Publications related to IEL - Gestion

Challenges and Opportunities of Edge AI for Next-Generation Implantable BMIs

Neuroscience and neurotechnology are currently being revolutionized by artificial intelligence (AI) and machine learning. AI is widely used to study and interpret neural signals (analytical applications), assist people with disabilities (prosthetic applications), and treat underlying neurological symptoms (therapeutic applications). In this brief, we will review the emerging opportunities of on-chip AI for the next-generation implantable brain machine interfaces (BMIs), with a focus on state-of-the-art prosthetic BMIs. Major technological challenges for the effectiveness of AI models will be discussed. Finally, we will present algorithmic and IC design solutions to enable a new generation of AI-enhanced and high-channel-count BMIs.

IEEE2022

Identifying uncertainty states during wayfinding in indoor environments: An EEG classification Study

Mahsa Shoaran, Bingzhao Zhu

The researchers used a machine-learning classification approach to better understand neurological features associated with periods of wayfinding uncertainty. The participants (n = 30) were asked to complete wayfinding tasks of varying difficulty in a virtual reality (VR) hospital environment. Time segments when participants experienced navigational uncertainty were first identified using a combination of objective measurements (frequency of inputs into the VR controller) and behavioral annotations from two independent observers. Uncertainty time-segments during navigation were ranked on a scale from 1 (low) to 5 (high). The machine-learning model, a Random Forest classifier implemented using scikit-learn in Python, was used to evaluate common spatial patterns of EEG spectral power across the theta, alpha, and beta bands associated with the researcher-identified uncertainty states. The overall predictive power of the resulting model was 0.70 in terms of the area under the Receiver Operating Characteristics curve (ROC-AUC). These findings indicate that EEG data can potentially be used as a metric for identifying navigational uncertainty states, which may provide greater rigor and efficiency in studies of human responses to architectural design variables and wayfinding cues.

ELSEVIER SCI LTD2022

Discretization Uncertainties of Flow and Fatigue Damage Simulations of a Reversible Francis Pump-Turbine at Off-Design Operation in Turbine Mode

Daniel Biner, Philippe Alexandre Bontemps, Drazen Dujic

In the framework of the XFLEX HYDRO H2020 European Project, a 5 MW reversible Francis pump-turbine prototype, equipped with an industrial full size frequency converter, is investigated to demonstrate enhanced grid regulation capacities of variable speed hydropower units. The targeted high flexibility operation modes imply increased numbers of start and stop cycles, frequent off-design operation and even power reversions. Thereby, the understanding of fatigue mechanisms at off-design conditions and during transitions is crucial to ensure safe long term flexibility operation of hydropower units. The present paper aims to investigate spatial and temporal discretization uncertainties of numerical flow and fatigue damage simulations. For a selected off-design operating point in turbine mode, unsteady CFD simulations and one way coupled FSI simulations are carried out to reveal stress fluctuations and fatigue damage on the Francis pump-turbine runner. First, a classical Grid Convergence Index (GCI) methodology is carried out. Moreover, a design of experiments methodology is applied, comprising a fractional factorial plan, to evaluate the individual effects from the guide vane, runner, and draft tube grid sizes as well as the simulation time step on the flow and fatigue damage results. The guide vane grid size is found to be the most impactful on torque, discharge, and efficiency results. The correlation of the grid sizes with the fatigue damage responses shows less pronounced statistical evidence. Nevertheless, the study reveals expectable numerical error ranges related to discretization uncertainties of runner fatigue damage simulations within practicable grid size and time step ranges.

