Early detection of prediabetes is crucial to preventing its progression to diabetes. Providing individuals with a personalized sense of their risk could improve prevention efforts. While complex mathematical models that simulate metabolic and inflammatory processes offer detailed and patient-specific insights, their computational cost usually makes them impractical for real-time prediction on mobile platforms.This work introduces a long short-term memory (LSTM) surrogate for the MT2D model, that simulates the main metabolic and inflammatory processes undergoing the transition to prediabetes. The model is developed using a dataset of 43 669 simulated subjects, each with lifestyle inputs and biomarker outputs over six months. Using 8 time series inputs, the surrogate predicts the dynamics of 11 key metabolic and inflammatory outputs, closely replicating the behaviour of the MT2D model.After training, the proposed LSTM model reduces computational time from an average of 8.4 hours to 0.1 seconds per simulation, making it suitable for mobile device deployment. The model achieves root mean squared errors on the order of 10−2on scaled data, and shows promise for prediabetes risk assessment by capturing trends in inflammatory biomarkers.This surrogate model can provide real-time and patient-specific insights into the metabolic health, potentially improving the understanding of prediabetes risk.

Joseph Cornelius, Oscar Lithgow-Serrano, Sandra Mitrovic, Ljiljana Dolamic, Fabio Rinaldi. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

Novel joint techniques capture both the microscopic context and the mesoscopic structure of networks by leveraging two previously separated fields of research: node representation learning (NRL) and community detection (CD). However, several limitations exist in the literature. First, a comprehensive comparison between these joint NRL-CD techniques is non-existent. Second, baseline techniques, datasets, evaluation metrics, and classification algorithms differ significantly between each method. Thirdly, the literature lacks a synchronized experimental approach, thus rendering comparison between these methods strenuous. To overcome these limitations, we present a uni-fied experimental setup mutually comparing six joint NRL-CD techniques and comparing them with corresponding NRL/CD baselines in three different settings: non-overlapping and over-lapping CD and node classification. Our results show that joint methods underperform on the node classification task but achieve relatively solid results for overlapping community detection. Our research contribution is two-fold: first, we show specific weaknesses of selected joint techniques in different tasks and data sets; and second, we suggest a more thorough experimental setup to benchmark joint techniques with simpler NRL and CD techniques.

In many real-world networks, it is important to explicitly differentiate between positive and negative links, thus considering the observed networks as signed. To derive useful features, just as in the case of unsigned networks, representation learning can be used to learn meaningful representations of a network that characterize its underlying topology. Several methods for learning representations on signed networks have already been proposed but have not been systematically benchmarked together before. Hence, in this paper, we bridge this literature gap providing a quantitative and qualitative benchmark of the four most prominent representation learning methods for signed networks. Results on three different datasets for link sign prediction showcase the superiority of the StEM method over its competitors both from a predictive performance and runtime perspective.

In many real-life applications it is crucial to be able to, given a collection of link states of a network in a certain time period, accurately predict the link state of the network at a future time. This is known as dynamic link prediction, which compared to its static counterpart is more complex, as capturing the temporal characteristics is a non-trivial task. This explains while still majority of today's research in network representation learning focuses on static setting ignoring temporal information. In this work, we focus on one such case and aim at extending node2vec, representation learning method successfully applied for static link prediction, to a dynamic setup. This extended method is applied and validated on several real-life networks with different properties. Results show that taking into account dynamic aspect outperforms static approach. Additionally, based on the network properties, recommendations are given for the node2vec parameters.

Most of the recent studies on churn prediction in telco utilize social networks built on top of the call (and/or SMS) graphs to derive informative features. However, extracting features from large graphs, especially structural features, is an intricate process both from a methodological and computational perspective. Due to the former, feature extraction in the current literature has mainly been addressed in an ad-hoc and hand-crafted manner. Due to the latter, the full potential of the structural information is unexploited. In this work, we incorporate both interaction and structural information by devising two different ways of enriching original graphs with interaction information, delineated by the well-known RFM model. We circumvent the process of extensive manual feature engineering by enriching the networks and improving the scalability of the renowned node2vec approach to learn node representations. The obtained results demonstrate that our enriched network outperforms baseline RFM-based methods.