Validation of an automated machine learning algorithm for the detection and analysis of cerebral aneurysms.

OBJECTIVE: Machine learning algorithms have shown groundbreaking results in neuroimaging. The authors herein evaluated the performance of a newly developed convolutional neural network (CNN) to detect and analyze intracranial aneurysms (IAs) on CTA. METHODS: Consecutive patients with CTA studies between January 2015 and July 2021 at a single center were identified. The ground truth determination of cerebral aneurysm presence or absence was made from the neuroradiology report. The primary outcome was the performance of the CNN in detecting IAs in an external validation set, measured using area under the receiver operating characteristic curve statistics. Secondary outcomes included accuracy for location and size measurement. RESULTS: The independent validation imaging data set consisted of 400 patients with CTA studies, median age 40 years (IQR 34 years) and 141 (35.3%) of whom were male; 193 patients (48.3%) had a diagnosis of IA on neuroradiologist evaluation. The median maximum IA diameter was 3.7 mm (IQR 2.5 mm). In the independent validation imaging data set, the CNN performed well with 93.8% sensitivity (95% CI 0.87-0.98), 94.2% specificity (95% CI 0.90-0.97), and a positive predictive value of 88.2% (95% CI 0.80-0.94) in the subgroup with an IA diameter ≥ 4 mm. CONCLUSIONS: The described Viz.ai Aneurysm CNN performed well in identifying the presence or absence of IAs in an independent validation imaging set. Further studies are necessary to investigate the impact of the software on detection rates in a real-world setting.

Drug Repurposing Using Deep Embeddings of Gene Expression Profiles.

Computational drug repositioning requires assessment of the functional similarities among compounds. Here, we report a new method for measuring compound functional similarity based on gene expression data. This approach takes advantage of deep neural networks to learn an embedding that substantially denoises expression data, making replicates of the same compound more similar. Our method uses unlabeled data in the sense that it only requires compounds to be labeled by identity rather than detailed pharmacological information, which is often unavailable and costly to obtain. Similarity in the learned embedding space accurately predicted pharmacological similarities despite the lack of any such labels during training and achieved substantially improved performance in comparison with previous similarity measures applied to gene expression measurements. Our method could identify drugs with shared therapeutic and biological targets even when the compounds were structurally dissimilar, thereby revealing previously unreported functional relationships between compounds. Thus, our approach provides an improved engine for drug repurposing based on expression data, which we have made available through the online tool DeepCodex ( http://deepcodex.org ).

Effect of Caffeine on Attention and Alertness Measured in a Home-Setting, Using Web-Based Cognition Tests.

BACKGROUND: There is an increasing interest among nutritional researchers to perform lifestyle and nutritional intervention studies in a home setting instead of testing subjects in a clinical unit. The term used in other disciplines is 'ecological validity' stressing a realistic situation. This becomes more and more feasible because devices and self-tests that enable such studies are more commonly available. Here, we present such a study in which we reproduced the effect of caffeine on attention and alertness in an at-home setting. OBJECTIVE: The study was aimed to reproduce the effect of caffeine on attention and alertness using a Web-based study environment of subjects, at home, performing different Web-based cognition tests. METHODS: The study was designed as a randomized, placebo-controlled, double-blind, crossover study. Subjects were provided with coffee sachets (2 with and 2 without caffeine). They were also provided with a written instruction of the test days. Healthy volunteers consumed a cup of coffee after an overnight fast. Each intervention was repeated once. Before and 1 hour after coffee consumption subjects performed Web-based cognitive performance tests at home, which measured alertness and attention, established by 3 computerized tests provided by QuantifiedMind. Each test was performed for 5 minutes. RESULTS: Web-based recruitment was fast and efficient. Within 2 weeks, 102 subjects applied, of whom 70 were eligible. Of the 66 subjects who started the study, 53 completed all 4 test sessions (80%), indicating that they were able to perform the do it yourself tests, at home, correctly. The Go-No Go cognition test performed at home showed the same significant improvement in reaction time with caffeine as found in controlled studies in a metabolic ward (P=.02). For coding and N-back the second block was performed approximately 10% faster. No effect was seen on correctness. CONCLUSIONS: The study showed that the effects of caffeine consumption on a cognition test in an at-home setting revealed similar results as in a controlled setting. The Go-No Go test applied showed improved results after caffeine intake, similar as seen in clinical trials. This type of study is a fast, reliable, economical, and easy way to demonstrate effectiveness of a supplement and is rapidly becoming a viable alternative for the classical randomized control trial to evaluate life style and nutritional interventions. TRIAL REGISTRATION: Clinicaltrials.gov NCT02061982; https://clinicaltrials.gov/ct2/show/NCT02061982 (Archived by WebCite at https://clinicaltrials.gov/ct2/show/NCT02061982).

HLA-DPB1 as a Risk Factor for Relapse in Antineutrophil Cytoplasmic Antibody-Associated Vasculitis: A Cohort Study.

