Statistical Link Prediction for Temporal Networks with Epilepsy MEG Data
Poster No:
1501
Submission Type:
Abstract Submission
Authors:
Jaehee Kim1, Heaji Lee1
Institutions:
1Duksung Women's University, Seoul, Seoul
First Author:
Co-Author:
Introduction:
Networks have become a fundamental approach to understanding systems of interacting objects, unifying the study of diverse phenomena, including biological organisms and human brain systems. With the rise of large-scale temporal networks, such as social and neuroimaging networks, temporal link prediction has become an interesting issue. Brain neuroimaging network data are dynamic or temporal networks where entities and relationships appear, disappear, strengthen, and weaken over time. Nodes and their relationships represent links and entities. Each link contains information on the time when it is active and other possible characteristics. The evolutionary behavior of temporal networks got the attention of interest. As adding or removing new links or edges over time leads to network evolution, learning the evolutionary behavior of networks is directly related to the link prediction problem.
Different types of edges represent different relationships, including connectivity and similarity among nodes. Adopting a specific definition of edges can fundamentally alter how we analyze and interpret a brain network. Temporal link prediction (TLP) is a classic yet challenging inference task on dynamic graphs, which predicts possible future linkage based on historical topology. The predicted future topology can be used to support some advanced applications on real-world systems (e.g., resource pre-allocation) for better system performance or information. It aims to predict possible linkages in specific future time steps based on the observed historical topology, essential in revealing brain systems' dynamic nature.
Different types of edges represent different relationships, including connectivity and similarity among nodes. Adopting a specific definition of edges can fundamentally alter how we analyze and interpret a brain network. Temporal link prediction (TLP) is a classic yet challenging inference task on dynamic graphs, which predicts possible future linkage based on historical topology. The predicted future topology can be used to support some advanced applications on real-world systems (e.g., resource pre-allocation) for better system performance or information. It aims to predict possible linkages in specific future time steps based on the observed historical topology, essential in revealing brain systems' dynamic nature.
Methods:
We develop a link prediction method based on a stochastic block model as a probabilistic approach and another method based on the exponentially weighted moving average for node centrality as a time series approach combined with the neighbor similarity. We apply the proposed methods to MEG data for temporal link prediction. Link prediction aims to estimate the evolving likelihood of the existence of a link in a given network based on observed information in the network's evolving process.
We apply our method to the epilepsy patients' magnetoencephalography (MEG) data on 72 ROIs collected at Seoul National University Hospital (SNUH). The dataset contained MEG data from 44 mesial temporal lobe epilepsy (mTLE) with hippocampal sclerosis (HS) patients who underwent epilepsy surgery between 2005 and 2011.
Temporal link predictions are made and compared with the similarity-based methods such as CN (common neighbor), AA (Adamic-Adar), Jacard index and PA (preferential attachment).
We apply our method to the epilepsy patients' magnetoencephalography (MEG) data on 72 ROIs collected at Seoul National University Hospital (SNUH). The dataset contained MEG data from 44 mesial temporal lobe epilepsy (mTLE) with hippocampal sclerosis (HS) patients who underwent epilepsy surgery between 2005 and 2011.
Temporal link predictions are made and compared with the similarity-based methods such as CN (common neighbor), AA (Adamic-Adar), Jacard index and PA (preferential attachment).
Results:
Static and temporal link prediction is done with epilepsy MEG data. We compare link prediction for the temporal MEG network data with the proposed methods. Table 1 shows the comparison results (omitted), for example. Figure 1 reveals the link prediction networks at the next time-point via the proposed method using node centrality and common neighbors for exemplary subject in each group. We further demonstrate the utility of our proposed methodology via simulation studies. Our methods perform reasonably well for the MEG network link prediction.
·Figure 1. Observed and Predicted Network in LT, RT and HC
Conclusions:
Networks have become increasingly important to model complex brain systems with interacting elements such as links. Link prediction aims to infer the behavior of the network link formation process by predicting missed or future relationships based on currently observed connections. It has become an attractive study area since it allows us to predict how networks will evolve. Network neuroscience explorations can benefit from edge-centric perspectives.
Modeling and Analysis Methods:
Connectivity (eg. functional, effective, structural) 1
Novel Imaging Acquisition Methods:
MEG 2
Keywords:
Epilepsy
MEG
1|2Indicates the priority used for review
Provide references using author date format
Bullmore, E. & Sporns, O. (2009) Complex brain networks: graph theoretical analysis of structural and functional systems. Nature reviews
Divakaran, A., & Mohan, A. (2020) Temporal link prediction: A survey. New Generation Computing, 38, 213-258.
Katz, L. (1953) A new status index derived from sociometric analysis. Psychometrika 18(1), 39–43.
Kossinets, G. (2006) Effects of missing data in social networks. Soc. Netw. 28(3), 247–268.
Liu, W., Lü, L. (2010) Link prediction based on local random walk. EPL (Europhysics Letters) 89(5), 58007.
Lü, L., Jin, C.H., Zhou, T. (2009) Similarity index based on local paths for link prediction of complex networks. Phys. Rev. E 80(4), 046122.
Lü, L., & Zhou, T. (2011). Link prediction in complex networks: A survey. Physica A: statistical mechanics and its applications, 390(6), 1150-1170.
Thompson, W. H., Brantefors, P. & Fransson, P. (2017). From static to temporal network theory: Applications to functional brain connectivity. Network Neuroscience, 1(2), 69-99.
Zhu, B., & Xia, Y. (2015) An information-theoretic model for link prediction in complex networks. Scientific reports, 5(1), 13707.