Explore the vast range of titles available.

For complex machine learning (ML) algorithms to gain widespread acceptance in decision making, we must be able to identify the features driving the predictions. Explainability models allow transparency of ML algorithms, however their reliability within high-dimensional data is unclear. To test the reliability of the explainability model SHapley Additive exPlanations (SHAP), we developed a convolutional neural network to predict tissue classification from Genotype-Tissue Expression (GTEx) RNA-seq data representing 16,651 samples from 47 tissues. Our classifier achieved an average F1 score of 96.1% on held-out GTEx samples. Using SHAP values, we identified the 2423 most discriminatory genes, of which 98.6% were also identified by differential expression analysis across all tissues. The SHAP genes reflected expected biological processes involved in tissue differentiation and function. Moreover, SHAP genes clustered tissue types with superior performance when compared to all genes, genes detected by differential expression analysis, or random genes. We demonstrate the utility and reliability of SHAP to explain a deep learning model and highlight the strengths of applying ML to transcriptome data.

Uncertainty estimation is crucial for understanding the reliability of deep learning (DL) predictions, and critical for deploying DL in the clinic. Differences between training and production datasets can lead to incorrect predictions with underestimated uncertainty. To investigate this pitfall, we benchmarked one pointwise and three approximate Bayesian DL models for predicting cancer of unknown primary, using three RNA-seq datasets with 10,968 samples across 57 cancer types. Our results highlight that simple and scalable Bayesian DL significantly improves the generalisation of uncertainty estimation. Moreover, we designed a prototypical metric—the area between development and production curve (ADP), which evaluates the accuracy loss when deploying models from development to production. Using ADP, we demonstrate that Bayesian DL improves accuracy under data distributional shifts when utilising ‘uncertainty thresholding’. In summary, Bayesian DL is a promising approach for generalising uncertainty, improving performance, transparency, and safety of DL models for deployment in the real world.

Trust and transparency are critical for deploying deep learning (DL) models into the clinic. DL application poses generalisation obstacles since training/development datasets often have different data distributions to clinical/production datasets that can lead to incorrect predictions with underestimated uncertainty. To investigate this pitfall, we benchmarked one pointwise and three approximate Bayesian DL models used to predict cancer of unknown primary with three independent RNA-seq datasets covering 10,968 samples across 57 primary cancer types. Our results highlight simple and scalable Bayesian DL significantly improves the generalisation of uncertainty estimation (e.g., p- value = 0.0013 for calibration). Moreover, we demonstrate Bayesian DL substantially improves accuracy under data distributional shifts when utilising \"uncertainty thresholding\" by designing a prototypical metric that evaluates the expected (accuracy) loss when deploying models from development to production, which we call the Area between Development and Production curve (ADP). In summary, Bayesian DL is a hopeful avenue of research for generalising uncertainty, which improves performance, transparency, and therefore safety of DL models for deployment in real-world. Competing Interest Statement M.Y., H.F., S.M., K.S. and M.T. are employed by Max Kelsen, which is a commercial company with an embedded research team. J.V.P. and N.W. are founders and shareholders of genomiQa Pty Ltd, and members of its Board. S.S., A.B., O.K., V.A., S.W, L.T.K. and R.L.J have no competing interests.

Differential microglial inflammatory responses play a role in regulation of differentiation and maturation of oligodendrocytes (OLs) in brain white matter. How microglia–OL crosstalk is altered by traumatic brain injury (TBI) and its impact on axonal myelination and neurological function impairment remain poorly understood. In this study, we investigated roles of a Na + /H + exchanger (NHE1), an essential microglial pH regulatory protein, in microglial proinflammatory activation and OL survival and differentiation in a murine TBI model induced by controlled cortical impact. Similar TBI-induced contusion volumes were detected in the Cx3cr1-Cre ERT2 control (Ctrl) mice and selective microglial Nhe1 knockout ( Cx3cr1-Cre ERT2 ;Nhe1 flox/flox , Nhe1 cKO) mice. Compared to the Ctrl mice, the Nhe1 cKO mice displayed increased resistance to initial TBI-induced white matter damage and accelerated chronic phase of OL regeneration at 30 days post-TBI. The cKO brains presented increased anti-inflammatory phenotypes of microglia and infiltrated myeloid cells, with reduced proinflammatory transcriptome profiles. Moreover, the cKO mice exhibited accelerated post-TBI sensorimotor and cognitive functional recovery than the Ctrl mice. These phenotypic outcomes in cKO mice were recapitulated in C57BL6J wild-type TBI mice receiving treatment of a potent NHE1 inhibitor HOE642 for 1–7 days post-TBI. Taken together, these findings collectively demonstrated that blocking NHE1 protein stimulates restorative microglial activation in oligodendrogenesis and neuroprotection, which contributes to accelerated brain repair and neurological function recovery after TBI.

