Semantic-Powered Explainable Model-Free Few-Shot Learning Scheme of Diagnosing COVID-19 on Chest X-ray.

Chest X-ray (CXR) is commonly performed as an initial investigation in COVID-19, whose fast and accurate diagnosis is critical. Recently, deep learning has a great potential in detecting people who are suspected to be infected with COVID-19. However, deep learning resulting with black-box models, which often breaks down when forced to make predictions about data for which limited supervised information is available and lack inter-pretability, still is a major barrier for clinical integration. In this work, we hereby propose a semantic-powered explainable model-free few-shot learning scheme to quickly and precisely diagnose COVID-19 with higher reliability and transparency. Specifically, we design a Report Image Explanation Cell (RIEC) to exploit clinically indicators derived from radiology reports as interpretable driver to introduce prior knowledge at training. Meanwhile, multi-task colla-borative diagnosis strategy (MCDS) is developed to construct [Formula: see text]-way [Formula: see text]-shot tasks, which adopts a cyclic and collaborative training approach for producing better generalization performance on new tasks. Extensive experiments demonstrate that the proposed scheme achieves competitive results (accuracy of 98.91%, precision of 98.95%, recall of 97.94% and F1-score of 98.57%) to diagnose COVID-19 and other pneumonia infected categories, even with only 200 paired CXR images and radiology reports for training. Furthermore, statistical results of comparative experiments show that our scheme provides an interpretable window into the COVID-19 diagnosis to improve the performance of the small sample size, the reliability and transparency of black-box deep learning mod-els. Our source codes will be released on https://github.com/AI-medical-diagnosis-team-of-JNU/SPEMFSL-Diagnosis-COVID-19.

Rapid point-of-care detection of SARS-CoV-2 RNA with smartphone-based upconversion luminescence diagnostics.

Accurate COVID-19 screening via molecular technologies is still hampered by bulky instrumentation, complicated procedure, high cost, lengthy testing time, and the need for specialized personnel. Herein, we develop point-of-care upconversion luminescence diagnostics (PULD), and a streamlined smartphone-based portable platform facilitated by a ready-to-use assay for rapid SARS-CoV-2 nucleocapsid (N) gene testing. With the complementary oligo-modified upconversion nanoprobes and gold nanoprobes specifically hybridized with the target N gene, the luminescence resonance energy transfer effect leads to a quenching of fluorescence intensity that can be detected by the easy-to-use diagnostic system. A remarkable detection limit of 11.46 fM is achieved in this diagnostic platform without the need of target amplification, demonstrating high sensitivity and signal-to-noise ratio of the assay. The capability of the developed PULD is further assessed by probing 9 RT-qPCR-validated SARS-CoV-2 variant clinical samples (B.1.1.529/Omicron) within 20 min, producing reliable diagnostic results consistent with those obtained from a standard fluorescence spectrometer. Importantly, PULD is capable of identifying the positive COVID-19 samples with superior sensitivity and specificity, making it a promising front-line tool for rapid, high-throughput screening and infection control of COVID-19 or other infectious diseases.

Multiplexed detection of SARS-CoV-2 based on upconversion luminescence nanoprobe/MXene biosensing platform for COVID-19 point-of-care diagnostics.

Multiplexed detection is essential in biomedical sciences since it is more efficient and accurate than single-analyte detection. For an accurate early diagnosis of COVID-19, a multiplexed detection strategy is required to avoid false negatives with the existing gold standard assay. Nb2CTx nanosheets were found to efficiently quench the fluorescence emission of lanthanide-doped upconversion luminescence nanoparticles at wavelengths ranging from visible to near-infrared spectrum. Using this broad-spectrum quencher, we developed a label-free FRET-based biosensor for rapid and accurate detection of SARS-CoV-2 RNA. To target ORF and N genes, two types of oligo-modified lanthanide-doped upconversion nanoparticles can be used simultaneously to identify-two sites in one assay via upconversion fluorescence enhancement intensity measurement with detection limits of 15 pM and 914 pM, respectively. Moreover, with multisite cross-validation, this multiplexed and sensitive biosensor is capable of simultaneous and multicolor analysis of two gene fragments of SARS-CoV-2 Omicron variant within minutes in a single homogeneous solution, which significantly improves the detection efficiency. The diagnosis result via our assay is consistent with the PCR result, demonstrating its application in the rapid and accurate screening of multiple genes of SARS-CoV-2 and other infectious diseases.

Deep Learning Mask Face Recognition with Annealing Mechanism

Face recognition (FR) has matured with deep learning, but due to the COVID-19 epidemic, people need to wear masks outside to reduce the risk of infection, making FR a challenge. This study uses the FaceNet approach combined with transfer learning using three different sizes of validated CNN architectures: InceptionResNetV2, InceptionV3, and MobileNetV2. With the addition of the cosine annealing (CA) mechanism, the optimizer can automatically adjust the learning rate (LR) during the model training process to improve the efficiency of the model in finding the best solution in the global domain. The mask face recognition (MFR) method is accomplished without increasing the computational complexity using existing methods. Experimentally, the three models of different sizes using the CA mechanism have a better performance than the fixed LR, step and exponential methods. The accuracy of the three models of different sizes using the CA mechanism can reach a practical level at about 93%.