SPRINGER INTERNATIONAL PUBLISHING AG2022

,

Context Long-term adherence to physical activity (PA) interventions is challenging. The Lifestyle-integrated Functional Exercise programmes were adapted Lifestyle-integrated Functional Exercise (aLiFE) to include more challenging activities and a behavioural change framework, and then enhanced Lifestyle-integrated Functional Exercise (eLiFE) to be delivered using smartphones and smartwatches. Objectives To (1) compare adherence measures, (2) identify determinants of adherence and (3) assess the impact on outcome measures of a lifestyle-integrated programme. Design, setting and participants A multicentre, feasibility randomised controlled trial including participants aged 61-70 years conducted in three European cities. Interventions Six-month trainer-supported aLiFE or eLiFE compared with a control group, which received written PA advice. Outcome measures Self-reporting adherence per month using a single question and after 6-month intervention using the Exercise Adherence Rating Scale (EARS, score range 6-24). Treatment outcomes included function and disability scores (measured using the Late-Life Function and Disability Index) and sensor-derived physical behaviour complexity measure. Determinants of adherence (EARS score) were identified using linear multivariate analysis. Linear regression estimated the association of adherence on treatment outcome. Results We included 120 participants randomised to the intervention groups (aLiFE/eLiFE) (66.3 +/- 2.3 years, 53% women). The 106 participants reassessed after 6 months had a mean EARS score of 16.0 +/- 5.1. Better adherence was associated with lower number of medications taken, lower depression and lower risk of functional decline. We estimated adherence to significantly increase basic lower extremity function by 1.3 points (p

BMJ PUBLISHING GROUP2022

Polydimethylsiloxane microstructure-induced acoustic streaming for enhanced ultrasonic DNA fragmentation on a microfluidic chip

Martinus Gijs, Thomas Lehnert, Lin Sun

Next-generation sequencing (NGS) is an essential technology for DNA identification in genomic research. DNA fragmentation is a critical step for NGS and doing this on-chip is of great interest for future integrated genomic solutions. Here we demonstrate fast acoustofluidic DNA fragmentation via ultrasound-actuated elastic polydimethylsiloxane (PDMS) microstructures that induce acoustic streaming and associated shear forces when placed in the field of an ultrasonic transducer. Indeed, acoustic streaming locally generates high tensile stresses that can mechanically stretch and break DNA molecule chains. The improvement in efficiency of the on-chip DNA fragmentation is due to the synergistic effect of these tensile stresses and ultrasonic cavitation phenomena. We tested these microstructure-induced effects in a DNA-containing microfluidic channel both experimentally and by simulation. The DNA fragmentation process was evaluated by measuring the change in the DNA fragment size over time. The chip works well with both long and short DNA chains; in particular, purified lambda (lambda) DNA was cut from 48.5 kbp to 3 kbp in one minute with selected microstructures and further down to 300 bp within two and a half minutes. The fragment size of mouse genomic DNA was reduced from 1.4 kbp to 400 bp in one minute and then to 200 bp in two and a half minutes. The DNA fragmentation efficiency of the chip equipped with the PDMS microstructures was twice that of the chip without the microstructures. Exhaustive comparison shows that the on-chip fragmentation performance reaches the level of high-end professional standards. Recently, DNA fragmentation was shown to be enhanced using vibrating air microbubbles when the chip was placed in an acoustic field. We think the microbubble-free microstructure-based device we present is easier to operate and more reliable, as it avoids microbubble preparation and maintenance, while showing high DNA fragmentation performance.

ROYAL SOC CHEMISTRY2022

Performance Analysis of Energy Balancing Methods for Matrix Modular Multilevel Converters

Philippe Alexandre Bontemps, Drazen Dujic, Stefan Milovanovic

Modular multilevel converters achieve voltage scalability through the series connection of cells, each with its own capacitance. However, this converter structure using floating energy storage elements increases the control complexity significantly. The voltages of each capacitance must be controlled to their respective setpoints to ensure correct operation of the converter. Relying on the use of circulating currents, which are unobservable at the converter terminals, the energy can be exchanged among the branches. Three control principles resulting in three different implementations are thoroughly explained and their dynamic performance is compared. The results are verified using a hardware-in-the-loop platform simulating a hydro power plant using matrix modular multilevel converter connected between a 6.6 kV grid and a 6 kV synchronous machine.

