Personalised modelling with spiking neural networks integrating temporal and static information

A State-of-Art Review of Digital Technologies for the Next Generation of Tinnitus Therapeutics

Background: Digital processing has enabled the development of several generations of technology for tinnitus therapy. The first digital generation was comprised of digital Hearing Aids (HAs) and personal digital music players implementing already established sound-based therapies, as well as text based information on the internet. In the second generation Smart-phone applications (apps) alone or in conjunction with HAs resulted in more therapy options for users to select from. The 3rd generation of digital tinnitus technologies began with the emergence of many novel, largely neurophysiologically-inspired, treatment theories that drove development of processing; enabled through HAs, apps, the internet and stand-alone devices. We are now of the cusp of a 4th generation that will incorporate physiological sensors, multiple transducers and AI to personalize therapies. Aim: To review technologies that will enable the next generations of digital therapies for tinnitus. Methods: A "state-of-the-art" review was undertaken to answer the question: what digital technology could be applied to tinnitus therapy in the next 10 years? Google Scholar and PubMed were searched for the 10-year period 2011-2021. The search strategy used the following key words: "tinnitus" and ["HA," "personalized therapy," "AI" (and "methods" or "applications"), "Virtual reality," "Games," "Sensors" and "Transducers"], and "Hearables." Snowballing was used to expand the search from the identified papers. The results of the review were cataloged and organized into themes. Results: This paper identified digital technologies and research on the development of smart therapies for tinnitus. AI methods that could have tinnitus applications are identified and discussed. The potential of personalized treatments and the benefits of being able to gather data in ecologically valid settings are outlined. Conclusions: There is a huge scope for the application of digital technology to tinnitus therapy, but the uncertain mechanisms underpinning tinnitus present a challenge and many posited therapeutic approaches may not be successful. Personalized AI modeling based on biometric measures obtained through various sensor types, and assessments of individual psychology and lifestyles should result in the development of smart therapy platforms for tinnitus.

Deep Learning of Explainable EEG Patterns as Dynamic Spatiotemporal Clusters and Rules in a Brain-Inspired Spiking Neural Network

The paper proposes a new method for deep learning and knowledge discovery in a brain-inspired Spiking Neural Networks (SNN) architecture that enhances the model’s explainability while learning from streaming spatiotemporal brain data (STBD) in an incremental and on-line mode of operation. This led to the extraction of spatiotemporal rules from SNN models that explain why a certain decision (output prediction) was made by the model. During the learning process, the SNN created dynamic neural clusters, captured as polygons, which evolved in time and continuously changed their size and shape. The dynamic patterns of the clusters were quantitatively analyzed to identify the important STBD features that correspond to the most activated brain regions. We studied the trend of dynamically created clusters and their spike-driven events that occur together in specific space and time. The research contributes to: (1) enhanced interpretability of SNN learning behavior through dynamic neural clustering; (2) feature selection and enhanced accuracy of classification; (3) spatiotemporal rules to support model explainability; and (4) a better understanding of the dynamics in STBD in terms of feature interaction. The clustering method was applied to a case study of Electroencephalogram (EEG) data, recorded from a healthy control group (n = 21) and opiate use (n = 18) subjects while they were performing a cognitive task. The SNN models of EEG demonstrated different trends of dynamic clusters across the groups. This suggested to select a group of marker EEG features and resulted in an improved accuracy of EEG classification to 92%, when compared with all-feature classification. During learning of EEG data, the areas of neurons in the SNN model that form adjacent clusters (corresponding to neighboring EEG channels) were detected as fuzzy boundaries that explain overlapping activity of brain regions for each group of subjects.

Design of MRI structured spiking neural networks and learning algorithms for personalized modelling, analysis, and prediction of EEG signals

This paper proposes a novel method and algorithms for the design of MRI structured personalized 3D spiking neural network models (MRI-SNN) for a better analysis, modeling, and prediction of EEG signals. It proposes a novel gradient-descent learning algorithm integrated with a spike-time-dependent-plasticity algorithm. The models capture informative personal patterns of interaction between EEG channels, contrary to single EEG signal modeling methods or to spike-based approaches which do not use personal MRI data to pre-structure a model. The proposed models can not only learn and model accurately measured EEG data, but they can also predict signals at 3D model locations that correspond to non-monitored brain areas, e.g. other EEG channels, from where data has not been collected. This is the first study in this respect. As an illustration of the method, personalized MRI-SNN models are created and tested on EEG data from two subjects. The models result in better prediction accuracy and a better understanding of the personalized EEG signals than traditional methods due to the MRI and EEG information integration. The models are interpretable and facilitate a better understanding of related brain processes. This approach can be applied for personalized modeling, analysis, and prediction of EEG signals across brain studies such as the study and prediction of epilepsy, peri-perceptual brain activities, brain-computer interfaces, and others.

Behavioral Outcomes and Neural Network Modeling of a Novel, Putative, Recategorization Sound Therapy

The mechanisms underlying sound's effect on tinnitus perception are unclear. Tinnitus activity appears to conflict with perceptual expectations of "real" sound, resulting in it being a salient signal. Attention diverted towards tinnitus during the later stages of object processing potentially disrupts high-order auditory streaming, and its uncertain nature results in negative psychological responses. This study investigated the benefits and neurophysiological basis of passive perceptual training and informational counseling to recategorize phantom perception as a more real auditory object. Specifically, it examined underlying psychoacoustic correlates of tinnitus and the neural activities associated with tinnitus auditory streaming and how malleable these are to change with targeted intervention. Eighteen participants (8 females, 10 males, mean age = 61.6 years) completed the study. The study consisted of 2 parts: (1) An acute exposure over 30 min to a sound that matched the person's tinnitus (Tinnitus Avatar) that was cross-faded to a selected nature sound (Cicadas, Fan, Water Sound/Rain, Birds, Water and Bird). (2) A chronic exposure for 3 months to the same "morphed" sound. A brain-inspired spiking neural network (SNN) architecture was used to model and compare differences between electroencephalography (EEG) patterns recorded prior to morphing sound presentation, during, after (3-month), and post-follow-up. Results showed that the tinnitus avatar generated was a good match to an individual's tinnitus as rated on likeness scales and was not rated as unpleasant. The five environmental sounds selected for this study were also rated as being appropriate matches to individuals' tinnitus and largely pleasant to listen to. There was a significant reduction in the Tinnitus Functional Index score and subscales of intrusiveness of the tinnitus signal and ability to concentrate with the tinnitus trial end compared to baseline. There was a significant decrease in how strong the tinnitus signal was rated as well as ratings of how easy it was to ignore the tinnitus signal on severity rating scales. Qualitative analysis found that the environmental sound interacted with the tinnitus in a positive way, but participants did not experience change in severity, however, characteristics of tinnitus, including pitch and uniformity of sound, were reported to change. The results indicate the feasibility of the computational SNN method and preliminary evidence that the sound exposure may change activation of neural tinnitus networks and greater bilateral hemispheric involvement as the sound morphs over time into natural environmental sound; particularly relating to attention and discriminatory judgments (dorsal attention network, precentral gyrus, ventral anterior network). This is the first study that attempts to recategorize tinnitus using passive auditory training to a sound that morphs from resembling the person's tinnitus to a natural sound. These findings will be used to design future-controlled trials to elucidate whether the approach used differs in effect and mechanism from conventional Broadband Noise (BBN) sound therapy.