For human children as well as machine learning systems, a key challenge in learning a word is linking the word to the visual phenomena it describes. We explore this aspect of word learning by using the performance of computer vision systems as a proxy for the difficulty of learning a word from visual cues. We show that the age at which children acquire different categories of words is correlated with the performance of visual classification and captioning systems, over and above the expected effects of word frequency. The performance of the computer vision systems is correlated with human judgments of the concreteness of words, which are in turn a predictor of children's word learning, suggesting that these models are capturing the relationship between words and visual phenomena.

Halide perovskites are promising light-absorbing materials for high-efficiency solar cells, while the crystalline phase of halide perovskites may influence the device's efficiency and stability. In this work, we investigated the thermally driven phase transition of perovskite (CsPbI_xBr_3-x), which was confirmed by electron diffraction and high-resolution transmission electron microscopy results. CsPbI_xBr_3-x transitioned from δ phase to α phase when heated, and the γ phase was obtained when the sample was cooled down. The γ phase was stable as long as it was isolated from humidity and air. A template matching-based data analysis method enabled visualization of the thermally driven phase evolution of perovskite during heating. We also proposed a possible atomic movement in the process of phase transition based on our in situ heating experimental data. The results presented here may improve our understanding of the thermally driven phase transition of perovskite as well as provide a protocol for big-data analysis of in situ experiments.

Gastric cancer (GC) is one of the most common and lethal types of cancer affecting over one million people, leading to 768,793 deaths globally in 2020 alone. The key for improving the survival rate lies in reliable screening and early diagnosis. Existing techniques including barium-meal gastric photofluorography and upper endoscopy can be costly and time-consuming and are thus impractical for population screening. We look instead for small extracellular vesicles (sEVs, currently also referred as exosomes) sized ⌀ 30-150 nm as a candidate. sEVs have attracted a significantly higher level of attention during the past decade or two because of their potentials in disease diagnoses and therapeutics. Here, we report that the composition information of the collective Raman-active bonds inside sEVs of human donors obtained by surface-enhanced Raman spectroscopy (SERS) holds the potential for non-invasive GC detection. SERS was triggered by the substrate of gold nanopyramid arrays we developed previously. A machine learning-based spectral feature analysis algorithm was developed for objectively distinguishing the cancer-derived sEVs from those of the non-cancer sub-population. sEVs from the tissue, blood, and saliva of GC patients and non-GC participants were collected (n = 15 each) and analyzed. The algorithm prediction accuracies were reportedly 90, 85, and 72%. Leave-a-pair-of-samples out validation was further performed to test the clinical potential. The area under the curve of each receiver operating characteristic curve was 0.96, 0.91, and 0.65 in tissue, blood, and saliva, respectively. In addition, by comparing the SERS fingerprints of individual vesicles, we provided a possible way of tracing the biogenesis pathways of patient-specific sEVs from tissue to blood to saliva. The methodology involved in this study is expected to be amenable for non-invasive detection of diseases other than GC.

High-precision isolation of small extracellular vesicles (sEVs) from biofluids is essential toward developing next-generation liquid biopsies and regenerative therapies. However, current methods of sEV separation require specialized equipment and time-consuming protocols and have difficulties producing highly pure subpopulations of sEVs. Here, we present Acoustic Nanoscale Separation via Wave-pillar Excitation Resonance (ANSWER), which allows single-step, rapid (<10 min), high-purity (>96% small exosomes, >80% exomeres) fractionation of sEV subpopulations from biofluids without the need for any sample preprocessing. Particles are iteratively deflected in a size-selective manner via an excitation resonance. This previously unidentified phenomenon generates patterns of virtual, tunable, pillar-like acoustic field in a fluid using surface acoustic waves. Highly precise sEV fractionation without the need for sample preprocessing or complex nanofabrication methods has been demonstrated using ANSWER, showing potential as a powerful tool that will enable more in-depth studies into the complexity, heterogeneity, and functionality of sEV subpopulations.

Previous efforts to evaluate the detection of human papilloma viral (HPV) DNA in whole saliva as a diagnostic measure for HPV-associated oropharyngeal cancer (HPV-OPC) have not shown sufficient clinical performance. We hypothesize that salivary exosomes are packaged with HPV-associated biomarkers, and efficient enrichment of salivary exosomes through isolation can enhance diagnostic and prognostic performance for HPV-OPC. In this study, an acoustofluidic (the fusion of acoustics and microfluidics) platform was developed to perform size-based isolation of salivary exosomes. These data showed that this platform is capable of consistently isolating exosomes from saliva samples, regardless of viscosity variation and collection method. Compared with the current gold standard, differential centrifugation, droplet digital RT-PCR analysis showed that the average yield of salivary exosomal small RNA from the acoustofluidic platform is 15 times higher. With this high-yield exosome isolation platform, we show that HPV16 DNA could be detected in isolated exosomes from the saliva of HPV-associated OPC patients at 80% concordance with tissues/biopsies positive for HPV16. Overall, these data demonstrated that the acoustofluidic platform can achieve high-purity and high-yield salivary exosome isolation for downstream salivary exosome-based liquid biopsy applications. Additionally, HPV16 DNA sequences in HPV-OPC patients are packaged in salivary exosomes and their isolation will enhance the detection of HPV16 DNA.

The extracellular RNA communication consortium (ERCC) is an NIH-funded program aiming to promote the development of new technologies, resources, and knowledge about exRNAs and their carriers. After Phase 1 (2013-2018), Phase 2 of the program (ERCC2, 2019-2023) aims to fill critical gaps in knowledge and technology to enable rigorous and reproducible methods for separation and characterization of both bulk populations of exRNA carriers and single EVs. ERCC2 investigators are also developing new bioinformatic pipelines to promote data integration through the exRNA atlas database. ERCC2 has established several Working Groups (Resource Sharing, Reagent Development, Data Analysis and Coordination, Technology Development, nomenclature, and Scientific Outreach) to promote collaboration between ERCC2 members and the broader scientific community. We expect that ERCC2's current and future achievements will significantly improve our understanding of exRNA biology and the development of accurate and efficient exRNA-based diagnostic, prognostic, and theranostic biomarker assays.