An efficient pattern mining approach for event detection in multivariate temporal data

Abstract

This work proposes a pattern mining approach to learn event detection models from complex multivariate temporal data, such as electronic health records. We present Recent Temporal Pattern mining, a novel approach for efficiently finding predictive patterns for event detection problems. This approach first converts the time series data into time-interval sequences of temporal abstractions. It then constructs more complex time-interval patterns backward in time using temporal operators. We also present the Minimal Predictive Recent Temporal Patterns framework for selecting a small set of predictive and non-spurious patterns. We apply our methods for predicting adverse medical events in real-world clinical data. The results demonstrate the benefits of our methods in learning accurate event detection models, which is a key step for developing intelligent patient monitoring and decision support systems.

Mining and exploring care pathways from electronic medical records with visual analytics

“…In real-world EMR systems, duration information is often not captured, so we choose to use techniques that do not require this information. However, it is also possible to abstract point-based data by applying temporal knowledge which results in a more abstract representation of the data, in the form of symbolic time intervals Batal et al provide several pattern mining techniques that uses a time interval-related representation of a sequence, which requires either the events have continuous values that can be quantized or the duration of every event is available [36,42]. Moskovitch et al provide several approaches for discretizing continuous event values to derive more discriminative time-interval related patterns [40,41].…”

Extended vertical lists for temporal pattern mining from multivariate time series

“…An idea of assigning to each FTP a list of record identifiers that contain it: P.ids = {i ∶ P ∈ Z i }, reduces the search space drastically (Batal et al, 2016). It is based on the vertical data format (Zaki, 2000(Zaki, , 2001.…”

“…For other concepts of TP like recent TP (RTP) in Batal et al (2016), this assumption does not hold. Thus, the candidate elimination phase will not work here, and only less efficient techniques like the vertical data format should be utilized instead.…”

“…Continuously expanding resources for computing, data storage, and transmission have enabled pattern mining in complex data sets emerging from various domains such as transaction databases (Agrawal, Imielinski, & Swami, ; Agrawal, Imieliński, & Swami, ), web mining (Srivastava, Cooley, Deshpande, & Tan, ), Internet of Things (Chen et al, ; Tsai, Lai, Chiang, & Yang, ), medicine (Batal et al, ; Hauskrecht et al, ), fraud detection (Seeja & Zareapoor, ), finance (Tiple, Cavique, & Cavalheiro Marques, ), and so on. This paper demonstrates the problem of extracting temporal patterns (TPs) from multivariate time series records.…”

RobustSPAM for Inference from Noisy Longitudinal Data and Preservation of Privacy

“…Electronic Health Records, machine log files, credit/debit card use). The increase in the number of complex temporal datasets has prompted the development of methods that extend applicability of classical statistical, machine learning and data mining methods to those datasets [1]- [3]. This is particularly important in monitoring or detection problems such as: patient monitoring [4] or fraud detection [5].…”