The growing threat of antimicrobial resistance makes rapid and sensitive detection of bacterial pathogens in human samples an urgent need. In the case of blood infection, or bacteremia, standard pathogen...

Artificial intelligence (AI) is gradually transforming healthcare. However, despite its promised benefits, AI in healthcare also raises a number of ethical, legal and social concerns. Compliance by design (CbD) has been proposed as one way of addressing some of these concerns. In the context of healthcare, CbD efforts could focus on building compliance with existing clinical guidelines (CGs), given that they provide the best practices identified according to evidence-based medicine. In this paper we use the example of AI-based clinical decision support systems (CDSS) to theoretically examine whether medical AI tools could be designed to be inherently compliant with CGs, and implication for ethics and trust. We argue that AI-based CDSS systematically complying with CGs when applied to specific patient cases are not desirable, as CGs, despite their usefulness in guiding medical decision-making, are only recommendations on how to diagnose and treat medical conditions. We thus propose a new understanding of CbD for CGs as a sociotechnical program supported by AI that applies to the whole clinical decision-making process rather than just understanding CbD for CGs as a process located only within the AI tool. This implies taking into account emerging knowledge from actual clinical practices to put CGs in perspective, reflexivity from users regarding the information needed for decision-making, as well as a shift in the design culture, from AI as a stand-alone tool to AI as an in-situ service located within particular healthcare settings.

This one-page abstract briefly presents the current status of my doctoral research in philosophy and ethics about the introduction of AI-based systems (AIS) in the field of healthcare. My main aim is to explore the ways in which ethical AIS for medicine could be ethically designed. I will briefly present the background of my research, my main activities so far and the upcoming avenues for research I envision.

Bacterial resistance is becoming a global issue due to its rapid growth. Potential new drugs as antimicrobial peptides (AMPs) are considered for several decades as promising candidates to circumvent this threat. Nonetheless, AMPs have also been used more recently in other settings such as molecular probes grafted on biosensors able to detect whole bacteria. Rapid, reliable and cost-efficient diagnostic tools for bacterial infection could prevent the spread of the pathogen from the earliest stages. Biosensors based on AMPs would enable easy monitoring of potentially infected samples, thanks to their powerful versatility and integrability in pre-existent settings. AMPs, which show a broad spectrum of interactions with bacterial membranes, can be tailored in order to design ubiquitous biosensors easily adaptable to clinical settings. This review aims to focus on the state of the art of AMPs used as the recognition elements of whole bacteria in label-free biosensors with a particular focus on the characteristics obtained in terms of threshold, volume of sample analysable and medium, in order to assess their workability in real-world applications.

L’analyse microbiologique pour confirmer l’absence de bactéries dans des échantillons biologiques normalement sains, comme le sang, est une routine dans de nombreux laboratoires. En effet, la présence de bactéries dans le sang, appelée bactériémie, peut avoir des conséquences très graves, voire mortelles pour le patient. Le protocole standard pour la détection des bactériémies repose jusqu’ici sur l’enrichissement des échantillons sanguins prélevés sur les patients lors de l’hémoculture, afin d’obtenir une population suffisante pour analyse. La lenteur de ce procédé retarde ainsi de parfois plusieurs jours le diagnostic et donc l’adaptation du traitement antibiotique administré au patient. Ces dernières décennies, des techniques comme l’identification par spectrométrie de masse ou les analyses moléculaires, ont permis de diminuer le délai requis pour identifier les pathogènes en cause. Dans ce contexte, l’emploi de biocapteurs est également une alternative. Ce travail propose d’inclure des sondes à large spectre dans un capteur optique par imagerie SPR (résonance de plasmons de surface). Ce système est déjà développé pour la reconnaissance spécifique de pathogènes au cours de leur croissance dans le sang. Les nouveaux ligands proposés et évalués sont les peptides antimicrobiens (PAM). Ces courts peptides cationiques et amphiphiles, présentent l’avantage d’un large spectre d’interaction couplé à une haute stabilité (chimique, thermique et séchage) comparativement aux anticorps employés jusqu’ici. Leur immobilisation sur des prismes SPRI permet d’évaluer simultanément l’affinité de plusieurs PAM à la même souche bactérienne. Les biocapteurs ainsi préparés ont permis de détecter des souches pathogènes d’Escherichia coli et Staphylococcus aureus en milieu de culture simple, comme en plasma et en sang dilué au milieu d’hémoculture. Le système obtenu permet la détection des pathogènes présents à une concentration initiale de l’ordre de 1 UFC.ml-1, en moins de 24 heures et quel que soit le milieu. Enfin, la mise en place d’analyses statistiques multidimensionnelles a abouti à une classification cohérente des espèces ciblées en milieu simple, comme en sang. Ces résultats montrent le potentiel de ce système pour parvenir à développer un biocapteur à large spectre capable à la fois de détecter mais aussi d’identifier par affinité croisée des pathogènes bactériens.

Fast detection of bacteria in samples presumed to be un-contaminated, such as blood, is of great importance. Indeed, rapid diagnosis allows the set-up of appropriate antibiotic treatment. Besides clinical issues, there are many other domains, such as food processing or drug manufacturing, where the strict absence of any bacteria has to be assessed. Because the bacterial load found in most contaminated samples is often below the limit of detection for currently validated assays, a preliminary enrichment step is required to allow bacterial multiplication before proceeding to the analysis step, whatever it might be cultural, immunological or molecular methods. In this study, we describe the use of a biosensor for single-step bacteria detection. The whole analysis is performed in less than 20 h, during the growth phase of the micro-organisms, using an array of antimicrobial peptides (AMPs) coupled with a surface plasmon resonance imager (SPRI). A wide range of bacterial strains are assayed, showing differentiated affinity patterns with the immobilized peptides, which are confirmed by multivariate analysis. This work establishes the evidence that antimicrobial peptides, mostly used so far in the antibiotic drug industry, are suited for the wide-spectrum detection of unknown bacteria in samples, even at very low initial loads. Moreover, the small set of AMPs that were assayed provided a specific affinity profile for each pathogen, as confirmed by multivariate analyses. Furthermore, this work opens up the possibility of applying this method in more complex and relevant samples such as foodstuff, urine or blood.

In this work we demonstrate the use of Fully Depleted Silicon On Insulator (FDSOI) transistors as pH sensors with a 23 nm silicon nitride sensing layer built in the Back-End-Of-Line (BEOL). The back end process to deposit the sensing layer and fabricate the electrical structures needed for testing is detailed. A series of tests employing different pH buffer solutions has been performed on transistors of different geometries, controlled via the back gate. The main findings show a shift of the drain current (I-D) as a function of the back gate voltage (V-B) when different pH buffer solutions are probed in the range of pH 6 to pH 8. This shift is observed at V-B voltages swept from 0 V to 3 V, demonstrating the sensor operation at low voltage. A high sensitivity of up to 250 mV/pH unit (more than 4-fold larger than Nernstian response) is observed on FDSOI MOS transistors of 0.06 mu m gate length and 0.08 mu m gate width. (C) 2017 Elsevier Ltd. All rights reserved.