Rapid point-of-care detection of SARS-CoV-2 RNA with smartphone-based upconversion luminescence diagnostics

Accurate COVID-19 screening via molecular technologies is still hampered by bulky instrumentation, complicated procedure, high cost, lengthy testing time, and the need for specialized personnel. Herein, we develop point-of-care upconversion luminescence diagnostics (PULD), and a streamlined smartphone-based portable platform facilitated by a ready-to-use assay for rapid SARS-CoV-2 nucleocapsid (N) gene testing. With the complementary oligo-modified upconversion nanoprobes and gold nanoprobes specifically hybridized with the target N gene, the luminescence resonance energy transfer effect leads to a quenching of fluorescence intensity that can be detected by the easy-to-use diagnostic system. A remarkable detection limit of 11.46 fM is achieved in this diagnostic platform without the need of target amplification, demonstrating high sensitivity and signal-to-noise ratio of the assay. The capability of the developed PULD is further assessed by probing 9 RT-qPCR-validated SARS-CoV-2 variant clinical samples (B.1.1.529/Omicron) within 20 mins, producing reliable diagnostic results consistent with those obtained from a standard fluorescence spectrometer. Importantly, PULD is capable of identifying the positive COVID-19 samples with superior sensitivity and specificity, making it a promising front-line tool for rapid, high-throughput screening and infection control of COVID-19 or other infectious diseases.

Rapid and ultrasensitive detection of SARS-CoV-2 spike protein based on upconversion luminescence biosensor for COVID-19 point-of-care diagnostics.

Here, we firstly introduce a detection system consisting of upconversion nanoparticles (UCNPs) and Au nanorods (AuNRs) for an ultrasensitive, rapid, quantitative and on-site detection of SARS-CoV-2 spike (S) protein based on Förster resonance energy transfer (FRET) effect. Briefly, the UCNPs capture the S protein of lysed SARS-CoV-2 in the swabs and subsequently they are bound with the anti-S antibodies modified AuNRs, resulting in significant nonradiative transitions from UCNPs (donors) to AuNRs (acceptors) at 480 nm and 800 nm, respectively. Notably, the specific recognition and quantitation of S protein can be realized in minutes at 800 nm because of the low autofluorescence and high Yb-Tm energy transfer in upconversion process. Inspiringly, the limit of detection (LOD) of the S protein can reach down to 1.06 fg mL-1, while the recognition of nucleocapsid protein is also comparable with a commercial test kit in a shorter time (only 5 min). The established strategy is technically superior to those reported point-of-care biosensors in terms of detection time, cost, and sensitivity, which paves a new avenue for future on-site rapid viral screening and point-of-care diagnostics.

Changes in Refractive Error Under COVID-19: A 3-Year Follow-up Study.

INTRODUCTION: To investigate changes in refractive error in schoolchildren before and during the coronavirus disease 2019 (COVID-19) pandemic. METHODS: This study included 2792 students, who underwent a 3-year follow-up from 2018 to 2020. All participants underwent yearly noncycloplegic refraction and ocular examinations. Time-related changes in sphere, cylinder, and spherical equivalent (SE) measurements in both genders were analyzed. RESULTS: The myopic sphere (- 0.78 ± 1.83 vs. - 1.03 ± 1.91 D; P = 0.025) and SE (- 1.04 ± 1.90 vs. - 1.32 ± 1.99 D; P = 0.015) progressed significantly from 2018 to 2019. Female participants had a significantly greater change in SE than male participants (P < 0.05), and the low hyperopia, emmetropia, and mild myopia groups significantly deteriorated (P < 0.001) from 2018 to 2019. Significant differences in sphere change (- 0.21 ± 0.97 vs. - 0.36 ± 0.96 D; P < 0.001) and SE change (- 0.23 ± 0.99 vs. - 0.38 ± 0.98 D; P < 0.001) were noted between 2019-2018 and 2020-2019, respectively. The respective changes in cylinder were statistically similar (- 0.03 ± 0.53 vs. - 0.05 ± 0.62 D; P = 0.400). CONCLUSIONS: The refractive status of schoolchildren showed an increasing myopic shift trend before and during the COVID-19 pandemic. The low hyperopia, emmetropia, and mild myopia groups were more sensitive to environmental changes during COVID-19 than before. The myopic shift was greater in female participants than male participants.

Evaluating the Impact of New York's Executive Order on Face Mask Use on COVID-19 Cases and Mortality: a Comparative Interrupted Times Series Study.

