Abstracts
Critical transitions and self-organization in sleep dynamics
Ronny Bartsch
Bar-Ilan University, Israel
ABSTRACT:
Traditionally, sleep is considered to operate according to the classical principle of homeostasis, which postulates that a system returns to equilibrium after perturbation, and that linear causality controls physiological interactions. However, the complex dynamics of sleep-stage transitions and arousals, which occur at short time scales during sleep and constitute the sleep micro-architecture, cannot be understood within this framework.
Our recent findings indicate that arousals represent previously unrecognized intrinsic aspects of sleep and exhibit complex temporal organization and scale-invariant behavior characterized by a power-law probability distribution for their durations. In contrast, sleep-stage durations exhibit exponential behavior. The co-existence of these two very different processes in the sleep regulatory mechanism has not been observed in other physiological systems and resembles a particular class of non-equilibrium physical systems exhibiting self-organized criticality (SOC).
This talk will present preliminary results suggesting that SOC can be seen as a new paradigm of sleep dynamics. We will show that the SOC-type dynamics persist throughout the sleep period in humans and across several mammalian and non-mammalian species with different sleep architectures. We will also discuss the link between sleep arousals and temperature regulation, which may be relevant for the clinically observed higher risk for sudden infant death syndrome at increased ambient temperatures.
The Conductome: A New Paradigm for Predicting and Explaining Human Behaviour
Christopher R. Stephens
National University of Mexico, Mexico
ABSTRACT:
As almost every major problem that humankind faces are a consequence of human behaviour, predicting behaviour and behaviour change is fundamental. Given the multitude of factors that affect our decision making, a transdisciplinary understanding of behaviour is impossible without the integration of data that cross disciplinary boundaries. The concept of Conduct-“ome” is an analog of those holistic –“omic”-approaches found in the biological sciences which take a “totality of factors” approach, and provides a framework for studying human behaviour in a multifactorial, multidisciplinary context, accounting for a wealth of potential causes of behaviour, from the genetic and epigenetic to psychological, neurological, social, physiological, clinical, socio-economic, socio-demographic, socio-political and ethical factors. We argue that behaviour can only be understood probabilistically, through a process of statistical inference that constructs P(A|X), the probability for a conduct A conditioned on the large set of factors, X, that predict it. This inference process can be based on an “external” ensemble of objective, countable events, using a frequentist interpretation of probabilities, or on an “internal” ensemble, implicit in our mental models and based on a Bayesian interpretation. Including these two approaches allows one to compare objective, observable reality with the subjective perception of reality constructed within a mental model, allowing for the identification of discrepancies between the two in the form of cognitive biases. A key component for constructing the Conductome is the obtention of data that transcends disciplines, and which can be used to link a range of relevant behaviours, as effects, to their causes, both internal and external. A second component is the use of advanced modelling tools, such as machine learning, for the analysis of such multi-scale data and the construction of explicit prediction models for a given conduct. In this talk, the feasibility of the Conductome approach is illustrated by considering obesity-related behaviours; as obesity has become one of the key social problems that affects a growing segment of the population worldwide. In summary: The objective is to understand, interpret and provide an interdisciplinary, computational and data-based framework for generating prediction models for addressing problems that originate in human behaviour.
Complex topological features of reservoirs shape learning performances in bio-inspired recurrent neural networks
Valeria D'Andrea
Fondazione Bruno Kessler, Trento, Italy
ABSTRACT:
Recurrent networks are a class of artificial neural systems that use their internal states to perform computing tasks. One of its state-of-the-art developments, i.e. reservoir computing (RC), uses the internal structure -- usually a static network with random structure -- to map an input signal into a nonlinear dynamical system defined in a higher dimensional space. However, it is fundamentally unknown how the random topology of the reservoir affects the learning performance. Here, we fill this gap by considering multiple synthetic networks -- characterized by different topological features -- and 45 empirical connectomes -- sampled from brain regions of organisms belonging to 8 different species -- to build the reservoir and test the learning performance against a prediction task with a variety of input signals. We find correlations between RC performances and both the number of nodes and rank of the covariance matrix of activation states, with performance depending on the nature -- stochastic or deterministic -- of input signals. The modularity and the number of independent states are found to affect RC performances. Overall, our findings highlight that the complex topological features characterizing biophysical computing systems such as connectomes can be used to design efficient bio-inspired artificial neural networks.
