MIDAS Seminar Series Presents: Luonan Chen, PhD – Chinese Academy of Sciences

Luonan Chen, PhD

Professor

Executive Director, Shanghai Institutes for Biological Sciences

Chinese Academy of Sciences

Criticality detection and time-series prediction from small samples by dynamics-based model-free approaches

Abstract: In this talk, Dr. Chen will present two works for small-sample data analysis by dynamics-based model-free approaches, i.e. (1) detecting the tipping points of dynamic processes by DNB [1][2][4], and (2) predicting future dynamics of short-term time-series data by RDE [3].

(1) Considerable evidence suggests that during the progression of complex diseases, the deteriorations are not necessarily smooth but are abrupt, and may cause a critical transition from one state to another at a tipping point. Here, we develop a model-free method to detect early-warning signals of such critical transitions (un-occurred diseases), even with only a small number of samples. Specifically, we theoretically derive an index based on a dynamic network biomarker (DNB) that serves as a general early-warning signal indicating an imminent sudden deterioration before the critical transition occurs [1][2][4]. Based on theoretical analyses, we show that predicting a sudden transition from small samples is achievable provided that there are a large number of measurements for each sample, e.g., high-throughput data. We employ gene expression data of three diseases to demonstrate the effectiveness of our method. The relevance of DNBs with the diseases was also validated by related experimental data (e.g., liver cancer, lung injury, influenza, type-2 diabetes) and functional analysis. DNB can also be used for the analysis of nonlinear biological processes, e.g., cell differentiation process [1][2].

(2) Future state prediction for nonlinear dynamical systems is a challenging task, particularly when only a few time-series samples for high-dimensional variables are available from real-world systems. In this work, we propose a novel model-free framework, named Randomly Distributed Embedding (RDE), to achieve accurate future state prediction based on short-term, high-dimensional data [3]. Specifically, from the observed data of high-dimensional variables, the RDE framework randomly generates a sufficient number of low dimensional “non-delay embeddings”, and maps each of them to a “delay embedding” which is constructed from the data of a to-bepredicted target variable. Any of these mappings can perform as a low-dimensional weak predictor for future state prediction, and all of such mappings generate a distribution of predicted future states. This distribution actually patches all pieces of association information from various embeddings unbiasedly or biasedly into the whole dynamics of the target variable, which, after operated by appropriate estimation strategies, creates a stronger predictor for achieving prediction in a more reliable and robust form. Through applying the RDE framework to data from both representative models and real world systems, we reveal that a high-dimension feature is no longer an obstacle but a source of information crucial to accurate prediction for short-term data even under noise deterioration. Actually, RDE can be applied to machine learning for small-sample training [3].

References:

[1] Xiaoping Liu, Xiao Chang, Siyang Leng, Hui Tang, Kazuyuki Aihara, Luonan Chen. Detection for disease tipping points by landscape dynamic network biomarkers. Natl Sci Rev, 2019, https://doi.org/10.1093/nsr/nwy162.

[2] Biwei Yang, Meiyi Li, Wenqing Tang, Weixin Liu, Si Zhang, Luonan Chen, Jinglin Xia. Dynamic network biomarker indicates pulmonary metastasis at the tipping point of hepatocellular carcinoma. Nature Communications, 9, 678, DOI: 10.1038/s41467-018-03024-2, 2018.

[3] Huanfei Ma, Siyang Leng, Kazuyuki Aihara, Wei Lin, Luonan Chen. Randomly Distributed Embedding Making Short-term High-dimensional Data Predictable. Proc Natl Acad Sci USA, 115 (43) E9994-E10002, https://doi.org/10.1073/pnas.1802987115, 2018.

[4] Juan Zhao, Yiwei Zhou, Xiujun Zhang, Luonan Chen. Part mutual information for quantifying direct associations in networks. Proc Natl Acad Sci USA, 2016, 113, 5130-5135.

Bio: Luonan Chen received BS degree in the Electrical Engineering, from Huazhong University of Science and Technology, China, and the M.E. and Ph.D. degrees in the electrical engineering, from Tohoku University, Japan, in 1988 and 1991, respectively. From 1997, he was an associate professor of the Osaka Sangyo University, Osaka, Japan, and then a full Professor. Since 2010, he has been a professor and executive director at Key Laboratory of Systems Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences. He was elected as the founding president of Computational Systems Biology Society of OR China, and Chair of Technical Committee of Systems Biology at IEEE SMC Society. His research areas include nonlinear dynamics, causality inference, machine learning and systems biology, in particular, dynamics-based data analysis. In recent years, he published over 300 journal papers and two monographs in the area of network and systems biology.

For more information on MIDAS or the Seminar Series, please contact midas-contact@umich.edu.