An important difference between hospital environments and the community is resource availability. The successful implementation of community-based care requires the development of affordable, compact, and user-friendly solutions. This introduces trade-offs between data resolution and the cost and expertise required to collect, process, and interpret it. Thus, data analysis methods for supporting diagnosis and treatment of neurological conditions cannot be limited to cases where state-of-the-art equipment and a high level of technical expertise are readily available.
The development of bespoke techniques to maximise the potential of low-resolution data is critically important to improve patient care in the community. This will require an understanding of the clinical questions addressed by these modalities, the main characteristics of data used in community-based care, and the development of methods for maximising the information derived from those data.
In this workshop, we will bring together quantitative scientists, clinicians, and representatives from industry to discuss technologies that facilitate community-based care of neurological disorders. Topics will include:
– Exploration of the most relevant healthcare challenges in the community,
– Development of predictive models based on low-resolution data,
– Optimisation techniques (maximal information from minimal infrastructure),
– Development of automated pipelines for data handling.
The workshop will begin with a series of short presentations where key stakeholders will discuss current challenges in the acquisition, processing, and/or analysis of neurological data outside of a clinical setting and the potential of recent data-analytic tools to address them (optimisation techniques, machine learning, AI, etc.). This will be followed by a networking activity, where participants will discuss these challenges, develop collaborative links, and initiate a wider discussion on the importance of interdisciplinary approaches for facilitating community-based care.
13:00 – 13:10 – Arrival
13:10 – 13:20 – Welcome from the organisers
13:20 – 14:50 – Presentations (4 talks, 15mins + 5 mins Q&A)
14:50 – 15:00 – Break
15:00 – 16:00 – Focused Networking Activity
16:00 – 16:10 – Concluding remarks
16:10 – 17:00 – Time for 1-to-1 networking
Prof. David Grayden (University of Melbourme – Australia)
Title: Epileptic seizure forecasting using mobile and wearable devices.
Abstract: Surveys conducted by the Epilepsy Foundation of America and others have revealed that one of the most distressing aspect of epilepsy for individuals living with the condition is the uncertainty of when their seizures will occur. For approximately 30% of those affected by epilepsy, conventional medical interventions and therapies prove inadequate in managing their seizures, consequently impeding their ability to participate in activities such as driving and swimming. Thus, there is an imperative to develop methodologies that empower individuals with epilepsy to gain a deeper comprehension of their condition and potentially exert some level of control over it. Significant strides have been made over the past decade in harnessing the potential of phone-based and wearable technologies for predicting seizures. This presentation will examine various approaches that we are developing and explore the potential future landscape of epileptic seizure forecasting using mobile and wearable devices.
Dr. Andrew Quinn (University of Birmingham)
Title: Controlling for covariates and confounds when estimating EEG power spectra.
Abstract: The power spectrum of EEG data provides a summary of the oscillatory content of neuronal activity. Unfortunately, it also contains non neuronal dynamics relating to muscle artefacts, heart beats and movement. In an ideal setting, an experimenter can minimise these artefact sources by controlling the experimental set-up and communicating with the participant. However, this is not possible in many data recordings that take place in natural and clinical settings and are typically ‘noiser’ as a result. In addition, these recordings often have shorter time-series and fewer channels making standard data cleaning metrics challenging.
Here I will summarise the GLM-Spectrum as an approach for efficient power spectrum estimation which includes the possibility to include covariate and confound factors that can regress out potential noise sources. This is an adaptive approach that will not change the spectrum estimate unless the covariates actually impact the spectrum estimate. I will explain the theory behind the method and demonstrate its use in several EEG data recordings.
Dr. Ali Asghar Zarei (UNEEG Medical)
Title: Unmasking hidden patterns: insights from ultra long-term subcutaneous EEG data.
Abstract: Considering the clinical limitations and questions, we will discuss the reliability of subcutaneous electroencephalography (sqEEG) and potential in ultra long-term monitoring. We will dive into the practical implications of the novel, minimally-invasive sqEEG solution from UNEEG medical for outpatient recording, as well as discussing the automated seizure detection implementation and the potential of forecasting seizures. Join us in exploring the practical significance of ultra long-term sqEEG monitoring, ushering in a new era of comprehensive and personalized monitoring strategies with potential impacts on epilepsy management and hopefully increasing the quality of life for people with epilepsy.
Prof. Christopher James (University of Warwick)
Information coming soon.