•A lightweight deep learning model is capable of classifying phonocardiogram data on edge devices.•Self-supervised learning can significantly increase the classification accuracy and robustness of ...the model.•The classification algorithm can be deployed on consumer-grade smartphones with substantially improved runtime.
Detection and classification of heart murmur using mobile-phone-collected sound is an emerging approach to the scale-up screening of valvular heart disease at a population level. Nonetheless, the widespread adoption of artificial intelligence (AI) methods for this type of mobile health (mHealth) application requires highly accurate and lightweight AI models that can be deployed in consumer-grade mobile devices. This study presents a lightweight deep learning model and a self-supervised learning (SSL) method to utilise unlabelled data to improve the accuracy of valvular heart disease classification using phonocardiogram data.
This study proposes a lightweight convolutional neural network (CNN) that consists of ten times fewer parameters than other deep learning models to classify phonocardiogram data. SSL is applied to harness a large collection of unlabelled data as pre-training to enhance the accuracy and robustness of the model and reduce the number of epochs required to converge. A mobile application prototype that encapsulates the model is developed to perform in-device inference and fine-turning.
The proposed lightweight model achieves an average accuracy of 98.65% in 10-fold cross-validation. When coupled with SSL using unlabelled data, the pre-trained model can reach an average accuracy higher than 99.4% in 10-fold cross-validation. Furthermore, SSL-trained models have a 4–20% improvement in classification accuracy over non-SSL-trained models when tested with perturbed or noisy data, suggesting that SSL improves robustness of the model. When deployed on common smartphones, in-device fine-tuning and inference of the model can be completed within 0.03–0.37 s, which is considerably faster than 0.22–5.7 s by a standard CNN model that have ten times the number of parameters. Our lightweight model also consumes only a third of the power compared to the larger standard model.
This work presents a lightweight and accurate phonocardiogram classifier that supports near real-time performance on standard mobile devices.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Artificial intelligence (AI) has accelerated novel discoveries across multiple disciplines including medicine. Clinical medicine suffers from a lack of AI-based applications, potentially due to lack ...of awareness of AI methodology. Future collaboration between computer scientists and clinicians is critical to maximize the benefits of transformative technology in this field for patients. To illustrate, we describe AI-based advances in the diagnosis and management of gliomas, the most common primary central nervous system (CNS) malignancy.
Presented is a succinct description of foundational concepts of AI approaches and their relevance to clinical medicine, geared toward clinicians without computer science backgrounds. We also review novel AI approaches in the diagnosis and management of glioma.
Novel AI approaches in gliomas have been developed to predict the grading and genomics from imaging, automate the diagnosis from histopathology, and provide insight into prognosis.
Novel AI approaches offer acceptable performance in gliomas. Further investigation is necessary to improve the methodology and determine the full clinical utility of these novel approaches.
Abstract Point defects in two-dimensional materials are of key interest for quantum information science. However, the parameter space of possible defects is immense, making the identification of ...high-performance quantum defects very challenging. Here, we perform high-throughput (HT) first-principles computational screening to search for promising quantum defects within WS 2 , which present localized levels in the band gap that can lead to bright optical transitions in the visible or telecom regime. Our computed database spans more than 700 charged defects formed through substitution on the tungsten or sulfur site. We found that sulfur substitutions enable the most promising quantum defects. We computationally identify the neutral cobalt substitution to sulfur (Co $_{{{\rm{S}}}^{0}$$ S 0 ) and fabricate it with scanning tunneling microscopy (STM). The Co $_{{{\rm{S}}}^{0}$$ S 0 electronic structure measured by STM agrees with first principles and showcases an attractive quantum defect. Our work shows how HT computational screening and nanoscale synthesis routes can be combined to design promising quantum defects.
