Linguistic summaries of categorical time series for septic shock patient data

Abstract

Linguistic summarization is a data mining and knowledge discovery approach to extract patterns and sum up large volume of data into simple sentences. There is a large research in generating linguistic summaries which can be used to better understand and communicate about patterns, evolution and long trends in numerical, time series or labelled data. The objective of this work is to develop a computational system capable of automatically generating linguistic descriptions of time series data of septic shock patients containing labelled data, not only of the whole series, but also on the differences between subsets of the data. This is of particular interest in septic shock, as the differences between patients are not well understood. For this purpose we propose a new type of differential summaries, based on a numerical criterion assessing the characteristics of the summary on each subset of interest. Furthermore, this paper proposes an extension of linguistic summaries to provide temporal and categorical contextualization. This is of particular interest in healthcare to detect differences related to a condition or illness as well as the effectiveness of the administered treatment.

Interpretability of fuzzy linguistic summaries

“…For instance, if the sentence Z 3 denotes "Most old people are tall", it is not in contradiction with Z since there is no particular constraint on the fact that "old people" must not be as tall as "young people" are. In the case where one of the two summaries holds and not the other, a differential summary can be established (see [15] and paragraph 3.1.1).…”

“…In [15], the protoform "QBy's are A Q τ time" is proposed, illustrated by a sentence such as "few patients have a medium value of heart rate most of the time". In order to evaluate this protoform, the values of each time series for different time stamps are first summarised by the protoform "Q τ time By's are A ", where Q τ is a quantifier applied to the time attribute.…”

“…The evaluation of "Q τ time QBy's are A" [15] is also a two-step process where the set of time series is first summarised at each time stamp and then summarised across all time stamps.…”

“…of the form "QBy's are A Q τ time", measure the summary quality as a twolevel aggregation [15]: each series x = (x t ) t∈T , T being the universe of time stamps, is individually summarised as "Q τ By's are A" using the temporal quantifier Q τ applied to the data time stamps t; at a second level, a classic protoform lying on the quantifier Q is applied to all series that make up the whole data set. Thus, each time stamp of each time series is first evaluated:…”

“…The differentiation score [15] aims at favouring sentences that highlight differences between subgroups of data. Its particularity is that its evaluation does not only depend on the sentence to be assessed but also on other similar sentences, requiring the definition of a similarity measure.…”