Neural interfaces for monitoring and targeted treatment
Dries Braeken (imec): Printing human microcircuits on CMOS chips
The key limitation to understanding and treating neurodegenerative disorders is the failure of non-human models to reproduce clinically-relevant aspects of these diseases. We are developing a technology platform to print mature human neuronal microcircuits relevant to Parkinson’s disease on a 16-well multi-electrode array (MEA) complementary metal oxide semiconductor (CMOS) chip. Our strategy is based on the local electroporation of defined transcription factors in each well using spatially restricted activation of the chip’s electrodes. This will endow local neuronal progenitors with the capacity to differentiate into defined neuronal cell types. Using the chip’s built-in recording circuitry, we will measure electrophysiological changes between induced neuronal circuits of cells obtained from healthy people and from an extensive collection of Parkinson patients. We will perturb key pathways based on our prior extensive mechanistic insights to correct the defects, and we will deliver a unique screening platform to test novel therapeutic approaches. Our CMOS-MEA chip with induced microcircuits is adaptable (ie different transcription factors can be used) and scalable (ie other cell types can be added) and thus applicable to other diseases as well and will make this tool available for the community.
Christopher Grigsby (Karolinska Institutet): Bioactive melt electrospun scaffolds for neuronal reprogramming and brain organoids
The production of precisely specified, mature, and functional neuronal cells is critical for both the modeling and treatment of neurodegenerative disease. Relative to the biochemical and genetic cues provided to cells in vitro, the physical and electrical components of their microenvironment are often neglected. However, new biomaterial engineering techniques have begun to enable more comprehensive optimization of the cellular microenvironment. Using melt electrospinning, an electrohydrodynamic additive manufacturing technique that enables layer-by-layer deposition of polymer microfibers with controlled geometries, we are able to generate 3D-engineered scaffolds comprised of poly-ε-caprolactone (PCL) and poly(3,4-ethylenedioxythiophene) (PEDOT). The higher viscosity and lower conductivity of polymer melts compared to solutions helps to eliminate instabilities during fabrication. While the technique gives more control than conventional electrospinning and smaller feature sizes than standard 3D-printing, it has only recently been explored for biomedical applications. When applied to the non-viral direct reprogramming of cells to neurons, 3D functionalized melt electrospun scaffolds increase reprogramming yields, encourage survival, and promote cell maturation. When applied to the production of brain organoids, the scaffolds improve both the quality and reproducibility of the organoids. Our ongoing work aims to further characterize the structure-function relationships between melt electrospun biomaterial scaffolds and their effects on cell and organoid engineering. Alongside traditional biochemical and genetic cues, an optimized physical and electrical microenvironment is critical to produce the highest quality neuronal cells and organoids for disease modeling and regenerative medicine.
Stephanie Lacour (École polytechnique fédérale de Lausanne): Soft microfabricated and multimodal neural interfaces
We are exploring novel device materials and their associated technologies to design and manufacture soft implantable neural interfaces. They are broadly defined as microfabricated devices with mechanical properties suited to comply the soft and dynamic biological tissues. The soft interfaces host micro- to macro-electrodes for neural recordings and electrical stimulation but also microLED arrays for optical stimulation and embedded microfluidic delivery for drugs. This talk will report on soft implants optimized for electrocorticography, multimodal stimulation of the spinal cord and the peripheral nerves.
James Fawcett (University of Cambridge): Novel cognitive enhancement treatments for dementia
The most effective treatments for CNS damage have their main effect through stimulating plasticity. The CNS then has to learn how to use new connections, so effective treatments combine rehabilitation with measures that affect plasticity or excitability. To promote recovery in animal models of spinal cord injury, stroke, Alzheimer’s disease and ageing a successful method has been to target the perineuronal nets by digesting of chondroitin sulphate proteoglycans or by inhibiting perineuronal net synthesis. This has allowed restoration of sensorimotor function and memory. Perineuronal nets exert their effects through the sulfation motifs on the chondroitin sulphate glycan chains, which bind to effectors which include semaphorin3A. The sulfation pattern of the nets changes with ageing, rendering the nets more inhibitory and preventing memory formation. Treatments that remove the nets or change their sulphation pattern can restore memory.
Payam Barnaghi (University of Surrey): AI and in-home monitoring technologies to improve dementia care
Network-enabled devices and sensing/actuation technologies, i.e. Internet of Things (IoT), are changing the way we live and interact with our environment. IoT is a rapidly growing domain and is quickly becoming a key enabler in various domains including healthcare. This talk will focus on the applications of IoT technologies and machine learning in remote healthcare monitoring for people with dementia. We will discuss an IoT platform that uses data analytics and machine learning to extract actionable information from raw observations in an in-home healthcare monitoring setting. The developed system is currently in use for supporting people with dementia and their caregivers in their home environment. The system supports data-driven risk assessment to enable effective clinical interventions. The architecture and analytics solutions developed in this system are also applicable to other healthcare conditions. The talk will discuss some of the key design elements and will briefly describe machine learning algorithms that process sensory data to provide clinical insights and risk assessment.
Aldo Faisal (Imperial College London): Wearable digital biomarkers in-the-wild
Behaviour is the only way for our brain to interact with the world, consequently we hypothesise that appropriate monitoring of natural human behaviour in daily-life yields allows us to infer with high resolution the internal physiological and mental state of users. We demonstrated already that natural behaviour in-the-wild is distinctly different from laboratory-based tasks and even so called "activities of daily living". Natural behaviour thus yields a principled pathway towards the definition of a data-driven health- and disease-state. These data-defined measures of disease state are not only more precise, but can also enable studies with smaller clinical populations and shorter duration. We will report on our ongoing NIHR clinical trials in the neurogenerative disease domains to develop these data-driven measures into EMA/FDA validated digital biomarkers. Furthermore, we show how our recent work on action grammars allows us not only to quantify the ability of patients to perform motor tasks, but allow us to aggregate and extract higher-order structure of daily activities and thus cognitive function.
Nick Van Helleputte (imec): How wearable health & ubiquitous technology can improve care in patients with dementia: a prospective outlook.
The onset of neurodegenerative diseases happens over the course of many years and early diagnosis remains challenging. Researchers are trying to identify early modifiers like sleep disorders and cognitive decline and behavioral change and linking them to disease progression. In this talk, we will discuss wearable and invisible technology that could help to quantify some of these early modifiers. Recent years have seen a number of important developments in ultra-low-power wearables that can monitor brain function, autonomous nervous system and various aspects of sleep and behavior. Leveraging the availability of a multitude of data with new developments in machine learning and data analytics allows us to develop personalized models and predictors. In this talk I will discuss how we have applied these tools to develop a method to measure acute stress and anxiety.
Ellie D’Hondt (imec): High-performance computing for the life sciences
Computational bottlenecks are prevalent in life science research. Use cases such as gene sequencing, omics, image processing and population modelling could all benefit from faster tools that can handle more data at a time. Data-driven analytics and machine learning are even more compute-hungry. In this talk I will provide a number of examples were high-performance computing techniques were able to bring life sciences use cases to the next level. This includes elPrep, a sequencing alignment software tool that is now running in clinical settings. These examples should convince you that computational challenges can be tackled with the right expertise, and should not stop you from taking your research one step further.