OBJECTIVE: The antineutrophil cytoplasmic antibody (ANCA)-associated vasculitides (AAVs) form a group of small-vessel vasculitides with systemic involvement. Although the etiology of AAVs remains largely unknown, both genetic and environmental factors have been implicated. Recently, certain alleles in the HLA-DPB1 region on chromosome 6 were shown to be associated with proteinase 3 (PR3)-ANCA-positive AAV but not with myeloperoxidase (MPO)-ANCA-positive AAV. The aim of this study was to investigate whether different alleles in the HLA-DPB1 region have clinical and/or prognostic implications in AAV. METHODS: One hundred seventy-four patients with a diagnosis of AAV were recruited at the Maastricht University Medical Centre between 2000 and 2009. Seventeen different HLA-DPB1 alleles were determined using the restriction fragment length polymorphism technique. A validation cohort of 170 AAV patients from the Vasculitis Centre of Luebeck/Bad Bramstedt was included. RESULTS: In the initial cohort, the distribution of HLA-DPB1 alleles was significantly different between PR3-ANCA-positive compared with MPO-ANCA-positive AAV patients, ANCA-negative AAV patients, and healthy controls. Importantly, HLA-DPB1*04:01 was present in 90% of PR3-ANCA-positive AAV patients compared with 63% of MPO-ANCA-positive AAV patients, 58% of ANCA-negative patients, and 63% of healthy controls. Patients homozygous for HLA-DPB1*04:01 had relapses more often compared with heterozygous patients and noncarrier patients. This association persisted after correction for ANCA subtype and diagnosis. In the validation cohort, patients homozygous for HLA-DPB1*04:01 and those heterozygous for HLA-DPB1*04:01 had relapses more often compared with noncarrier patients. When both patient cohorts were merged (n = 344), homozygous patients relapsed most often, followed by heterozygous patients and noncarrier patients. CONCLUSION: Carriage of HLA-DPB1*04:01 in patients with AAV is significantly associated with an increased risk of relapse compared with HLA-DPB1*04:01-negative patients, irrespective of ANCA status or clinical AAV entity.

Piecewise power laws in individual learning curves.

The notion that human learning follows a smooth power law (PL) of diminishing gains is well-established in psychology. This characteristic is observed when multiple curves are averaged, potentially masking more complex dynamics underpinning the curves of individual learners. Here, we analyzed 25,280 individual learning curves, each comprising 500 measurements of cognitive performance taken from four cognitive tasks. A piecewise PL (PPL) model explained the individual learning curves significantly better than a single PL, controlling for model complexity. The PPL model allows for multiple PLs connected at different points in the learning process. We also explored the transition dynamics between PL curve component pieces. Performance in later pieces typically surpassed that in earlier pieces, after a brief drop in performance at the transition point. The transition rate was negatively associated with age, even after controlling for overall performance. Our results suggest at least two processes at work in individual learning curves: locally, a gradual, smooth improvement, with diminishing gains within a specific strategy, which is modeled well as a PL; and globally, a discrete sequence of strategy shifts, in which each strategy is better in the long term than the ones preceding it. The piecewise extension of the classic PL of practice has implications for both individual skill acquisition and theories of learning.

Great Desire for Extended Life and Health amongst the American Public.

People want to live long, healthy lives. Previous surveys suggest very limited interest in much longer lifespans, but we show that stipulating good health changes responses to favor longer lives by an order of magnitude. Advances in aging research hold out hope for greatly slowed aging with associated good health. Understanding the public's desires correctly is important to avoid misallocation of resources for research.

Imputing gene expression from selectively reduced probe sets.

Measuring complete gene expression profiles for a large number of experiments is costly. We propose an approach in which a small subset of probes is selected based on a preliminary set of full expression profiles. In subsequent experiments, only the subset is measured, and the missing values are inputed. We developed several algorithms to simultaneously select probes and input missing values, and we demonstrate that these 'probe selection for imputation' (PSI) algorithms can successfully reconstruct missing gene expression values in a wide variety of applications, as evaluated using multiple metrics of biological importance. We analyze the performance of PSI methods under varying conditions, provide guidelines for choosing the optimal method based on the experimental setting, and indicate how to estimate imputation accuracy. Finally, we apply our approach to a large-scale study of immune system variation.

Unbiased reconstruction of a mammalian transcriptional network mediating pathogen responses.

Models of mammalian regulatory networks controlling gene expression have been inferred from genomic data but have largely not been validated. We present an unbiased strategy to systematically perturb candidate regulators and monitor cellular transcriptional responses. We applied this approach to derive regulatory networks that control the transcriptional response of mouse primary dendritic cells to pathogens. Our approach revealed the regulatory functions of 125 transcription factors, chromatin modifiers, and RNA binding proteins, which enabled the construction of a network model consisting of 24 core regulators and 76 fine-tuners that help to explain how pathogen-sensing pathways achieve specificity. This study establishes a broadly applicable, comprehensive, and unbiased approach to reveal the wiring and functions of a regulatory network controlling a major transcriptional response in primary mammalian cells.