BACKGROUND: On April 17, 2020, the State of New York (NY) implemented an Executive Order that requires all people in NY to wear a face mask or covering in public settings where social distancing cannot be maintained. Although the Centers for Disease Control and Prevention recommended face mask use by the general public, there is a lack of evidence on the effect of face mask policies on the spread of COVID-19 at the state level. OBJECTIVE: To assess the impact of the Executive Order on face mask use on COVID-19 cases and mortality in NY. DESIGN: A comparative interrupted time series analysis was used to assess the impact of the Executive Order in NY with Massachusetts (MA) as a comparison state. PARTICIPANTS: We analyzed data on COVID-19 in NY and MA from March 25 to May 6, 2020. INTERVENTION: The Executive Order on face mask use in NY. MAIN MEASURES: Daily numbers of COVID-19 confirmed cases and deaths. KEY RESULTS: The average daily number of confirmed cases in NY decreased from 8549 to 5085 after the Executive Order took effect, with a trend change of 341 (95% CI, 187-496) cases per day. The average daily number of deaths decreased from 521 to 384 during the same two time periods, with a trend change of 52 (95% CI, 44-60) deaths per day. Compared to MA, the decreasing trend in NY was significantly greater for both daily numbers of confirmed cases (P = 0.003) and deaths (P < 0.001). CONCLUSIONS: The Executive Order on face mask use in NY led to a significant decrease in both daily numbers of COVID-19 confirmed cases and deaths. Findings from this study provide important evidence to support state-level policies that require face mask use by the general public.

A Retrospective Study on the Use of Chinese Patent Medicine in 24 Medical Institutions for COVID-19 in China.

Objective: This research aims to analyze the application regularity of Chinese patent medicine during the COVID-19 epidemic by collecting the names of the top three Chinese patent medicines used by 24 hospitals in 14 provinces of China in four time periods (January 20-22, February 16-18, March 01-03, April 01-03, 2020), and explore its contribution to combating the disease. Methods: 1) We built a database of the top three Chinese patent medicines used by 24 hospitals. 2) The frequency and efficacy distribution of Chinese patent medicine were analyzed with risk areas, regions, and hospitals of different properties as three factors. 3) Finally, we analyzed the differences in the use of heat-clearing and non-heat-clearing medicines among the three factors (χ2 test) and the correlation between the Chinese patent medicine and COVID-19 epidemic (correlation analysis) with SPSS 23.0 statistical software. Results: 1) The heat-clearing medicine was the main use category nationwide during January 20-22, 2020. Meanwhile, there was a significant difference in the utilization rate of heat-clearing and non-heat-clearing medicine in different risk areas (p < 0.01). 2) The variety of Chinese patent medicine was increased nationwide during February 16-18, 2020, mainly including tonics, blood-activating and resolving-stasis, and heat-clearing medicines. Meanwhile, there was a significant difference in the utilization rate of heat-clearing and non-heat-clearing medicine in the southern and northern regions (p < 0.05). 3) Tonics, and blood-activating and resolving-stasis medicines became the primary use categories nationwide during March 01-03, 2020. 4) The tonics class, and blood-activating and resolving-stasis medicine were still the primary categories nationwide during April 01-03, 2020. Meanwhile, there was a significant difference in the utilization rate of heat-clearing and non-heat-clearing medicine in different risk areas (p < 0.01). Conclusion: Chinese patent medicine has a certain degree of participation in fighting against the COVID-19. The efficacy distribution is related to the risk area, region, and hospital of different properties, among which the risk area is the main influencing factor. It is hoped that future research can further collect the application amount of Chinese patent medicine used in hospitals all over the country, so as to perfectly reflect the relationship between Chinese patent medicine and the epidemic situation.

A Rapid, Accurate and Machine-Agnostic Segmentation and Quantification Method for CT-Based COVID-19 Diagnosis.

COVID-19 has caused a global pandemic and become the most urgent threat to the entire world. Tremendous efforts and resources have been invested in developing diagnosis, prognosis and treatment strategies to combat the disease. Although nucleic acid detection has been mainly used as the gold standard to confirm this RNA virus-based disease, it has been shown that such a strategy has a high false negative rate, especially for patients in the early stage, and thus CT imaging has been applied as a major diagnostic modality in confirming positive COVID-19. Despite the various, urgent advances in developing artificial intelligence (AI)-based computer-aided systems for CT-based COVID-19 diagnosis, most of the existing methods can only perform classification, whereas the state-of-the-art segmentation method requires a high level of human intervention. In this paper, we propose a fully-automatic, rapid, accurate, and machine-agnostic method that can segment and quantify the infection regions on CT scans from different sources. Our method is founded upon two innovations: 1) the first CT scan simulator for COVID-19, by fitting the dynamic change of real patients' data measured at different time points, which greatly alleviates the data scarcity issue; and 2) a novel deep learning algorithm to solve the large-scene-small-object problem, which decomposes the 3D segmentation problem into three 2D ones, and thus reduces the model complexity by an order of magnitude and, at the same time, significantly improves the segmentation accuracy. Comprehensive experimental results over multi-country, multi-hospital, and multi-machine datasets demonstrate the superior performance of our method over the existing ones and suggest its important application value in combating the disease.