Search for: All records

Sleep staging has a very important role in diagnosing patients with sleep disorders. In general, this task is very time-consuming for physicians to perform. Deep learning shows great potential to automate this process and remove physician bias from decision making. In this study, we aim to identify recent trends on performance improvement and the causes for these trends. Recent papers on sleep stage classification and interpretability are investigated to explore different modeling and data manipulation techniques, their efficiency, and recent advances. We identify an improvement in performance up to 12% on standard datasets over the last 5 years. The improvements in performance do not appear to be necessarily correlated to the size of the models, but instead seem to be caused by incorporating new architectural components, such as the use of transformers and contrastive learning.

 

Microphysiological systems (MPS) incorporate physiologically relevant microanatomy, mechanics, and cells to mimic tissue function. Reproducible and standardized in vitro models of tissue barriers, such as the blood-tissue interface (BTI), are critical for next-generation MPS applications in research and industry. Many models of the BTI are limited by the need for semipermeable membranes, use of homogenous cell populations, or 2D culture. These factors limit the relevant endothelial-epithelial contact and 3D transport, which would best mimic the BTI. Current models are also difficult to assemble, requiring precise alignment and layering of components. The work reported herein details the engineering of a BTI-on-a-chip (BTI Chip) that addresses current disadvantages by demonstrating a single layer, membrane-free design. Laminar flow profiles, photocurable hydrogel scaffolds, and human cell lines were used to construct a BTI Chip that juxtaposes an endothelium in direct contact with a 3D engineered tissue. A biomaterial composite, gelatin methacryloyl and 8-arm polyethylene glycol thiol, was used for in situ fabrication of a tissue structure within a Y-shaped microfluidic device. To produce the BTI, a laminar flow profile was achieved by flowing a photocurable precursor solution alongside phosphate buffered saline. Immediately after stopping flow, the scaffold underwent polymerization through a rapid exposure to UV light (<300 mJ/cm2). After scaffold formation, blood vessel endothelial cells were introduced and allowed to adhere directly to the 3D tissue scaffold, without barriers or phase guides. Fabrication of the BTI Chip was demonstrated in both an epithelial tissue model and blood-brain barrier (BBB) model. In the epithelial model, scaffolds were seeded with human dermal fibroblasts. For the BBB models, scaffolds were seeded with the immortalized glial cell line, SVGP12. The BTI Chip microanatomy was analyzed post facto by immunohistochemistry, showing the uniform production of a patent endothelium juxtaposed with a 3D engineered tissue. Fluorescent tracer molecules were used to characterize the permeability of the BTI Chip. The BTI Chips were challenged with an efflux pump inhibitor, cyclosporine A, to assess physiological function and endothelial cell activation. Operation of physiologically relevant BTI Chips and a novel means for high-throughput MPS generation was demonstrated, enabling future development for drug candidate screening and fundamental biological investigations. 

Most affinity-based biosensors are designed to be single-use devices, based on the measurement of irreversible binding events, which makes longitudinal monitoring resource-intensive, and typically prohibits the measurement of analyte fluctuations over time using the same device. Selective reversal of biorecognition events, i.e., regeneration, may enable repeated and longitudinal use of affinity-based biosensors; however, typical regeneration methods utilize additional chemical reagents, requiring longer processing times and increasing the likelihood of operator error. The development of a “solid-state” regeneration method provides significant value for extending the utility of affinity-based biosensors, such as electrochemical immunosensors and aptasensors. Herein, we report the characterization of a method for electronically controlling pH without additional reagents. Palladium was used to induce pH swings in aqueous electrolytes and buffers by application of an electric potential. The developed system was able to affect acidic and basic pH changes of ± 4. The efficacy of this method was further demonstrated by reversing common affinity-binding complexes and compared to conventional glycine-based regeneration. 

