Lucia Sacchi

Data mining with Temporal Abstractions: learning rules from time series

A large volume of research in temporal data mining is focusing on discovering temporal rules from time-stamped data. The majority of the methods proposed so far have been mainly devoted to the mining of temporal rules which describe relationships between data sequences or instantaneous events and do not consider the presence of complex temporal patterns into the dataset. Such complex patterns, such as trends or up and down behaviors, are often very interesting for the users. In this paper we propose a new kind of temporal association rule and the related extraction algorithm; the learned rules involve complex temporal patterns in both their antecedent and consequent. Within our proposed approach, the user defines a set of complex patterns of interest that constitute the basis for the construction of the temporal rule; such complex patterns are represented and retrieved in the data through the formalism of knowledge-based Temporal Abstractions. An Apriori-like algorithm looks then for meaningful temporal relationships (in particular, precedence temporal relationships) among the complex patterns of interest. The paper presents the results obtained by the rule extraction algorithm on a simulated dataset and on two different datasets related to biomedical applications: the first one concerns the analysis of time series coming from the monitoring of different clinical variables during hemodialysis sessions, while the other one deals with the biological problem of inferring relationships between genes from DNA microarray data.

Predictive value of baseline serum vascular endothelial growth factor and neutrophil gelatinase-associated lipocalin in advanced kidney cancer patients receiving sunitinib

To identify factors that might predict response to sunitinib in patients with renal cell carcinoma, we measured serum vascular endothelial growth factor (VEGF) and neutrophil gelatinase-associated lipocalin (NGAL) levels. A total of 85 patients were selected and, using the Motzer classification, 46 were assigned to the good- and 38 to the intermediate-risk groups. With univariate Cox analysis, both baseline serum VEGF and NGAL titers, determined by enzyme-linked immunosorbent assay, significantly predicted progression-free survival. For each biomarker, a threshold value was identified, which proved useful to classify patients into groups having titers above or below the thresholds. We then stratified patients according to the two dichotomous variables into good-, intermediate-, and poor-risk groups, and found significantly different progression-free survival rates ranging from 3.5 to 11.6 months. Both VEGF and NGAL maintained their predictive significance at bivariate analysis. Our study shows that serum levels of VEGF and NGAL are significant predictors of progression-free survival in patients with renal cell carcinoma treated with sunitinib.

Maternal Oct-4 is a potential key regulator of the developmental competence of mouse oocytes

BackgroundThe maternal contribution of transcripts and proteins supplied to the zygote is crucial for the progression from a gametic to an embryonic control of preimplantation development. Here we compared the transcriptional profiles of two types of mouse MII oocytes, one which is developmentally competent (MIISN oocyte), the other that ceases development at the 2-cell stage (MIINSN oocyte), with the aim of identifying genes and gene expression networks whose misregulated expression would contribute to a reduced developmental competence.ResultsWe report that: 1) the transcription factor Oct-4 is absent in MIINSN oocytes, accounting for 2) the down-regulation of Stella, a maternal-effect factor required for the oocyte-to-embryo transition and of which Oct-4 is a positive regulator; 3) eighteen Oct-4-regulated genes are up-regulated in MIINSN oocytes and are part of gene expression networks implicated in the activation of adverse biochemical pathways such as oxidative phosphorylation, mitochondrial dysfunction and apoptosis.ConclusionThe down-regulation of Oct-4 plays a crucial function in a sequence of molecular processes that leads to the developmental arrest of MIINSN oocytes. The use of a model study in which the MII oocyte ceases development consistently at the 2-cell stage has allowed to attribute a role to the maternal Oct-4 that has never been described before. Oct-4 emerges as a key regulator of the molecular events that govern the establishment of the developmental competence of mouse oocytes.

Changes in circulating pro-angiogenic cytokines, other than VEGF, before progression to sunitinib therapy in advanced renal cell carcinoma patients.

Objectives: This study included a cohort of advanced renal cell carcinoma patients treated with sunitinib. Since resistance to sunitinib may be mediated through angiogenic cytokines other than VEGF, we measured the circulating levels of three pro-angiogenic cytokines: basic fibroblast growth factor (bFGF), hepatocyte growth factor (HGF), and interleukin (IL)-6. Methods: Cytokines were measured at baseline and on the first day of each treatment cycle until progression in 85 advanced kidney cancer patients treated with sunitinib using a quantitative sandwich enzyme immunoassay (ELISA) technique. Results: Even though no statistically significant differences in the titers of the three cytokines were observed between baseline and the time of progression in the whole patient cohort, in 45.3, 46.6, and 37.3% of the patients a more than 50% increase between baseline and the time of progression was shown in circulating IL-6, bFGF, and HGF, respectively. Furthermore, this increase was more than 100% in 37.3, 44, and 30.6% of the patients, respectively. We also demonstrated that, in these patients, cytokines tended to increase and to remain high immediately before progression. Conclusions: In a large percentage of kidney cancer patients, progression is preceded by a significant increase in pro-angiogenic cytokines other than VEGF.

TimeClust: a clustering tool for gene expression time series

UNLABELLED TimeClust is a user-friendly software package to cluster genes according to their temporal expression profiles. It can be conveniently used to analyze data obtained from DNA microarray time-course experiments. It implements two original algorithms specifically designed for clustering short time series together with hierarchical clustering and self-organizing maps. AVAILABILITY TimeClust executable files for Windows and LINUX platforms can be downloaded free of charge for non-profit institutions from the following web site: http://aimed11.unipv.it/TimeClust.

