomic sequences with randomly generated intergenic regions. This test set, which should still present an easier problem than real human genomic sequence, mimics the approximately 200kb long BACs being sequenced. In our experiments with these longer genomic sequences, the accuracy of GENSCAN, one of the most accurate ab initio gene prediction programs, dropped significantly, although its sensitivity remained high. Conversely, the accuracy of similarity-based programs, such as GENEWISE, PROCRUSTES, and BLASTX was not affected significantly by the presence of random intergenic sequence, but depended on the strength of the similarity to the protein homolog. As expected, the accuracy dropped if the models were built using more distant homologs, and we were able to quantitatively estimate this decline. However, the specificities of these techniques are still rather good even when the similarity is weak, which is a desirable characteristic for driving expensive follow-up experiments. Our experiments suggest that though gene prediction will improve with every new protein that is discovered and through improvements in the current set of tools, we still have a long way to go before we can decipher the precise exonic structure of every gene in the human genome using purely computational methodology.

Comparative gene prediction in human and mouse

The completion of the sequencing of the mouse genome promises to help predict human genes with greater accuracy. While current ab initio gene prediction programs are remarkably sensitive (i.e., they predict at least a fragment of most genes), their specificity is often low, predicting a large number of false-positive genes in the human genome. Sequence conservation at the protein level with the mouse genome can help eliminate some of those false positives. Here we describe SGP2, a gene prediction program that combines ab initio gene prediction with TBLASTX searches between two genome sequences to provide both sensitive and specific gene predictions. The accuracy of SGP2 when used to predict genes by comparing the human and mouse genomes is assessed on a number of data sets, including single-gene data sets, the highly curated human chromosome 22 predictions, and entire genome predictions from ENSEMBL. Results indicate that SGP2 outperforms purely ab initio gene prediction methods. Results also indicate that SGP2 works about as well with 3x shotgun data as it does with fully assembled genomes. SGP2 provides a high enough specificity that its predictions can be experimentally verified at a reasonable cost. SGP2 was used to generate a complete set of gene predictions on both the human and mouse by comparing the genomes of these two species. Our results suggest that another few thousand human and mouse genes currently not in ENSEMBL are worth verifying experimentally.

Splitting statistical potentials into meaningful scoring functions: testing the prediction of near-native structures from decoy conformations

Background: Recent advances on high-throughput technologies have produced a vast amount of protein sequences, while the number of high-resolution structures has seen a limited increase. This has impelled the production of many strategies to built protein structures from its sequence, generating a considerable amount of alternative models. The selection of the closest model to the native conformation has thus become crucial for structure prediction. Several methods have been developed to score protein models by energies, knowledge-based potentials and combination of both.Results: Here, we present and demonstrate a theory to split the knowledge-based potentials in scoring terms biologically meaningful and to combine them in new scores to predict near-native structures. Our strategy allows circumventing the problem of defining the reference state. In this approach we give the proof for a simple and linear application that can be further improved by optimizing the combination of Zscores. Using the simplest composite score () we obtained predictions similar to state-of-the-art methods. Besides, our approach has the advantage of identifying the most relevant terms involved in the stability of the protein structure. Finally, we also use the composite Zscores to assess the conformation of models and to detect local errors.Conclusion: We have introduced a method to split knowledge-based potentials and to solve the problem of defining a reference state. The new scores have detected near-native structures as accurately as state-of-art methods and have been successful to identify wrongly modeled regions of many near-native conformations.

Prediction of enzyme function by combining sequence similarity and protein interactions

Background: A number of studies have used protein interaction data alone for protein function prediction. Here, we introduce a computational approach for annotation of enzymes, based on the observation that similar protein sequences are more likely to perform the same function if they share similar interacting partners. Results: The method has been tested against the PSI-BLAST program using a set of 3,890 protein sequences from which interaction data was available. For protein sequences that align with at least 40% sequence identity to a known enzyme, the specificity of our method in predicting the first three EC digits increased from 80% to 90% at 80% coverage when compared to PSI-BLAST. Conclusion: Our method can also be used in proteins for which homologous sequences with known interacting partners can be detected. Thus, our method could increase 10% the specificity of genome-wide enzyme predictions based on sequence matching by PSI-BLAST alone.

Strategies for sequential prediction of stationary time series

We present simple procedures for the prediction of a real valued sequence. The algorithms are based on a combinationof several simple predictors. We show that if the sequence is a realization of a bounded stationary and ergodic random process then the average of squared errors converges, almost surely, to that of the optimum, given by the Bayes predictor. We offer an analog result for the prediction of stationary gaussian processes.

Efficient interval scoring rules

Scoring rules that elicit an entire belief distribution through the elicitation of point beliefsare time-consuming and demand considerable cognitive e¤ort. Moreover, the results are validonly when agents are risk-neutral or when one uses probabilistic rules. We investigate a classof rules in which the agent has to choose an interval and is rewarded (deterministically) onthe basis of the chosen interval and the realization of the random variable. We formulatean e¢ ciency criterion for such rules and present a speci.c interval scoring rule. For single-peaked beliefs, our rule gives information about both the location and the dispersion of thebelief distribution. These results hold for all concave utility functions.

On prediction of individual sequences

Sequential randomized prediction of an arbitrary binary sequence isinvestigated. No assumption is made on the mechanism of generating the bit sequence. The goal of the predictor is to minimize its relative loss, i.e., to make (almost) as few mistakes as the best ``expert'' in a fixed, possibly infinite, set of experts. We point out a surprising connection between this prediction problem and empirical process theory. First, in the special case of static (memoryless) experts, we completely characterize the minimax relative loss in terms of the maximum of an associated Rademacher process. Then we show general upper and lower bounds on the minimaxrelative loss in terms of the geometry of the class of experts. As main examples, we determine the exact order of magnitude of the minimax relative loss for the class of autoregressive linear predictors and for the class of Markov experts.