AI-driven Discovery of Morphomolecular Signatures in Toxicology.

Early identification of drug toxicity is essential yet challenging in drug development. At the preclinical stage, toxicity is assessed with histopathological examination of tissue sections from animal models to detect morphological lesions. To complement this analysis, toxicogenomics is increasingly employed to understand the mechanism of action of the compound and ultimately identify lesion-specific safety biomarkers for which in vitro assays can be designed. However, existing works that aim to identify morphological correlates of expression changes rely on qualitative or semi-quantitative morphological characterization and remain limited in scale or morphological diversity. Artificial intelligence (AI) offers a promising approach for quantitatively modeling this relationship at an unprecedented scale. Here, we introduce GEESE, an AI model designed to impute morphomolecular signatures in toxicology data. Our model was trained to predict 1,536 gene targets on a cohort of 8,231 hematoxylin and eosin-stained liver sections from Rattus norvegicus across 127 preclinical toxicity studies. The model, evaluated on 2,002 tissue sections from 29 held-out studies, can yield pseudo-spatially resolved gene expression maps, which we correlate with six key drug-induced liver injuries (DILI). From the resulting 25 million lesion-expression pairs, we established quantitative relations between up and downregulated genes and lesions. Validation of these signatures against toxicogenomic databases, pathway enrichment analyses, and human hepatocyte cell lines asserted their relevance. Overall, our study introduces new methods for characterizing toxicity at an unprecedented scale and granularity, paving the way for AI-driven discovery of toxicity biomarkers. Live demo: https://mahmoodlab.github.io/tox-discovery-ui/.

Case study into the successful emergency production and certification of a filtering facepiece respirator for Belgian hospitals during the COVID-19 pandemic.

The SARS-CoV-2 pandemic presented European hospitals with chronic shortages of personal protective equipment (PPE) such as surgical masks and respirator masks. Demand outstripped the production capacity of certified European manufacturers of these devices. Hospitals perceived emergency local manufacturing of PPE as an approach to reduce dependence on foreign supply. The fact of a pandemic does not circumvent the hospital's responsibility to provide appropriate protective equipment to their staff, so the emergency production needed to result in devices that were certified by testing agencies. This paper is a case study of the emergency manufacturing of respirator masks during the first month of the first wave of SARS-CoV-2 pandemic and is separated into two distinct phases. Phase A describes the three-panel folding facepiece respirator design, material sourcing, performance testing, and an analysis of the folding facepiece respirator assembly process. Phase B describes the redevelopment of individual steps in the assembly process.

Vibrotactile feedback for correcting aerodynamic position of a cyclist.

Guidance to maintain an optimal aerodynamic position is currently unavailable during cycling. This study used real-time vibrotactile feedback to guide cyclists to a reference position with minimal projected frontal area as an indicator of aerodynamic drag, by optimizing torso, shoulder, head and elbow position without compromising comfort when sitting still on the bike. The difference in recapturing the aerodynamic reference position during cycling after predefined deviations from the reference position at different intensities was analysed for 14 participants between three interventions, consisting of 1) vibrotactile feedback with a margin of error of 1.5% above the calibrated reference projected frontal area, 2) vibrotactile feedback with a margin of 3%, and 3) no feedback. The reference position is significantly more accurately achieved using vibrotactile feedback compared to no feedback (p < 0.001), but there is no significant difference between the 1.5% and 3% margin (p = 0.11) in terms of relative projected frontal area during cycling compared to the calibrated reference position (1.5% margin -0.46 ± 1.76%, 3% margin -0.01 ± 2.01%, no feedback 2.59 ± 3.29%). The results demonstrate that vibrotactile feedback can have an added value in assisting and correcting cyclists in recapturing their aerodynamic reference position.

Vibrotactile Feedback During Physical Exercise: Perception of Vibrotactile Cues in Cycling.

The aim of this study was to investigate the perception of vibrotactile signals during physical exercise by comparing differences in recognition between stationary and cycling positions. The impact of physical exercise on the ability to perceive vibrotactile cues is unknown, whereas the recognition in stationary position has been shown previously. Vibrating elements were positioned at 3 locations on the thighs and spine of 9 athletes to apply various vibrotactile cues. Subjects performed at 0, 50, 70 and 90% of their maximal cycling power output and denoted the interpretation of the vibration signals on a touchscreen. The results show a similarity in correct recognition between stationary position and physical exercise for the thighs and spine (p>0.1) and demonstrate a decrease in response time for 70 and 90% levels of physical exercise compared to 0 and 50% (p<0.001). Furthermore, vibrotactile signals at the spine are noticed more accurately and more rapidly compared to the thighs (p<0.01). These results suggest that vibrotactile feedback also has potential in applications during physical exercise. The potential use of vibrotactile feedback can be in cycling for, among other, correcting the aerodynamic position. Applications in other sports and health-related domains are feasible as well.