Temporal abstraction for feature extraction: A comparative case study in prediction from intensive care monitoring data

OBJECTIVES To compare two temporal abstraction procedures for the extraction of meta features from monitoring data. Feature extraction prior to predictive modeling is a common strategy in prediction from temporal data. A fundamental dilemma in this strategy, however, is the extent to which the extraction should be guided by domain knowledge, and to which extent it should be guided by the available data. The two temporal abstraction procedures compared in this case study differ in this respect. METHODS AND MATERIAL The first temporal abstraction procedure derives symbolic descriptions from the data that are predefined using existing concepts from the medical language. In the second procedure, a large space of numerical meta features is searched through to discover relevant features from the data. These procedures were applied to a prediction problem from intensive care monitoring data. The predictive value of the resulting meta features were compared, and based on each type of features, a class probability tree model was developed. RESULTS The numerical meta features extracted by the second procedure were found to be more informative than the symbolic meta features of the first procedure in the case study, and a superior predictive performance was observed for the associated tree model. CONCLUSION The findings indicate that for prediction from monitoring data, induction of numerical meta features from data is preferable to extraction of symbolic meta features using existing clinical concepts.

Gatekeeper of pluripotency: A common Oct4 transcriptional network operates in mouse eggs and embryonic stem cells

BackgroundOct4 is a key factor of an expanded transcriptional network (Oct4-TN) that governs pluripotency and self-renewal in embryonic stem cells (ESCs) and in the inner cell mass from which ESCs are derived. A pending question is whether the establishment of the Oct4-TN initiates during oogenesis or after fertilisation. To this regard, recent evidence has shown that Oct4 controls a poorly known Oct4-TN central to the acquisition of the mouse egg developmental competence. The aim of this study was to investigate the identity and extension of this maternal Oct4-TN, as much as whether its presence is circumscribed to the egg or maintained beyond fertilisation.ResultsBy comparing the genome-wide transcriptional profile of developmentally competent eggs that express the OCT4 protein to that of developmentally incompetent eggs in which OCT4 is down-regulated, we unveiled a maternal Oct4-TN of 182 genes. Eighty of these transcripts escape post-fertilisation degradation and represent the maternal Oct4-TN inheritance that is passed on to the 2-cell embryo. Most of these 80 genes are expressed in cancer cells and 37 are notable companions of the Oct4 transcriptome in ESCs.ConclusionsThese results provide, for the first time, a developmental link between eggs, early preimplantation embryos and ESCs, indicating that the molecular signature that characterises the ESCs identity is rooted in oogenesis. Also, they contribute a useful resource to further study the mechanisms of Oct4 function and regulation during the maternal-to-embryo transition and to explore the link between the regulation of pluripotency and the acquisition of de-differentiation in cancer cells.

Mining Healthcare Data with Temporal Association Rules: Improvements and Assessment for a Practical Use

The Regional Healthcare Agency (ASL) of Pavia has been maintaining a central data repository which stores healthcare data about the population of Pavia area. The analysis of such data can be fruitful for the assessment of healthcare activities. Given the crucial role of time in such databases, we developed a general methodology for the mining of Temporal Association Rules on sequences of hybrid events. In this paper we show how the method can be extended to suitably manage the integration of both clinical and administrative data. Moreover, we address the problem of developing an automated strategy for the filtering of output rules, exploiting the taxonomy underlying the drug coding system and considering the relationships between clinical variables and drug effects. The results show that the method could find a practical use for the evaluation of the pertinence of the care delivery flow for specific pathologies.

Mining Health Care Administrative Data with Temporal Association Rules on Hybrid Events

OBJECTIVE The analysis of administrative health care data can be helpful to conveniently assess health care activities. In this context temporal data mining techniques can be suitably exploited to get a deeper insight into the processes underlying health care delivery. In this paper we present an algorithm for the extraction of temporal association rules (TARs) on sequences of hybrid events and its application on health care administrative databases. METHODS We propose a method that extends TAR mining by managing hybrid events, namely events characterized by a heterogeneous temporal nature. Hybrid events include both point-like events (e.g. ambulatory visits) and interval-like events (e.g. drug consumption). The definition of user-defined rule templates can be optionally used to constrain the search only to the extraction of a subset of interesting rules. A TAR post-pruning strategy, based on a case-control approach, is also presented. RESULTS We analyzed the administrative database of diabetic patients in charge to the regional health care agency (ASL) of Pavia. TAR mining allowed to find patterns specifically related to the diabetic population in comparison with a control group, as well as to check the compliance of the actual clinical careflow with the ASL recommendations. CONCLUSION The experimental results highlighted the main potentials of the algorithm, such as the opportunity to detect interesting temporal relationships between diagnostic or therapeutic patterns, or to check the adherence of past temporal behaviors to specific expected paths (e.g. guidelines) or to discover new knowledge that could be implicitly hidden in the data.

TA-clustering: Cluster analysis of gene expression profiles through Temporal Abstractions

This paper describes a new technique for clustering short time series of gene expression data. The technique is a generalization of the template-based clustering and is based on a qualitative representation of profiles which are labelled using trend Temporal Abstractions (TAs); clusters are then dynamically identified on the basis of this qualitative representation. Clustering is performed in an efficient way at three different levels of aggregation of qualitative labels, each level corresponding to a distinct degree of qualitative representation. The developed TA-clustering algorithm provides an innovative way to cluster gene profiles. We show the developed method to be robust, efficient and to perform better than the standard hierarchical agglomerative clustering approach when dealing with temporal dislocations of time series. Results of the TA-clustering algorithm can be visualized as a three-level hierarchical tree of qualitative representations and as such easy to interpret. We demonstrate the utility of the proposed algorithm on a set of two simulated data sets and on a study of gene expression data from S. cerevisiae.