Elderly humans show decreased humoral immunity to pathogens and vaccines, yet the effects of aging on B cells are not fully known. Chronic viral infection by cytomegalovirus (CMV) is implicated as a ...driver of clonal T cell proliferations in some aging humans, but whether CMV or Epstein-Barr virus (EBV) infection contributes to alterations in the B cell repertoire with age is unclear. We have used high-throughput DNA sequencing of immunoglobulin heavy chain (IGH) gene rearrangements to study the B cell receptor repertoires over two successive years in 27 individuals ranging in age from 20 to 89 years. Some features of the B cell repertoire remain stable with age, but elderly subjects show increased numbers of B cells with long CDR3 regions, a trend toward accumulation of more highly mutated IgM and IgG immunoglobulin genes, and persistent clonal B cell populations in the blood. Seropositivity for CMV or EBV infection alters B cell repertoires, regardless of the individual's age: EBV infection correlates with the presence of persistent clonal B cell expansions, while CMV infection correlates with the proportion of highly mutated antibody genes. These findings isolate effects of aging from those of chronic viral infection on B cell repertoires, and provide a baseline for understanding human B cell responses to vaccination or infectious stimuli.
The high-throughput sequencing revolution allows us to take millions of noisy short reads from the DNA in a sample, essentially taking a snapshot of the genomic material in the sample. To recover the ...true genomes, these reads are assembled by algorithms exploiting their high coverage and overlap. I focus on two scenarios for sequence assembly. The first is de novo assembly, where the reads come from an unknown and diverse population of genomes. The second is variant assembly, where the reads come from short but clonally related genomes, only slightly mutated from each other. In both cases I use the same principled Bayesian approach to design an algorithm that uncovers the composition of the genomic sequences that produced the reads. I will demonstrate the algorithms' performance on real data taken from various metagenomic environments, as well as the immune system B cells. On that latter dataset, collected from 10 organ donors each providing 4 tissue samples, the results show evidence of clone migration between tissues and provide new insights on the organization of the immune system.
In order to analyze a trained model performance and identify its weak spots, one has to set aside a portion of the data for testing. The test set has to be large enough to detect statistically ...significant biases with respect to all the relevant sub-groups in the target population. This requirement may be difficult to satisfy, especially in data-hungry applications. We propose to overcome this difficulty by generating synthetic test set. We use the face landmarks detection task to validate our proposal by showing that all the biases observed on real datasets are also seen on a carefully designed synthetic dataset. This shows that synthetic test sets can efficiently detect a model's weak spots and overcome limitations of real test set in terms of quantity and/or diversity.
From Neural Networks to Deep Learning Laserson, Jonathan
Crossroads (Association for Computing Machinery),
09/2011, Letnik:
18, Številka:
1
Journal Article
Pondering the brain with the help of machine learning expert Andrew Ng and researcher-turned-author-turned-entrepreneur Jeff Hawkins.
Toward personalized care management of patients at risk Neuvirth, Hani; Ozery-Flato, Michal; Hu, Jianying ...
Proceedings of the 17th ACM SIGKDD international conference on Knowledge discovery and data mining,
08/2011
Conference Proceeding
Chronic diseases constitute the leading cause of mortality in the western world, have a major impact on the patients' quality of life, and comprise the bulk of healthcare costs. Nowadays, healthcare ...data management systems integrate large amounts of medical information on patients, including diagnoses, medical procedures, lab test results, and more. Sophisticated analysis methods are needed for utilizing these data to assist in patient management and to enhance treatment quality at reduced costs. In this study, we take a first step towards better disease management of diabetic patients by applying state-of-the art methods to anticipate the patient's future health condition and to identify patients at high risk. Two relevant outcome measures are explored: the need for emergency care services and the probability of the treatment producing a sub-optimal result, as defined by domain experts. By identifying the high-risk patients our prediction system can be used by healthcare providers to prepare both financially and logistically for the patient needs. To demonstrate a potential downstream application for the identified high-risk patients, we explore the association between the physician treating these patients and the treatment outcome, and propose a system that can assist healthcare providers in optimizing the match between a patient and a physician.
