Université La Sagesse

Predicting the service life of structural parts with hydrophilic constituents requires accounting for moisture-induced aging during the design phase. This involves determining the moisture content distribution through the material’s thickness over time. Traditionally, composites with at most one hydrophilic constituent are considered. For materials with multiple heterogeneous hydrophilic constituents, traditional methods for estimating macroscopic moisture content are no longer adequate. To address this, this work introduces an original, general methodology to localize moisture at the constituent scale (matrix and hygroscopic reinforcements) from the knowledge of the macroscopic laminate moisture field and the sorption isotherms of each constituent. The proposed framework employs an algorithm to correlate macroscopic water content fields with local moisture distributions in composite constituents. Model validation was conducted by comparing predicted moisture profiles and diffusion characteristics with experimental data on a flax/epoxy composite from literature. Using the Hailwood-Horrobin equation, the model accurately represents the sorption isotherms of flax/epoxy composites, with a high correlation coefficient (R 2 = 0.9905) to experimental data. For the epoxy matrix, the predicted maximum moisture content capacities are 1.4 wt% and 2.5 wt% at 45% and 75% relative humidity, in close agreement with the experimental values of 1.2 wt% and 2.4 wt%, respectively. For flax fibers confined within the matrix (composite), the calculated maximum moisture content capacities are 2.7 wt% at 45% relative humidity and 6.8 wt% at 75% relative humidity, compared with 4.2 wt% and 7.0 wt% reported for free fibers, thereby highlighting the confinement effect of the surrounding matrix.

ABSTRACT This paper proposes a novel attention‐based fusion framework that integrates two self‐supervised learning models, the Point‐MAE and MaskPoint methods, to improve 3D point cloud classification. Each model is pretrained on a carefully selected dataset to optimize downstream performance. Their pretrained encoders are then fused using two attention strategies, namely multi‐layer attention and cross‐attention, to capture complementary features from both networks. A comparative evaluation of their original classification heads revealed that the Point‐MAE architecture, featuring a three‐layer MLP with batch normalization and dropout, consistently outperforms the MaskPoint design and was thus adopted in the proposed framework. Extensive experiments on both synthetic and real‐world benchmarks, including ModelNet40 and multiple variants of ScanObjectNN, demonstrate that attention‐based fusion improves classification robustness in challenging scenarios, with cross‐attention consistently achieving the highest accuracy among the evaluated fusion strategies. Controlled baseline experiments further confirm that these gains stem from meaningful feature interaction. The proposed approach demonstrates that attention‐guided fusion of self‐supervised point cloud representations is an effective and flexible strategy for enhancing downstream 3D recognition performance.

Artificial Neural Networks (ANNs) have become increasingly prevalent for classification and decision-making tasks across various engineering domains. However, their high computational resource requirements remain a significant barrier, particularly for portable or embedded applications. To address this challenge, this paper proposes an innovative method where the ANN learning phase is conducted using conventional software-based training, while the classification phase is implemented as an analog electrical circuit derived from the trained model. The method is validated on a medical diagnostic application: the detection of Multiple Sclerosis (MS) lesions from magnetic-resonance imaging (MRI) images. The dataset comprises 2,831 FLAIR and T2-weighted MRI images, partitioned into 70% training and 30% testing subsets. The proposed analog circuits achieve accuracies of 93.31% (MLP) and 94.82% (ResNet-10), while significantly reducing power consumption to 728 nW (MLP) and 967 nW (ResNet-10) and inference latency to 0.2 µs. Although the achieved accuracy slightly trails that of purely software-based implementations, the substantial reductions in power usage and response time make this hardware-based approach highly advantageous for real-time, low-power medical diagnostic devices.