Abstract Metronomic chemotherapy using cyclophosphamide (CPA) is widely associated with antiangiogenesis; however, recent studies implicate other immune-based mechanisms, including antitumor innate ...immunity, which can induce major tumor regression in implanted brain tumor models. This study demonstrates the critical importance of drug schedule: CPA induced a potent antitumor innate immune response and tumor regression when administered intermittently on a 6-day repeating metronomic schedule but not with the same total exposure to activated CPA administered on an every 3-day schedule or using a daily oral regimen that serves as the basis for many clinical trials of metronomic chemotherapy. Notably, the more frequent metronomic CPA schedules abrogated the antitumor innate immune and therapeutic responses. Further, the innate immune response and antitumor activity both displayed an unusually steep dose-response curve and were not accompanied by antiangiogenesis. The strong recruitment of innate immune cells by the 6-day repeating CPA schedule was not sustained, and tumor regression was abolished, by a moderate (25%) reduction in CPA dose. Moreover, an ~20% increase in CPA dose eliminated the partial tumor regression and weak innate immune cell recruitment seen in a subset of the every 6-day treated tumors. Thus, metronomic drug treatment must be at a sufficiently high dose but also sufficiently well spaced in time to induce strong sustained antitumor immune cell recruitment. Many current clinical metronomic chemotherapeutic protocols employ oral daily low-dose schedules that do not meet these requirements, suggesting that they may benefit from optimization designed to maximize antitumor immune responses.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Studies examining changes in skeletal muscle and adipose tissue during treatment for cancer in children, adolescents, and young adults and their effect on the risk of chemotherapy toxicity ...(chemotoxicity) are limited.
Among 78 patients with lymphoma (79.5%) and rhabdomyosarcoma (20.5%), changes were measured in skeletal muscle (skeletal muscle index SMI; skeletal muscle density SMD) and adipose tissue (height-adjusted total adipose tissue hTAT) between baseline and first subsequent computed tomography scans at the third lumbar vertebral level by using commercially available software. Body mass index (BMI; operationalized as a percentile BMI%ile) and body surface area (BSA) were examined at each time point. The association of changes in body composition with chemotoxicities was examined by using linear regression.
The median age at cancer diagnosis of this cohort (62.8% male; 55.1% non-Hispanic White) was 12.7 years (2.5-21.1 years). The median time between scans was 48 days (range, 8-207 days). By adjusting for demographics and disease characteristics, this study found that patients undergo a significant decline in SMD (β ± standard error SE = -4.1 ± 1.4; p < .01). No significant changes in SMI (β ± SE = -0.5 ± 1.0; p = .7), hTAT (β ± SE = 5.5 ± 3.9; p = .2), BMI% (β ± SE = 4.1 ± 4.8; p = .3), or BSA (β ± SE = -0.02 ± 0.01; p = .3) were observed. Decline in SMD (per Hounsfield unit) was associated with a greater proportion of chemotherapy cycles with grade ≥3 nonhematologic toxicity (β ± SE = 1.09 ± 0.51; p = .04).
This study shows that children, adolescents, and young adults with lymphoma and rhabdomyosarcoma undergo a decline in SMD early during treatment, which is associated with a risk of chemotoxicities. Future studies should focus on interventions designed at preventing the loss of muscle during treatment.
We show that among children, adolescents, and young adults with lymphoma and rhabdomyosarcoma receiving chemotherapy, skeletal muscle density declines early during treatment. Additionally, a decline in skeletal muscle density is associated with a greater risk of nonhematologic chemotoxicities.
Full text
Available for:
BFBNIB, FZAB, GIS, IJS, KILJ, NLZOH, NUK, OILJ, SAZU, SBCE, SBMB, UL, UM, UPUK
The electrical current blockade of a peptide or protein threading through a nanopore can be used as a fingerprint of the molecule in biosensor applications. However, threading of full-length proteins ...has only been achieved using enzymatic unfolding and translocation. Here we describe an enzyme-free approach for unidirectional, slow transport of full-length proteins through nanopores. We show that the combination of a chemically resistant biological nanopore, α-hemolysin (narrowest part is ~1.4 nm in diameter), and a high concentration guanidinium chloride buffer enables unidirectional, single-file protein transport propelled by an electroosmotic effect. We show that the mean protein translocation velocity depends linearly on the applied voltage and translocation times depend linearly on length, resembling the translocation dynamics of ssDNA. Using a supervised machine-learning classifier, we demonstrate that single-translocation events contain sufficient information to distinguish their threading orientation and identity with accuracies larger than 90%. Capture rates of protein are increased substantially when either a genetically encoded charged peptide tail or a DNA tag is added to a protein.