Gene therapies have shown great promise for the potential treatment of a broad range of diseases. Adeno-associated viruses (AAVs) are popular gene vectors because of their ability to target specific tissues, and they have demonstrated high transduction efficiencies in multiple neurological targets. While these therapeutics hold great promise, their biomanufacturing has limited potential cost-reduction and more widespread adoption. Herein, we report the preliminary development of an immunosensor for measuring the titer of adeno-associated virus 2 (AAV2), which may be deployed for rapid quantification of product yield during AAV biomanufacturing. We functionalized an interdigitated electrode array with anti-AAV2 antibodies, and electrochemical impedance spectroscopy was employed to investigate the response to AAV2 titer. A Faradaic sensing principle was utilized, in which the charge transfer resistance (Rct) of an electrochemical reporter was monitored after capture of AAV2 on the surface of the sensor. A linear response was measured over titers 1012 - 1013 capsids/mL. 

The recent uptick in the approval of ex vivo cell therapies highlights the relevance of lentivirus (LV) as an enabling viral vector of modern medicine. As labile biologics, however, LVs pose critical challenges to industrial biomanufacturing. In particular, LV purification—currently reliant on filtration and anion-exchange or size-exclusion chromatography—suffers from long process times and low yield of transducing particles, which translate into high waiting time and cost to patients. Seeking to improve LV downstream processing, this study introduces peptides targeting the enveloped protein Vesicular stomatitis virus G (VSV-G) to serve as affinity ligands for the chromatographic purification of LV particles. An ensemble of candidate ligands was initially discovered by implementing a dual-fluorescence screening technology and a targeted in silico approach designed to identify sequences with high selectivity and tunable affinity. The selected peptides were conjugated on Poros resin and their LV binding-and-release performance was optimized by adjusting the flow rate, composition, and pH of the chromatographic buffers. Ligands GKEAAFAA and SRAFVGDADRD were selected for their high product yield (50%–60% of viral genomes; 40%–50% of HT1080 cell-transducing particles) upon elution in PIPES buffer with 0.65 M NaCl at pH 7.4. The peptide-based adsorbents also presented remarkable values of binding capacity (up to 3·109 TU per mL of resin, or 5·1011 vp per mL of resin, at the residence time of 1 min) and clearance of host cell proteins (up to a 220-fold reduction of HEK293 HCPs). Additionally, GKEAAFAA demonstrated high resistance to caustic cleaning-in-place (0.5 M NaOH, 30 min) with no observable loss in product yield and quality. 

Recyclable and biodegradable microelectronics, i.e., “green” electronics, are emerging as a viable solution to the global challenge of electronic waste. Specifically, flexible circuit boards represent a prime target for materials development and increasing the utility of green electronics in biomedical applications. Circuit board substrates and packaging are good dielectrics, mechanically and thermally robust, and are compatible with microfabrication processes. Poly(octamethylene maleate (anhydride) citrate) (POMaC) – a citric acid-based elastomer with tunable degradation and mechanical properties – presents a promising alternative for circuit board substrates and packaging. Here, we report the characterization of Elastomeric Circuit Boards (ECBs). Synthesis and processing conditions were optimized to achieve desired degradation and mechanical properties for production of stretchable circuits. ECB traces were characterized and exhibited sheet resistance of 0.599 Ω cm⁻², crosstalk distance of <0.6 mm, and exhibited stable 0% strain resistances after 1000 strain cycles to 20%. Fabrication of single layer and encapsulated ECBs was demonstrated.

 

A new microphysiological system allows precise control and monitoring of oxygen levels at the cell surface to study the impact of hypoxia. Hypoxia pushes human intestinal stem cells (hISCs) into a dormant but reversible proliferative state and primes hISCs to respond to a subset of interleukins that rescues hISC activity. 

Longitudinal fetal health monitoring is essential for high-risk pregnancies. Heart rate and heart rate variability are prime indicators of fetal health. In this work, we implemented two neural network architectures for heartbeat detection on a set of fetal phonocardiogram signals captured using fetal Doppler and a digital stethoscope. We test the efficacy of these networks using the raw signals and the hand-crafted energy from the signal. The results show a Convolutional Neural Network is the most efficient at identifying the S1 waveforms in a heartbeat, and its performance is improved when using the energy of the Doppler signals. We further discuss issues, such as low Signal-to-Noise Ratios (SNR), present in the training of a model based on the stethoscope signals. Finally, we show that we can improve the SNR, and subsequently the performance of the stethoscope, by matching the energy from the stethoscope to that of the Doppler signal.