2022

Augmented Cooper test: Biomechanical contributions to endurance performance

Kamiar Aminian, Salil Apte, Vincent Gremeaux, Simone Troxler

Running mechanics are modifiable with training and adopting an economical running technique can improve running economy and hence performance. While field measurement of running economy is cumbersome, running mechanics can be assessed accurately and conveniently using wearable inertial measurement units (IMUs). In this work, we extended this wearables-based approach to the Cooper test, by assessing the relative contribution of running biomechanics to the endurance performance. Furthermore, we explored different methods of estimating the distance covered in the Cooper test using a wearable global navigation satellite system (GNSS) receiver. Thirty-three runners (18 highly trained and 15 recreational) performed an incremental laboratory treadmill test to measure their maximum aerobic speed (MAS) and speed at the second ventilatory threshold (sVT2). They completed a 12-minute Cooper running test with foot-worm IMUs and a chest-worn GNSS-IMU on a running track 1-2 weeks later. Using the GNSS receiver, an accurate estimation of the 12-minute distance was obtained (accuracy of 16.5 m and precision of 1.1%). Using this distance, we showed a reliable estimation [R-2 > 0.9, RMSE epsilon (0.07, 0.25) km/h] of the MAS and sVT2. Biomechanical metrics were extracted using validated algorithm and their association with endurance performance was estimated. Additionally, the high-/low-performance runners were compared using pairwise statistical testing. All performance variables, MAS, sVT2, and average speed during Cooper test, were predicted with an acceptable error (R-2 >= 0.65, RMSE

FRONTIERS MEDIA SA2022

Introducing Alternative Value Proposition Canvases for Collaborative and Blended Design Thinking Activities in Science and Engineering Education

Isabelle Barbara Marie-Hélène Cardia, Denis Gillet

The value proposition canvas introduced in 2015 by Pigneur et al. is broadly used by entrepreneurs and intrapreneurs to create new services and products. We decided to apply it for collaborative and blended design thinking activities in a specialization in human-Computer interaction for master students in computer and data sciences. While being very useful, the original value proposition canvas relies on a business-oriented vocabulary not familiar to science and engineering students. Moreover, its underlying profit-oriented objective makes it difficult to focus on impact and trust-oriented design activities. In this paper, we propose two alternative value proposition canvases, one dedicated to sustainability for development and one to explainability for artificial intelligence. The motivations why and the ways how these alternative canvases were elicited and used in higher education are also discussed.

IEEE2022

Correspondence Between Values of Vertical Loading Rate and Oxygen Consumption During Inclined Running

Kamiar Aminian, Mathieu Pascal Falbriard, Frédéric Meyer, Grégoire Millet

Purpose: The aim of this study was to provide a theoretical model to predict the vertical loading rate (VLR) at different slopes and speeds during incline running. Methods: Twenty-nine healthy subjects running at least once a week performed in a randomized order 4-min running trials on an instrumented treadmill at various speeds (8, 10, 12, and 14 km h(-1)) and slopes (- 20%, - 10%, - 5%, 0%, + 5%, + 10%,+ 15%,+ 20%). Heart rate, gas exchanges and ground reaction forces were recorded. The VLR was then calculated as the slope of the vertical force between 20 and 80% of the duration from initial foot contact to the impact peak. Results: There was no difference in VLR between the four different uphill conditions at given running speeds, but it was reduced by 27% at 5% slope and by 54% at 10% slope for the same metabolic demand (similar VO2), when compared to level running. The average VLR measured at maximal aerobic intensity during level running would be decreased by 52.7% at + 5%, by 63.0% at+ 10%, and by 73.3% at+ 15% slope. Moreover, VLR was dependent on the slope in downhill conditions. Conclusion: This study highlights the possibility to use uphill running to minimize rate of mechanical load (i.e., osteoarticular load) from foot impact on the ground and as a time-efficient exercise routine (i.e., same energy expenditure than in level running in less time).

SPRINGER2022

Utilizing RT-Boxes to Support the Development of Medium Voltage Modular Multilevel Converters

Drazen Dujic

2022