Full text
Available for:
GEOZS, IJS, IMTLJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBMB, UL, UM, UPUK, ZAGLJ
Microbes of the human respiratory tract are important in health and disease, but accurate sampling of the lung presents challenges. Lung microbes are commonly sampled by bronchoscopy, but to acquire ...samples the bronchoscope must pass through the upper respiratory tract, which is rich in microbes. Here we present methods to identify authentic lung microbiota in bronchoalveolar lavage (BAL) fluid that contains substantial oropharyngeal admixture. We studied clinical BAL samples from six selected subjects with potential heavy lung colonization. A single sample of BAL fluid was obtained from each subject along with contemporaneous oral wash (OW) to sample the oropharynx, and then DNA was extracted from three separate aliquots of each. Bacterial 16S rDNA sequences were amplified and products analyzed by 454 pyrosequencing. By comparing replicates, we were able to specify the depth of sequencing needed to reach a 95% chance of identifying a bacterial lineage of a given proportion--for example, at a depth of 5,000 tags, OTUs of proportion 0.3% or greater would be called with 95% confidence. We next constructed a single-sided outlier test that allowed lung-enriched organisms to be quantified against a background of oropharyngeal admixture, and assessed improvements available with replicate sequence analysis. This allowed identification of lineages enriched in lung in some BAL specimens. Finally, using samples from healthy volunteers collected at multiple sites in the upper respiratory tract, we show that OW provides a reasonable but not perfect surrogate for bacteria carried into to the lung by a bronchoscope. These methods allow identification of microbes that can replicate in the lung despite the background due to oropharyngeal microbes derived from aspiration and bronchoscopic carry-over.
Full text
Available for:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
Interferons (IFNs) are cytokines that play a critical role in limiting infectious and malignant diseases
. Emerging data suggest that the strength and duration of IFN signaling can differentially ...impact cancer therapies, including immune checkpoint blockade
. Here, we characterize the output of IFN signaling, specifically IFN-stimulated gene (ISG) signatures, in primary tumors from The Cancer Genome Atlas. While immune infiltration correlates with the ISG signature in some primary tumors, the existence of ISG signature-positive tumors without evident infiltration of IFN-producing immune cells suggests that cancer cells per se can be a source of IFN production. Consistent with this hypothesis, analysis of patient-derived tumor xenografts propagated in immune-deficient mice shows evidence of ISG-positive tumors that correlates with expression of human type I and III IFNs derived from the cancer cells. Mechanistic studies using cell line models from the Cancer Cell Line Encyclopedia that harbor ISG signatures demonstrate that this is a by-product of a STING-dependent pathway resulting in chronic tumor-derived IFN production. This imposes a transcriptional state on the tumor, poising it to respond to the aberrant accumulation of double-stranded RNA (dsRNA) due to increased sensor levels (MDA5, RIG-I and PKR). By interrogating our functional short-hairpin RNA screen dataset across 398 cancer cell lines, we show that this ISG transcriptional state creates a novel genetic vulnerability. ISG signature-positive cancer cells are sensitive to the loss of ADAR, a dsRNA-editing enzyme that is also an ISG. A genome-wide CRISPR genetic suppressor screen reveals that the entire type I IFN pathway and the dsRNA-activated kinase, PKR, are required for the lethality induced by ADAR depletion. Therefore, tumor-derived IFN resulting in chronic signaling creates a cellular state primed to respond to dsRNA accumulation, rendering ISG-positive tumors susceptible to ADAR loss.
Full text
Available for:
EMUNI, FIS, FZAB, GEOZS, GIS, IJS, IMTLJ, KILJ, KISLJ, MFDPS, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, SBMB, SBNM, UKNU, UL, UM, UPUK, VKSCE, ZAGLJ
Healthy aging can lead to impairments in learning that affect many laboratory and real-life tasks. These tasks often involve the acquisition of dynamic contingencies, which requires adjusting the ...rate of learning to environmental statistics. For example, learning rate should increase when expectations are uncertain (uncertainty), outcomes are surprising (surprise) or contingencies are more likely to change (hazard rate). In this study, we combine computational modelling with an age-comparative behavioural study to test whether age-related learning deficits emerge from a failure to optimize learning according to the three factors mentioned above. Our results suggest that learning deficits observed in healthy older adults are driven by a diminished capacity to represent and use uncertainty to guide learning. These findings provide insight into age-related cognitive changes and demonstrate how learning deficits can emerge from a failure to accurately assess how much should be learned.