Increased cellular degradation by autophagy is a feature of many interventions that delay ageing. We report here that increased autophagy is necessary for reduced insulin-like signalling (IIS) to extend lifespan in Drosophila and is sufficient on its own to increase lifespan. We first established that the well-characterised lifespan extension associated with deletion of the insulin receptor substrate chico was completely abrogated by downregulation of the essential autophagy gene Atg5 . We next directly induced autophagy by over-expressing the major autophagy kinase Atg1 and found that a mild increase in autophagy extended lifespan. Interestingly, strong Atg1 up-regulation was detrimental to lifespan. Transcriptomic and metabolomic approaches identified specific signatures mediated by varying levels of autophagy in flies. Transcriptional upregulation of mitochondrial-related genes was the signature most specifically associated with mild Atg1 upregulation and extended lifespan, whereas short-lived flies, possessing strong Atg1 overexpression, showed reduced mitochondrial metabolism and up-regulated immune system pathways. Increased proteasomal activity and reduced triacylglycerol levels were features shared by both moderate and high Atg1 overexpression conditions. These contrasting effects of autophagy on ageing and differential metabolic profiles highlight the importance of fine-tuning autophagy levels to achieve optimal healthspan and disease prevention.

Dysregulation of redox homeostasis is implicated in the ageing process and the pathology of age-related diseases. To study redox signalling by H 2 O 2 in vivo, we established a redox-shifted model by manipulating levels of the H 2 O 2 -degrading enzyme catalase in Drosophila . Here we report that ubiquitous over-expression of catalase robustly extends lifespan in females. As anticipated, these flies are strongly resistant to a range of oxidative stress challenges, but interestingly are sensitive to starvation, which could not be explained by differences in levels of energy reserves. This led us to explore the contribution of autophagy, which is an important mechanism for organismal survival in response to starvation. We show that autophagy is essential for the increased lifespan by catalase upregulation, as the survival benefits are completely abolished upon global autophagy knock-down. Furthermore, using a specific redox-inactive knock-in mutant, we highlight the in vivo role of a key regulatory cysteine residue in Atg4a, which is required for the lifespan extension in our catalase model. Altogether, these findings confirm the redox regulation of autophagy in vivo as an important modulator of longevity. Redox signalling is emerging as an important regulator of metabolism and physiology, which is dysregulated in ageing and disease. Here, the authors show that redox regulation of a key redox sensitive cysteine in Atg4a induces autophagy in vivo and extends lifespan in female Drosophila .

Background A new class of antibody-based drug therapy with the potential for disease modification is now available for Alzheimer’s disease (AD). However, the complexity of drug eligibility, administration, cost, and safety of such disease modifying therapies (DMTs) necessitates adopting new treatment and care pathways. A working group was convened in Ireland to consider the implications of, and health system readiness for, DMTs for AD, and to describe a service model for the detection, diagnosis, and management of early AD in the Irish context, providing a template for similar small-medium sized healthcare systems. Methods A series of facilitated workshops with a multidisciplinary working group, including Patient and Public Involvement (PPI) members, were undertaken. This informed a series of recommendations for the implementation of new DMTs using an evidence-based conceptual framework for health system readiness based on [1] material resources and structures and [2] human and institutional relationships, values, and norms. Results We describe a hub-and-spoke model, which utilises the existing dementia care ecosystem as outlined in Ireland’s Model of Care for Dementia, with Regional Specialist Memory Services (RSMS) acting as central hubs and Memory Assessment and Support Services (MASS) functioning as spokes for less central areas. We provide criteria for DMT referral, eligibility, administration, and ongoing monitoring. Conclusions Healthcare systems worldwide are acknowledging the need for advanced clinical pathways for AD, driven by better diagnostics and the emergence of DMTs. Despite facing significant challenges in integrating DMTs into existing care models, the potential for overcoming challenges exists through increased funding, resources, and the development of a structured national treatment network, as proposed in Ireland’s Model of Care for Dementia. This approach offers a replicable blueprint for other healthcare systems with similar scale and complexity.

