Bipolar disorder is characterized by repeated episodes of mania and depression, and can be understood as pathological complex system behaviour involving cognitive, affective and psychomotor disturbance. Accurate prediction of episode transitions in the long-term pattern of mood changes in bipolar disorder could improve the management of the disorder by providing an objective early warning of relapse. In particular, circadian activity changes measured via actigraphy may contain clinically relevant signals of imminent systemic dysregulation. In this study, we propose a mathematical index to investigate the correlation between apparently irregular circadian activity rhythms and critical transitions in episodes of bipolar disorder. Not only does the proposed index illuminate the effects of pharmacological and psychological therapies in control over the state, but it also provides a framework to understand the dynamic (or state-dependent) control strategies. Modelling analyses using our new approach suggest that key clinical goals are minimizing side effects of mood stabilizers as well as increasing the efficiency of other therapeutic strategies.

No abstract

Although neuroimaging research has strongly implicated a reciprocal interaction between cortical and subcortical regions as pathogenic in bipolar disorder, this is the first model to mathematically represent this multilevel explanation of the phenomena of bipolar disorder.

Analog, unclocked, spiking neuromorphic microchips open new perspectives for implantable or wearable biosensors and biocontrollers, due to their low energy consumption and heat dissipation. However, the challenges from a computational point of view are formidable. Here we outline our solutions to realize the reservoir computing paradigm on such hardware and address the combined problems of low bit resolution, device mismatch, approximate neuron models, and timescale mismatch. The main contribution is a computational scheme, called Reservoir Transfer, which enables us to transfer the dynamical properties of a well-performing neural network which has been optimized on a digital computer, onto neuromorphic hardware that displays the abovementioned problematic properties. Here we present a case study of implementing an ECG heartbeat abnormality detector to showcase the proposed method.

No abstract