Correlation Transfer in Cortical Neurons
Michele Giugliano
SISSA, Trieste, Italy
ABSTRACT:
No matter how we probe the brain, we find correlated neuronal activity over a variety of spatial and temporal scales. For the cerebral cortex, significant evidence has accumulated on trial-to-trial covariability in synaptic inputs activation, subthreshold membrane potential fluctuations, and output spike trains. Although we do not yet fully understand their origin and whether they are detrimental or beneficial for information processing, we believe that clarifying how correlations emerge is pivotal for understanding large-scale neuronal network dynamics and computation. In this talk, I will report about quantitative differences between excitatory and inhibitory cells, as they relay input correlations into output correlations. I will explain this heterogeneity by a simple biophysical model and briefly explain how correlation transfer is linked to the neuron's dynamical linear transfer properties. As there is an excellent agreement between the theory and the experiment, I will conclude arguing that our experimental data represent the most validated test for a theory in explaining the emergence of correlations.
Deconstructing the molecular landscape of pediatric acute lymphoblastic leukemia using machine learning: A Divide-and-Conquer Approach
Olga Krali
Uppsala University, Sweden
ABSTRACT:
Acute lymphoblastic leukemia (ALL) is the most common childhood cancer. ALL is an extremely heterogeneous disease, characterized by numerous recurrent structural genomic aberrations. These aberrations define molecular subtypes, which have diagnostic and prognostic value. Therefore, accurate molecular classification at the time of diagnosis facilitates risk grouping and disease monitoring during treatment.
In recent years, the use of genomics for molecular classification has revolutionized how we diagnose and treat pediatric ALL. Basic research initiatives and clinical sequencing efforts around the world have demonstrated the power of this approach for detecting new subtypes, and it is becoming increasingly clear that these methods will eventually replace standard cytogenetics, FISH, and PCR. However, the successful implementation of these genomic approaches for upfront diagnostics in the clinic will require the development of robust machine learning algorithms that can effectively classify and interpret complex genomic data. As these types of data are characterized by high dimensionality, complexity, and large size, building machine learning models using biological input is challenging. To tackle this issue, we developed a model for ALL subtype Identification Using Machine learning (ALLIUM). ALLIUM uses nearest shrunken centroid (NSC) classification as its core component. We trained ALLIUM on 17 ALL subtypes present in our Nordic pediatric ALL cohort (n = 1131 patients) using DNA methylation (DNAm) and gene expression (GEX) data as input. ALLIUM achieved high performance on unseen DNAm or GEX data (>80%), resulting in highly accurate predictions regardless the data modality used as input. When we applied ALLIUM on 280 patients from this cohort without known molecular subtype (B-other ALL), we determined the subtype of as many as ~85% of the patients. In summary, we demonstrated the power of machine learning to resolve biological and data-driven complexity in pediatric ALL.
The physiology of Sleep: its interplay with epilepsy
Ani Kaplanian
Hofmann-La Roche, Basel, Switzerland
ABSTRACT:
Humans spend about one-third of their lives asleep, however its functions remains to be fully elucidated. Considerable advances in our understanding of the mechanisms and functions of sleep have occurred over the past twenty years and reveal its crucial role in normal brain physiology. Circadian rhythms regulate the sleep-wake cycle. In addition, the sleep-wake system is thought to be regulated by the interplay of two major processes, one that promotes sleep and one that maintains wakefulness. In this talk, we present the overview of sleep physiology, its characteristics, its role in memory consolidation and brain homeostasis and also how dysfunction of neural circuits that control sleep mechanisms underlie disorders in neural network level. As growing evidence shows bidirectional interactions between sleep, circadian rhythms and epilepsy, we focus on how our better understanding for sleep-epilepsy interaction may help to develop new diagnostic biomarkers for epilepsy as well as new treatment strategies for seizure control.