Massive open online courses (MOOCs) have the potential to make a broader educational impact because many learners undertake these courses. Despite their reach, there is a lack of knowledge about which methods are used for evaluating these courses. The aim of this review was to identify current MOOC evaluation methods to inform future study designs. We systematically searched the following databases for studies published from January 2008 to October 2018: (1) Scopus, (2) Education Resources Information Center, (3) IEEE (Institute of Electrical and Electronic Engineers) Xplore, (4) PubMed, (5) Web of Science, (6) British Education Index, and (7) Google Scholar search engine. Two reviewers independently screened the abstracts and titles of the studies. Published studies in the English language that evaluated MOOCs were included. The study design of the evaluations, the underlying motivation for the evaluation studies, data collection, and data analysis methods were quantitatively and qualitatively analyzed. The quality of the included studies was appraised using the Cochrane Collaboration Risk of Bias Tool for randomized controlled trials (RCTs) and the National Institutes of Health-National Heart, Lung, and Blood Institute quality assessment tool for cohort observational studies and for before-after (pre-post) studies with no control group. The initial search resulted in 3275 studies, and 33 eligible studies were included in this review. In total, 16 studies used a quantitative study design, 11 used a qualitative design, and 6 used a mixed methods study design. In all, 16 studies evaluated learner characteristics and behavior, and 20 studies evaluated learning outcomes and experiences. A total of 12 studies used 1 data source, 11 used 2 data sources, 7 used 3 data sources, 4 used 2 data sources, and 1 used 5 data sources. Overall, 3 studies used more than 3 data sources in their evaluation. In terms of the data analysis methods, quantitative methods were most prominent with descriptive and inferential statistics, which were the top 2 preferred methods. In all, 26 studies with a cross-sectional design had a low-quality assessment, whereas RCTs and quasi-experimental studies received a high-quality assessment. The MOOC evaluation data collection and data analysis methods should be determined carefully on the basis of the aim of the evaluation. The MOOC evaluations are subject to bias, which could be reduced using pre-MOOC measures for comparison or by controlling for confounding variables. Future MOOC evaluations should consider using more diverse data sources and data analysis methods. RR2-10.2196/12087.

Long-term survival of nonreplicating Mycobacterium tuberculosis (Mtb) is ensured by the coordinated shutdown of active metabolism through a broad transcriptional program called the stringent response. In Mtb, this response is initiated by the enzymatic action of RelMtband deletion of relMtbproduces a strain (H37RvΔ relMtb) severely compromised in the maintenance of long-term viability. Although aerosol inoculation of mice with H37RvΔ relMtbresults in normal initial bacterial growth and containment, the ability of this strain to sustain chronic infection is severely impaired. Significant histopathologic differences were noted in lungs and spleens of mice infected with H37RvΔ relMtbcompared with controls throughout the course of the infection. Microarray analysis revealed that H37RvΔ relMtbsuffers from a generalized alteration of the transcriptional apparatus, as well as specific changes in the expression of virulence factors, cell-wall biosynthetic enzymes, heat shock proteins, and secreted antigens that may alter immune recognition of the recombinant organism. Thus, RelMtbis critical for the successful establishment of persistent infection in mice by altering the expression of antigenic and enzymatic factors that may contribute to successful latent infection.

The current understanding of the activity of mammalian pheromones is that endocrine and behavioural effects are limited to the exposed individuals. Here, we demonstrate that the nasal exposure of female mice to a male murine pheromone stimulates expansion of mammary glands, leading to prolonged nursing of pups. Subsequent behavioural testing of the pups from pheromone-exposed dams exhibited enhanced learning. Sialic acid components in the milk are known to be involved in brain development. We hypothesized that the offspring might have received more of this key nutrient that promotes brain development. The mRNA for polysialyltransferase, which produces polysialylated neural cell adhesion molecules related to brain development, was increased in the brain of offspring of pheromone-exposed dams at post-natal day 10, while it was not different at embryonic stages, indicating possible differential brain development during early post-natal life.