Distributed Tensor Decomposition for Large Scale Health Analytics

Abstract

In the past few decades, there has been rapid growth in quantity and variety of healthcare data. These large sets of data are usually high dimensional (e.g. patients, their diagnoses, and medications to treat their diagnoses) and cannot be adequately represented as matrices. Thus, many existing algorithms can not analyze them. To accommodate these high dimensional data, tensor factorization, which can be viewed as a higher-order extension of methods like PCA, has attracted much attention and emerged as a promising solution. However, tensor factorization is a computationally expensive task, and existing methods developed to factor large tensors are not flexible enough for real-world situations. To address this scaling problem more efficiently, we introduce , a distributed, scalable, and sparse tensor factorization method fit through stochastic gradient descent. offers three contributions: (1) Scalability: it employs a block partitioning and parallel processing design and thus scales to large tensors, (2) Accuracy: we show that our method can achieve results faster without sacrificing the quality of the tensor decomposition, and (3) FlexibleConstraints: we show our approach can encompass various kinds of constraints including l2 norm, l1 norm, and logistic regularization. We demonstrate ’s capabilities in two real-world use cases. In the first, we use Google searches for flu-like symptoms to characterize and predict influenza patterns. In the second, we use to extract clinically interesting sets (i.e., phenotypes) of patients from electronic health records. Through these case studies, we show has the potential to be used to rapidly characterize, predict, and manage a large multimodal datasets, thereby promising a novel, data-driven solution that can benefit very large segments of the population.

Deep representation learning of patient data from Electronic Health Records (EHR): A systematic review

“…These advantages have made tensor decomposition a promising modeling approach to learn abstract patient representations from EHR data and provide good interpretability and scalability [42]. A tensor-based patient representation allows for the capture of complex interactions and relationships between clinical events (especially phenotypes, comorbidities, and medications) that are not evident in flattened EHR data [43].…”

A Decoupled Blockchain Approach for Edge-Envisioned IoT-Based Healthcare Monitoring

“…To reduce the storage resources required by this data, various authors have used tensor-based data decomposition and storage techniques. For instance, He et al [20] proposed a distributed, scalable, and sparse tensor factorization method to provide scalability and accuracy while performing healthcare analytics. Furthermore, Sandhu et al [21] showed the applicability of tensor-based data representation and storage approach on the healthcare diabetes data.…”

Efficient, Out-of-Memory Sparse MTTKRP on Massively Parallel Architectures

“…Tensors are higher-order generalization of matrices, and they provide a natural abstraction for complex and inter-related data. Many critical applications, such as data mining [25,38], social network analytics [14,41], cybersecurity [9,13], and healthcare [16,18], generate massive amounts of multidimensional (multi-modal) data as sparse tensors that can be analyzed quickly and efficiently using tensor decomposition (TD). The most popular TD method is the canonical polyadic decomposition (CPD), which approximates a tensor as a sum of a finite number of rank-one tensors such that each rank-one tensor corresponds to a useful data property [5,24].…”