The concept of dynamical disease [i.e., the qualitative transition (bifurcation) to a diseased state through changes in control parameters], proposed almost a couple of decades ago, is intuitively useful in disease forecasting, prevention and control. However, its verification has been difficult because it requires large amount of data for the state variables to gain quantitative insights into the dynamical state of the system. Now, it is increasingly possible to quantitatively study disease dynamics(e.g.,[1], [2], [3]), especially near bifurcation points, based on large amount of data owing to recent advances in wearable and/or biomedical sensing technologies. In this mini-symposium, we aim at providing a forum to discuss how the concepts of dynamical disease and predictive medicine can be clinically beneficial in the era of "data science".

Abstract:
Dynamical diseases arise because of abnormalities in underlying physiological control mechanisms. They are manifested by the sudden onset of a qualitative change in dynamics in physiological variables: an oscillation may appear where it is not normally observed; an existing oscillation may change its periodicity or even disappear altogether. Although such changes can occur in mathematical models of control when critical parameters are altered, translating theory into therapy has been difficult since the most relevant parameters in patients are often difficult to identify. Particularly problematic are those diseases in which medical emergencies recur in a paroxysmal and seemingly unpredictable manner, e.g., cardiac and respiratory arrhythmias, seizures in patients with epilepsy and falls in the elderly. Here I review evidence that supports the development of �ephysiological defibrillators�f to treat dynamical diseases, namely devices that monitor relevant physiological variables and deliver treatment when it is needed. The increasing availability and use of wearable physiological monitoring devices may make it possible to identify individuals at risk even before they have sought medical attention.

Abstract:
In this talk, I review our recent research on DNB(Dynamical Network Biomarkers) toward predictive medicine. By this research, we extended the concept of the conventional biomarkers to that of DNB on the basis of complex dynamical systems theory(1,2). This DNB analysis makes it possible to detect early-warning signals peculiar to pre-disease states, or "Mibyo" states in the oriental medicine, just before sudden deterioration of diseases, namely imminent transitions from healthy states to disease states, and start to treat and even control the pre-disease states.
References
(1) L. Chen, R. Liu, Z.-P. Liu, M. Li, and K. Aihara: Scientific Reports, Vol.2, Article No.342, pp.1-8 (2012).
(2) K.Aihara, J.Imura, and T.Ueta(Eds.), Analysis and Control of Complex Dynamical Systems: Robust Bifurcation, Dynamic Attractors, and Network Complexity, Springer, Tokyo (2015).

Junichiro Hayano¹⁾, Ken Kiyono²⁾, Eiichi Watanabe³⁾, Yoshiharu Yamamoto⁴⁾

1) Graduate School of Medical Sciences, Nagoya City University, Nagoya, Japan 2) Graduate School of Engineering Sciences, Osaka University, Osaka, Japan 3) Division of Cardiology, Department of Internal Medicine, Fujita Health University School of Medicine, Toyoake, Japan 4) Graduate School of Education, The University of Tokyo, Tokyo, Japan

Abstract:
TBA

Taishin Nomura, Yasuyuki Suzuki, Fu Chunjiang, Ken Kiyono

Graduate School of Engineering Science, Osaka University, Osaka, Japan

Abstract:
Studies on human motor control aim to understand how the central nervous system controls mechanical dynamics of our body in a stable and energetically efficient manner in order to achieve goals of control. Moreover, it is of significant importance to understand how such motor control capability is lost due to neurological diseases that exhibit motor dysfunction. Here, we consider a motor control during human bipedal standing and walking in healthy individuals that can achieve two apparently contrasting goals at the same time, namely dynamic stability and movement flexibility. We also contrast this interesting motor capability in healthy individuals with motor dysfunction, including movement instability in patients with Parkinson's disease. To this end, we address issues associated with movement variability, such as postural sway and gait cycle variability, which can be characterized by magnitude and fractality (power-law distributed long-range correlation) of the variability, among other indices. Based on our quantitative characterization of movement variability, we claim that the intermittent time-delayed feedback control that switches between "off-phases" and "on-phases" of active control in a state-dependent manner might be a promising control strategy that can achieve stability and flexibility simultaneously. An interesting feature of the proposed intermittent feedback control model is that dynamics of the movement might be unstable not only during "off-phases" without active feedback control but also during "on-phases" with time-delayed active feedback control, nonetheless the overall switching dynamics can exhibit bounded stability. Moreover, we will try to show that severity of postural instability and gait dysfunction in patients can be characterized by a loss of fractality in their movement variability. Comparisons between experimental and modelled dynamics suggest that a loss of intermittency in the on-off switching of the active feedback control is one of the major causes of motor dysfunction in patients with Parkinson's disease.

Jerome C. Foo¹⁾, Hamid R. Noori²⁾, Ikuhiro Yamaguchi¹⁾, Valentina Vengeliene²⁾, Alejandro Cosa-Linan²⁾, Toru Nakamura¹⁾, Kenji Morita¹⁾, Rainer Spanagel²⁾, Yoshiharu Yamamoto¹⁾

1) Graduate School of Education, The University of Tokyo, Tokyo, Japan 2) Institute of Psychopharmacology, Central Institute of Mental Health, Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany

Abstract:
Disease dynamics can be characterized by features of complex systems such as critical phase transitions, but in the biomedical field little evidence has been provided for this concept so far. Technological advancements are now making it possible to measure the intensive longitudinal data (ILD) necessary to capture pathologically-relevant signal components exhibiting the multiscale complexity of disease dynamics. Using a well-established model of alcohol relapse in rats as an example of disease onset and progression, we applied a multiscale computational approach to extract dynamical characteristics of massive high-resolution measurements of rat drinking behavior and locomotor activity. We show a stage-by-stage dynamical phase transition into relapse behavior preceded by early warning signals such as instability of drinking behavior and circadian rhythms, and a resultant increase in low frequency, ultradian rhythms. This study provides a blueprint for processing ILD from clinical studies and will help to predict disease dynamics in general.