Tilling, a common agricultural practice, is being done excessively on farms leading to about 2.35 billion tons of soil erosion from US croplands annually.  This causes soil erosion, soil infertility, carbon release, nutrient runoff, and fertilizer over-usage. This paper evaluates whether optimizing tillage intensity, timing, and fertilizer quantity will address these problems. A convolutional neural network based machine learning model utilizes a camera-captured field image to determine existing tilling intensity on a 7-point scale. This machine learning output, along with soil sensor and external forecast data, flows into a 10-parameter algorithm that determines optimal tilling and fertilizer levels. A fully functional tractor prototype demonstrates the above. A 30-year simulation comparing conventionally-tilled and algorithm-tilled farms showed a reduction in carbon emission by 57%, fertilizer usage by 43%, and runoff by 86% demonstrating the transformative potential of this algorithm. Additionally, a stationary prototype was deployed in 155 farms across 5 countries. 

There are 2.2 billion visually impaired individuals and 285 million blind people worldwide. The vestibular system plays a fundamental role in the balance of a person related to sight and hearing, and thus blind people require physical therapy to improve their balance. Several clinical tests have been developed to evaluate balance, such as the mini-BESTest. This test has been used to evaluate the balance of people with neurological diseases, but there have been no studies that evaluate the balance of blind individuals before. Furthermore, despite the scoring of these tests being not subjective, the performance of some activities are subject to the physiotherapist's bias. Tele-rehabilitation is a growing field that aims to provide physical therapy to people with disabilities. Among the technologies used in tele-rehabilitation are inertial measurement units that can be used to monitor the balance of individuals. The amount of data collected by these devices is large and the use of deep learning models can help in analyzing these data. Therefore, the objective of this study is to analyze for the first time the balance of blind individuals using the mini-BESTest and inertial measurement units and to identify the activities that best differentiate between blind and sighted individuals. We use the OpenSense RT monitoring device to collect data from the inertial measurement unit, and we develop machine learning and deep learning models to predict the score of the most relevant mini-BESTest activities. In this study 29 blind and sighted individuals participated. The one-legged stance is the activity that best differentiates between blind and sighted individuals. An analysis on the acceleration data suggests that the evaluation of physiotherapists is not completely adjusted to the test criterion. Cluster analysis suggests that inertial data are not able to distinguish between three levels of evaluation. However, the performance of our models shows an F1-score of 85.6% in predicting the score evaluated by the mini-BESTest in a binary classification problem. The results of this study can help physiotherapists have a more objective evaluation of the balance of their patients and to develop tele-rehabilitation systems for blind individuals.

To the Editor: We note the publication of the post-colonoscopy colorectal cancer (PCCRC) article (Fourie et al. [1] It denotes a baseline for South Africa (SA), and confirms that our PCCRC rates are globally on par.We have several concerns we wish to note regarding inferred conclusions: (i) The manner in which the PCCRC rate for medical gastroenterologists was established: It is seemingly unclear, but assumed, that the operator for the second diagnostic colonoscopy also performed the previous scope where the lesion was not diagnosed.This needs clarity and/or correction, as medical gastroenterologists, in the authors' conclusions, have a higher PCCRC rate.This seems methodologically unsound and casts unneeded aspersions.(ii) The vast majority of the cohort (14 773) were excluded from analysis, and of these, most (40%) were on the basis of a background high risk of CRC (polyposis syndromes, hereditary non-polyposis CRC, previous CRC, previous adenomas and inflammatory bowel disease (IBD)).This cannot be accurately done on the basis of ICD-10 codes only, as per methodology, as many of the above do not have ICD-10 codes (e.g.serrated polyposis syndrome; no distinction made for adenomas -rather the use of a broad category of 'benign neoplasms').It is anticipated that polyposis syndromes and IBD account for a minority of CRC cases, hence most were likely excluded given inadequate coding.The adenoma detection rate is a well-recognised and validated key performance indicator for colonoscopy quality, and not PCCRC.Adenoma detection rates are inversely correlated to PCCRC rates.This exclusionary approach potentially excludes scopes that detect adenomas, and by extension, lower rates of PCCRC.Not including these data in the analysis may inherently create significant bias.High-risk groups are of particular interest in our country, where we lack capacity to perform community-based screening, so that targeted screening is possibly a more feasible approach.(iii) The indications for colonoscopy were not considered.It is expected that those performing endoscopy in patients with alarm symptoms will have higher rates of CRC and subsequently PCCRC identification.Conversely, performing colonoscopies for screening or low-yield indications could lead to lower rates of PCCRC.Without proper matching for indications, the PCCRC rates cannot be compared between groups.Similarly, the utility of endoscopy needs to be considered.Repeating annual colonoscopies in low-risk patients will lead to lower rates of PCCRC and increased cost.Other validated key performance indicators were not considered (bowel preparation, caecal intubation rates).These data are required to correctly interpret the quality of colonoscopies to match CRC rates.(iv) The comparative groups were apportioned into surgeons, gastroenterologists and physicians/general practitioners.While it is acknowledged as a limitation, this leads to the data being uninterpretable.Most general specialist physicians do not receive 1.

