A machine learning-based framework to design capillary-driven networks.

We present a novel approach for the design of capillary-driven microfluidic networks using a machine learning genetic algorithm (ML-GA). This strategy relies on a user-friendly 1D numerical tool specifically developed to generate the necessary data to train the ML-GA. This 1D model was validated using analytical results issued from a Y-shaped capillary network and experimental data. For a given microfluidic network, we defined the objective of the ML-GA to obtain the set of geometric parameters that produces the closest matching results against two prescribed curves of delivered volume against time. We performed more than 20 generations of 10 000 simulations to train the ML-GA and achieved the optimal solution of the inverse design problem. The optimisation took less than 6 hours, and the results were successfully validated using experimental data. This work establishes the utility of the presented method for the fast and reliable design of complex capillary-driven devices, enabling users to optimise their designs via an easy-to-use 1D numerical tool and machine learning technique.

Ventilator output splitting interface 'ACRA': Description and evaluation in lung simulators and in an experimental ARDS animal model.

The current COVID-19 pandemic has led the world to an unprecedented global shortage of ventilators, and its sharing has been proposed as an alternative to meet the surge. This study outlines the performance of a preformed novel interface called 'ACRA', designed to split ventilator outflow into two breathing systems. The 'ACRA' interface was built using medical use approved components. It consists of four unidirectional valves, two adjustable flow-restrictor valves placed on the inspiratory limbs of each unit, and one adjustable PEEP valve placed on the expiratory limb of the unit that would require a greater PEEP. The interface was interposed between a ventilator and two lung units (phase I), two breathing simulators (phase II) and two live pigs with heterogeneous lung conditions (phase III). The interface and ventilator adjustments tested the ability to regulate individual pressures and the resulting tidal volumes. Data were analyzed using Friedman and Wilcoxon tests test (p < 0.05). Ventilator outflow splitting, independent pressure adjustments and individual tidal volume monitoring were feasible in all phases. In all experimental measurements, dual ventilation allowed for individual and tight adjustments of the pressure, and thus volume delivered to each paired lung unit without affecting the other unit's ventilation-all the modifications performed on the ventilator equally affected both paired lung units. Although only suggested during a dire crisis, this experiment supports dual ventilation as an alternative worth to be considered.

Exhalatory dynamic interactions between patients connected to a shared ventilation device.

In this work a shared pressure-controlled ventilation device for two patients is considered. By the use of different valves incorporated to the circuit, the device enables the restriction of possible cross contamination and the individualization of tidal volumes, driving pressures, and positive end expiratory pressure PEEP. Possible interactions in the expiratory dynamics of different pairs of patients are evaluated in terms of the characteristic exhalatory times. These characteristic times can not be easily established using simple linear lumped element models. For this purpose, a 1D model using the Hydraulic and Mechanical libraries in Matlab Simulink was developed. In this sense, experiments accompany this study to validate the model and characterize the different valves of the circuit. Our results show that connecting two patients in parallel to a ventilator always resulted in delays of time during the exhalation. The size of this effect depends on different parameters associated with the patients, the circuit and the ventilator. The dynamics of the exhalation of both patients is determined by the ratios between patients exhalatory resistances, compliances, driving pressures and PEEPs. Adverse effects on exhalations became less noticeable when respiratory parameters of both patients were similar, flow resistances of valves added to the circuit were negligible, and when the ventilator exhalatory valve resistance was also negligible. The asymmetries of driving pressures, compliances or resistances exacerbated the possibility of auto-PEEP and the increase in relaxation times became greater in one patient than in the other. In contrast, exhalatory dynamics were less sensitive to the ratio of PEEP imposed to the patients.

Self-sustaining process through streak generation in a flat-plate boundary layer.

The existence of a self-sustaining process between streamwise vortices and streaks has been suggested at moderate Reynolds numbers. Such a mechanism has never been demonstrated experimentally. Using small cylinders as vortex generators to create streamwise counterrotating vortices, we show, through the characterization of the spatial transient growth of the energy of the longitudinal and spanwise velocity perturbations, that such a mechanism exists above a given Reynolds number. From instantaneous particle image velocimetry fields in a horizontal plane, we show that the self-sustaining process can also be associated with the longitudinal destabilization of streamwise streaks.