Magneto-Nanomechanical Array Biosensor for Ultrasensitive Detection of Oncogenic Exosomes for Early Diagnosis of Cancers.

Exosomes are a class of nanoscale vesicles secreted by cells, which contain abundant information closely related to parental cells. The ultrasensitive detection of cancer-derived exosomes is highly significant for early non-invasive diagnosis of cancer. Here, an ultrasensitive nanomechanical sensor is reported, which uses a magnetic-driven microcantilever array to selectively detect oncogenic exosomes. A magnetic force, which can produce a far greater deflection of microcantilever than that produced by the intermolecular interaction force even with very low concentrations of target substances, is introduced. This method reduced the detection limit to less than 10 exosomes mL-1 . Direct detection of exosomes in the serum of patients with breast cancer and in healthy people showed a significant difference. This work improved the sensitivity by five orders of magnitude as compared to that of traditional nanomechanical sensing based on surface stress mode. This method can be applied parallelly for highly sensitive detection of other microorganisms (such as bacteria and viruses) by using different probe molecules, which can provide a supersensitive detection approach for cancer diagnosis, food safety, and SARS-CoV-2 infection.

Nanomechanical assay for ultrasensitive and rapid detection of SARS-CoV-2 based on peptide nucleic acid.

The massive global spread of the COVID-19 pandemic makes the development of more effective and easily popularized assays critical. Here, we developed an ultrasensitive nanomechanical method based on microcantilever array and peptide nucleic acid (PNA) for the detection of severe acute respiratory syndrome-coronavirus-2 (SARS-CoV-2) RNA. The method has an extremely low detection limit of 0.1 fM (105 copies/mL) for N-gene specific sequence (20 bp). Interestingly, it was further found that the detection limit of N gene (pharyngeal swab sample) was even lower, reaching 50 copies/mL. The large size of the N gene dramatically enhances the sensitivity of the nanomechanical sensor by up to three orders of magnitude. The detection limit of this amplification-free assay method is an order of magnitude lower than RT-PCR (500 copies/mL) that requires amplification. The non-specific signal in the assay is eliminated by the in-situ comparison of the array, reducing the false-positive misdiagnosis rate. The method is amplification-free and label-free, allowing for accurate diagnosis within 1 h. The strong specificity and ultra-sensitivity allow single base mutations in viruses to be distinguished even at very low concentrations. Also, the method remains sensitive to fM magnitude lung cancer marker (miRNA-155). Therefore, this ultrasensitive, amplification-free and inexpensive assay is expected to be used for the early diagnosis of COVID-19 patients and to be extended as a broad detection tool. Electronic Supplementary Material: Supplementary material (experimental section, N gene sequences and all nucleic acid sequences used in the study, Figs. S1-S6, and Tables S1-S3) is available in the online version of this article at 10.1007/s12274-022-4333-3.

Capturing Time Dynamics From Speech Using Neural Networks for Surgical Mask Detection.

The importance of detecting whether a person wears a face mask while speaking has tremendously increased since the outbreak of SARS-CoV-2 (COVID-19), as wearing a mask can help to reduce the spread of the virus and mitigate the public health crisis. Besides affecting human speech characteristics related to frequency, face masks cause temporal interferences in speech, altering the pace, rhythm, and pronunciation speed. In this regard, this paper presents two effective neural network models to detect surgical masks from audio. The proposed architectures are both based on Convolutional Neural Networks (CNNs), chosen as an optimal approach for the spatial processing of the audio signals. One architecture applies a Long Short-Term Memory (LSTM) network to model the time-dependencies. Through an additional attention mechanism, the LSTM-based architecture enables the extraction of more salient temporal information. The other architecture (named ConvTx) retrieves the relative position of a sequence through the positional encoder of a transformer module. In order to assess to which extent both architectures can complement each other when modelling temporal dynamics, we also explore the combination of LSTM and Transformers in three hybrid models. Finally, we also investigate whether data augmentation techniques, such as, using transitions between audio frames and considering gender-dependent frameworks might impact the performance of the proposed architectures. Our experimental results show that one of the hybrid models achieves the best performance, surpassing existing state-of-the-art results for the task at hand.