The gaming landscape has undergone significant evolution over the years, with technological advancements consistently transforming how players engage with virtual environments. One pivotal aspect of this evolution lies in the interface between the player and the game - the controller. Traditional controllers, while effective, often present challenges such as high costs and limited life cycles. The proposed device is characterized by its low-cost, accuracy, and customization as it can adhere to the user motion. Shooting games, characterized by dynamic and fast-paced interactions, require precise and responsive control mechanisms for an immersive player experience. Therefore, this paper presents a comprehensive exploration of implementing sensor-based controls in shooting games. The selected sensors, coupled with the Arduino-based microcontroller integration, form the backbone of a system designed for accuracy, responsiveness, and adaptability. By providing players with a tangible connection to the virtual realm, the system aims to redefine the boundaries of player interaction, elevating the game beyond a mere gaming experience to a visceral and captivating journey where the line between player and character becomes delightfully blurred. The challenge lies in seamlessly translating physical gestures into in-game actions, redefining the player's relationship with the digital battlefield. Following a successful implementation, the system was subjected to extensive testing by a group of 20 users, with a total testing time exceeding 15 hours. This thorough evaluation yielded overwhelmingly positive feedback, affirming the system's functionality, reliability, and performance.

Abstract Background Caesarean section is one of the commonest surgical procedures worldwide. It is an important medical intervention for reducing the risk of poor perinatal outcomes. However, trends in caesarean section in sub-Saharan Africa continue to increase, while maternal and neonatal mortality and morbidity remain high. This study aims to determine the prevalence and factors associated with caesarean section in Tanzania. Methodology This was a secondary data analysis of 4,768 women of reproductive age (15-49) in Tanzania. The data utilized was from the Tanzania Demographic Health and Survey (TDHS) 2022. The factors associated with caesarean section were identified using multivariable binary logistic regression. Results Out of 4,768 women of reproductive age in Tanzania, 497 (10.4%) underwent a cesarean section. Attained primary (aOR: 1.79 ,95% CI 1.23-2.60), secondary (aOR:2.07,95% CI 1.36-3.14) and higher education (aOR: 2.35, 95% CI 1.08-5.12) or having a husband/partner with higher education ,being in richer household wealth quintile (aOR:1.98 ,95% CI (1.31-3.00) or in the middle (aOR :1.91 ,95% CI (1.28-2.85), having a job (aOR : 1.29, 95% CI: 1.05-1.58)and attended more than 4ANC(aOR: 1.36, 95% CI: 1.11-1.67) were associated with a higher odds of undergoing caesarean section compared to their respective counterparts. However, living in rural areas (aOR: 0.74, 95% CI:0.58- 0.94), being multiparous women with 2-4 births (aOR: 0.67, 95% CI: 0.53-0.84) and 5 and more births (aOR: 0.44, 95% CI: 0.32-0.60) were associated with lower odds of caesarean section. Conclusion The overall prevalence of caesarean section among women of reproductive age in Tanzania was 10.4%. The highest educational level, husband/partner's education level, household wealth quintile, type of residence, employment status, increased ANC number, and high parity were associated with caesarean section. There is an urgent need for continued efforts to ensure universal access to essential critical care services for all populations, thereby preventing premature maternal and infant mortality.