Search

The prevalence of machine learning (ML) tools combined with the ever-growing number of prediction algorithms has created the need for efficient approaches to their selection for small-scale datasets. In this paper, we present a novel Automated Machine Learning (AutoML) solution that efficiently searches a prediction space formed from thousands of ML algorithm pipelines. We distill the high-dimensional data to a low-dimensional configuration space that is efficiently sampled through Bayesian optimization. We demonstrate how the automatically organized space serves as a strong prior to build an AutoML system and deploy it on smaller scale data. We further demonstrate how the space organization can optimize performance while minimizing computational and time resources. As a proof of principle, we apply the prediction space approach to EEG data from a prospective study of n=216 infants with and without a family history of autism.

Lithography-based three-dimensional (3D) printing technologies allow high spatial resolution that exceeds that of typical extrusion-based bioprinting approaches, allowing to better mimic the complex architecture of biological tissues. Additionally, lithographic printing via digital light processing (DLP) enables fabrication of free-form lattice and patterned structures which cannot be easily produced with other 3D printing approaches. While significant progress has been dedicated to the development of cell-laden bioinks for extrusion-based bioprinting, less attention has been directed towards the development of cyto-compatible bio-resins and their application in lithography-based biofabrication, limiting the advancement of this promising technology. In this study, we developed a new bio-resin based on methacrylated poly(vinyl alcohol) (PVA-MA), gelatin-methacryloyl (Gel-MA) and a transition metal-based visible light photoinitiator. The utilization of a visible light photo-initiating system displaying high molar absorptivity allowed the bioprinting of constructs with high resolution features, in the range of 25-50 μm. Biofunctionalization of the resin with 1 wt% Gel-MA allowed long term survival (>90%) of encapsulated cells up to 21 d, and enabled attachment and spreading of endothelial cells seeded on the printed hydrogels. Cell-laden hydrogel constructs of high resolution with complex and ordered architecture were successfully bioprinted, where the encapsulated cells remained viable, homogenously distributed and functional. Bone and cartilage tissue synthesis was confirmed by encapsulated stem cells, underlining the potential of these DLP-bioprinted hydrogels for tissue engineering and biofabrication. Overall, the PVA-MA/Gel-MA bio-resin is a promising material for biofabrication and provides important cues for the further development of lithography-based bioprinting of complex, free-form living tissue analogues.

Polymer engineering, such as in three-dimensional (3D) printing, is rapidly gaining popularity, not only in the scientific and medical fields but also in the community in general. However, little is known about the toxicity of engineered materials. Therefore, we assessed the toxicity of 3D-printed and molded parts from five different polymers commonly used for prototyping, fabrication of organ-on-a-chip platforms, and medical devices. Toxic effects of PIC100, E-Shell200, E-Shell300, polydimethylsiloxane, and polystyrene (PS) on early bovine embryo development, on the transactivation of estrogen receptors were assessed, and possible polymer-leached components were identified by mass spectrometry. Embryo development beyond the two-cell stage was inhibited by PIC100, E-Shell200, and E-Shell300 and correlated to the released amount of diethyl phthalate and polyethylene glycol. Furthermore, all polymers (except PS) induced estrogen receptor transactivation. The released materials from PIC100 inhibited embryo cleavage across a confluent monolayer culture of oviduct epithelial cells and also inhibited oocyte maturation. These findings highlight the need for cautious use of engineered polymers for household 3D printing and bioengineering of culture and medical devices and the need for the safe disposal of used devices and associated waste.

International audience

The oviduct provides the natural micro-environment for gamete interaction, fertilization and early embryo development in mammals, such as the cow. In conventional culture systems, bovine oviduct epithelial cells (BOEC) undergo a rapid loss of essential differentiated cell properties; we aimed to develop a more physiological in vitro oviduct culture system capable of supporting fertilization. U-shaped chambers were produced using stereo-lithography and mounted with polycarbonate membranes, which were used as culture inserts for primary BOECs. Cells were grown to confluence and cultured at an air–liquid interface for 4 to 6 weeks and subsequently either fixed for immune staining, incubated with sperm cells for live-cell imaging, or used in an oocyte penetration study. Confluent BOEC cultures maintained polarization and differentiation status for at least 6 weeks. When sperm and oocytes were introduced into the system, the BOECs supported oocyte penetration in the absence of artificial sperm capacitation factors while also preventing polyspermy and parthenogenic activation, both of which occur in classical in vitro fertilization systems. Moreover, this “oviduct-on-a-chip” allowed live imaging of sperm-oviduct epithelium binding and release. Taken together, we describe for the first time the use of 3D-printing as a step further on bio-mimicking the oviduct, with polarized and differentiated BOECs in a tubular shape that can be perfused or manipulated, which is suitable for live imaging and supports in vitro fertilization.