Assessment of Forces in Intradermal Injection Devices: Hydrodynamic Versus Human Factors.

PURPOSE: The force that has to be exerted on the plunger for administering a given amount of fluid in a given time, has an important influence on comfort for the subject and usability for the administrator in intradermal drug delivery. The purpose of this study is to model those forces that are subject-independent, by linking needle and syringe geometry to the force required for ejecting a given fluid at a given ejection rate. MATERIAL AND METHODS: We extend the well-known Hagen-Poiseuille formula to predict pressure drop induced by a fluid passing through a cylindrical body. The model investigates the relation between the pressure drop in needles and the theoretic Hagen-Poiseuille prediction and is validated in fifteen needles from 26G up to 33G suited for intradermal drug delivery. We also provide a method to assess forces exerted by operators in real world conditions. RESULTS: The model is highly linear in each individual needle with R-square values ranging from 75% up to 99.9%. Ten out of fifteen needles exhibit R-square values above 99%. A proof-of-concept for force assessment is provided by logging forces in operators in real life conditions. CONCLUSIONS: The force assessment method and the model can be used to pinpoint needle geometry for intradermal injection devices, tuning comfort for subjects and usability for operators.

Vibrotactile feedback as a tool to improve motor learning and sports performance: a systematic review.

BACKGROUND: Evidence concerning the use of vibrotactile feedback for acquiring and learning new motor skills is limited. Although various concepts and applications for tactile feedback have been proposed, little is known about the suitability of this feedback mechanism in sports training. AIM: The goal of this systematic review was to gather knowledge on the efficacy of the use of vibrotactile feedback in improving sports performance skills. DESIGN: Systematic review. METHODS: Comprehensively searched databases were: PubMed, Cochrane and Web of Science. Studies investigating the effects of using vibrotactile feedback in sports training in healthy subjects were included in this review. RESULTS: No consensus was found regarding the positive effectiveness on performance using vibrotactile feedback in a sports context. No evidence was found that the addition of tactile feedback is effective for acquiring new motor skills. None of the studies show a significant learning effect.

Individual Impact Magnitude vs. Cumulative Magnitude for Estimating Concussion Odds.

Helmeted impact devices have allowed researchers to investigate the biomechanics of head impacts in vivo. While increased impact magnitude has been associated with greater concussion risk, a definitive concussive threshold has not been established. It is likely that concussion risk is not determined by a single impact itself, but a host of predisposing factors. These factors may include genetics, fatigue, and/or prior head impact exposure. The objective of the current paper is to investigate the association between cumulative head impact magnitude and concussion risk. It is hypothesized that increased cumulative magnitudes will be associated with greater concussion risk. This retrospective analysis included participants that were recruited from regional high-schools in Illinois and Michigan from 2007 to 2014 as part of an ongoing study on concussion biomechanics. Across seven seasons, 185 high school football athletes were instrumented with the Head Impact Telemetry system. Out of 185 athletes, 31 (17%) sustained a concussion, with two athletes sustaining two concussions over the study period, yielding 33 concussive events. The system recorded 78,204 impacts for all concussed players. Linear acceleration, rotational acceleration, and head impact telemetry severity profile (HITsp) magnitudes were summed within five timeframes: the day of injury, three days prior to injury, seven days prior to injury, 30 days prior to injury, and prior in-season exposure. Logistic regressions were modeled to explain concussive events based on the singular linear acceleration, rotational acceleration, and HITsp event along with the calculated summations over time. Linear acceleration, rotational acceleration, and HITsp all produced significant models estimating concussion (p < 0.05). The strongest estimators of a concussive impact were the linear acceleration (OR = 1.040, p < 0.05), rotational acceleration (OR = 1.001, p < 0.05), and HITsp (OR = 1.003, p < 0.05) for the singular impact rather than any of the cumulative magnitude calculations. Moreover, no cumulative count measure was significant for linear or rotational acceleration. Results from this investigation support the growing literature indicating cumulative magnitude is not related to concussion likelihood. Cumulative magnitude is a simplistic measure of the total exposure sustained by a player over a given period. However, this measure is limited as it assumes the brain is a static structure unable to undergo self-repair. Future research should consider how biological recovery between impacts may influence concussion risk.