In July 2020, the European Commission's High-Level Expert Group on AI (HLEG-AI) published the Assessment List for Trustworthy Artificial Intelligence (ALTAI) tool, enabling organizations to perform self-assessments of the fit of their AI systems and surrounding governance to the “7 Principles for Trustworthy AI.” Prior research on ALTAI has focused primarily on specific application areas, but there has yet to be a comprehensive analysis and broader recommendations aimed at proto-regulators and industry practitioners. This paper therefore starts with an overview of this tool, including an assessment of its strengths and limitations. The authors then consider the success by which the ALTAI tool is likely to be of utility to industry in improving understanding of the risks inherent in AI systems and best practices to mitigate such risks. It is highlighted how research and practices from fields such as Environmental Sustainability, Social Justice, and Corporate Governance (ESG) can be of benefit for addressing similar challenges in ethical AI development and deployment. Also explored is the extent to which the tool is likely to be successful in being taken up by industry, considering various factors pertaining to its likely adoption. Finally, the authors also propose recommendations applicable internationally to similar bodies to the HLEG-AI regarding the gaps needing to be addressed between high-level principles and practical support for those on the front-line developing or commercializing AI tools. In all, this work provides a comprehensive analysis of the ALTAI tool, as well as recommendations to relevant stakeholders, with the broader aim of promoting more widespread adoption of such a tool in industry.

Breast cancer is one of the most prevalent types of cancers worldwide and yet, its pathophysiology is poorly understood. Single-cell electrophysiological studies have provided evidence that membrane depolarization is implicated in the proliferation and metastasis of breast cancer. However, metastatic breast cancer cells are highly dynamic microscopic systems with complexities beyond a single-cell level. There is an urgent need for electrophysiological studies and technologies capable of decoding the intercellular signaling pathways and networks that control proliferation and metastasis, particularly at a population level. Hence, we present for the first time non-invasive in vitro electrical recordings of strongly metastatic MDA-MB-231 and weakly/non-metastatic MCF-7 breast cancer cell lines. To accomplish this, we fabricated an ultra-low noise sensor that exploits large-area electrodes, of 2 mm², which maximizes the double-layer capacitance and concomitant detection sensitivity. We show that the current recorded after adherence of the cells is dominated by the opening of voltage-gated sodium channels (VGSCs), confirmed by application of the highly specific inhibitor, tetrodotoxin (TTX). The electrical activity of MDA-MB-231 cells surpasses that of the MCF-7 cells, suggesting a link between the cells’ bioelectricity and invasiveness. We also recorded an activity pattern with characteristics similar to that of Random Telegraph Signal (RTS) noise. RTS patterns were less frequent than the asynchronous VGSC signals. The RTS noise power spectral density showed a Lorentzian shape, which revealed the presence of a low-frequency signal across MDA-MB-231 cell populations with propagation speeds of the same order as those reported for intercellular Ca²⁺ waves. Our recording platform paves the way for real-time investigations of the bioelectricity of cancer cells, their ionic/pharmacological properties and relationship to metastatic potential.

Organ-on-Chip technology is commonly used as a tool to replace animal testing in drug development. Cells or tissues are cultured on a microchip to replicate organ-level functions, where measurements of the electrical activity can be taken to understand how the cell populations react to different drugs. Microfluidic structures are integrated in these devices to replicate more closely an in vivo microenvironment. Research has provided proof of principle that more accurate replications of the microenvironment result in better micro-physiological behaviour, which in turn results in a higher predictive power. This work shows a transition from a no-flow (static) multi-electrode array (MEA) to a continuous-flow (dynamic) MEA, assuring a continuous and homogeneous transfer of an electrolyte solution across the measurement chamber. The process through which the microfluidic system was designed, simulated, and fabricated is described, and electrical characterisation of the whole structure under static solution and a continuous flow rate of 80 µL/min was performed. The latter reveals minimal background disturbance, with a background noise below 30 µVpp for all flow rates and areas. This microfluidic MEA, therefore, opens new avenues for more accurate and long-term recordings in Organ-on-Chip systems.

Peripheral nerve interfaces (PNIs) can enable communication with the peripheral nervous system and have a broad range of applications including in bioelectronic medicine and neuroprostheses. They can modulate neural activity through stimulation or monitor conditions by recording from the peripheral nerves. The recent growth of Machine Learning (ML) has led to the application of a wide variety of ML techniques to PNIs, especially in circumstances where the goal is classification or regression. However, the extent to which ML has been applied to PNIs or the range of suitable ML techniques has not been documented. Therefore, a scoping review was conducted to determine and understand the state of ML in the PNI field. The review searched five databases and included 63 studies after full-text review. Most studies incorporated a supervised learning approach to classify activity, with the most common algorithms being some form of neural network (artificial neural network, convolutional neural network or recurrent neural network). Unsupervised, semi-supervised and reinforcement learning approaches are currently underutilized and could be better leveraged to improve performance in this domain.

Breast cancer is a leading cause of death in women worldwide, and yet its pathophysiology is poorly understood. Although single-cell studies have highlighted the contribution of membrane depolarisation to the proliferation of breast cancer, dynamic signalling at a network level has not been extensively researched. It is urgent therefore to decode the intercellular signalling patterns linked to metastasis, particularly at a cell cohort level. This paper introduces a novel strategy for conducting such recordings on highly metastatic MDA-MB-231 cells, via an ultra-low noise biosensor based on a large electrode area which maximises the Helmholtz double-layer capacitance. The extracellular sensitivity of our biosensor allows the detection of pA-level random telegraph signal (RTS) noise superimposed with an omnipresent 1/f noise. The RTS noise is validated and modelled using a Markov chain. The analysis of slow cooperative potentials across the large area electrode suggests the involvement of cohort calcium signalling, and the 1/f noise analysis suggests a strong contribution of resting membrane noise. Overall, this work shows the potential of the new recording platform and statistical analysis for better understanding and predicting the underlying signalling mechanisms of metastatic breast cancer cells. In future, this platform could highlight the effects of compounds, or drugs, on the underlying activity of cancer cell cohorts in a clinical setting.

Little research exists on extending ex-vivo systems to large animal nerves, and to the best of our knowledge, there has yet to be a study comparing these against in-vivo data. This paper details the first ex-vivo system for large animal peripheral nerves to be compared with in-vivo results. Detailed ex-vivo and in-vivo closed-loop neuromodulation experiments were conducted on pig ulnar nerves. Temperatures from 20 °C to 37 °C were evaluated for the ex-vivo system. The data were analysed in the time and velocity domains, and a regression analysis established how evoked compound action potential amplitude and modal conduction velocity (CV) varied with temperature and time after explantation. Pig ulnar nerves were sustained ex-vivo up to 5 h post-explantation. CV distributions of ex-vivo and in-vivo data were compared, showing closer correspondence at 37 °C. Regression analysis results also demonstrated that modal CV and time since explantation were negatively correlated, whereas modal CV and temperature were positively correlated. Previous ex-vivo systems were primarily aimed at small animal nerves, and we are not aware of an ex-vivo system to be directly compared with in-vivo data. This new approach provides a route to understand how ex-vivo systems for large animal nerves can be developed and compared with in-vivo data. Conclusion The proposed ex-vivo system results were compared with those seen in-vivo, providing new insights into large animal nerve activity post-explantation. Such a system is crucial for complementing in-vivo experiments, maximising collected experimental data, and accelerating neural interface development.