Our work formulates the problem and examines the performance of several learning models on data from several thousands of patients. We further describe a pilot system built on the results of this analysis. We show that the risk for the two considered outcomes can be evaluated from patients' characteristics and that features of the patient-physician match improve the prediction accuracy for the treatment's success. These results suggest that personalized medicine can be valuable for high risk patients and raise interesting questions for future improvements.
The chest X-ray (CXR) is by far the most commonly performed radiological examination for screening and diagnosis of many cardiac and pulmonary diseases. There is an immense world-wide shortage of ...physicians capable of providing rapid and accurate interpretation of this study. A radiologist-driven analysis of over two million CXR reports generated an ontology including the 40 most prevalent pathologies on CXR. By manually tagging a relatively small set of sentences, we were able to construct a training set of 959k studies. A deep learning model was trained to predict the findings given the patient frontal and lateral scans. For 12 of the findings we compare the model performance against a team of radiologists and show that in most cases the radiologists agree on average more with the algorithm than with each other.
BACKGROUND:In sub-Saharan Africa, most people with HIV do not know they are infected.
METHODS:We conducted door-to-door home-based testing and counseling (HBTC) in rural western Kenya (Lwak) and an ...informal urban settlement in Nairobi (Kibera) in 2008. After consent, eligible persons (adults and adolescents aged 13 years or older and children aged 12 years or younger, whose biologic mother was HIV-infected or deceased) received parallel fingerstick HIV rapid testing and counseling. Persons newly diagnosed with HIV were referred to care services, fingerstick blood for CD4 testing was collected, and a one-month follow-up home visit was conducted.
RESULTS:Among 24,450 people who were offered HBTC, 19,966 (81.7%) accepted; 65.4% of whom were HIV-tested for the first time. Prevalence in adults aged 18 years or older being HIV-tested for the first time was 13.5% (12.6%, Kibera; 14.2%, Lwak). Among adults who reported a previously negative test, HIV prevalence was 7.4% (7.2%, Kibera; 7.6%, Lwak). Among all persons with HIV in these communities, two-thirds were newly diagnosed through HBTC. Median CD4 count among newly diagnosed adults was 403 interquartile range (IQR) = 252–594. Among couples, 38.0% in Kibera and 51.7% in Lwak were counseled together. Among HIV-infected people in a couple, 34.6% had an HIV-uninfected partner. Among newly diagnosed HIV-infected persons, at the one-month follow-up visit, 53.6% in Kibera and 43.6% in Lwak reported having visited an HIV patient support center.
CONCLUSIONS:HBTC acceptance was high and most HIV-infected persons did not previously know they had HIV. HBTC can be an effective strategy for early HIV diagnosis and treatment referral.
Malaria transmission is highly heterogeneous in most settings, resulting in the formation of recognizable malaria hotspots. Targeting these hotspots might represent a highly efficacious way of ...controlling or eliminating malaria if the hotspots fuel malaria transmission to the wider community.
Hotspots of malaria will be determined based on spatial patterns in age-adjusted prevalence and density of antibodies against malaria antigens apical membrane antigen-1 and merozoite surface protein-1. The community effect of interventions targeted at these hotspots will be determined. The intervention will comprise larviciding, focal screening and treatment of the human population, distribution of long-lasting insecticide-treated nets and indoor residual spraying. The impact of the intervention will be determined inside and up to 500 m outside the targeted hotspots by PCR-based parasite prevalence in cross-sectional surveys, malaria morbidity by passive case detection in selected facilities and entomological monitoring of larval and adult Anopheles populations.
This study aims to provide direct evidence for a community effect of hotspot-targeted interventions. The trial is powered to detect large effects on malaria transmission in the context of ongoing malaria interventions. Follow-up studies will be needed to determine the effect of individual components of the interventions and the cost-effectiveness of a hotspot-targeted approach, where savings made by reducing the number of compounds that need to receive interventions should outweigh the costs of hotspot-detection.
NCT01575613. The protocol was registered online on 20 March 2012; the first community was randomized on 26 March 2012.
HIV stigma is considered to be a major driver of the HIV/AIDS pandemic, yet there is a limited understanding of its occurrence. We describe the geographic patterns of two forms of HIV stigma in a ...cross-sectional sample of women of childbearing age from western Kenya: internalized stigma (associated with shame) and externalized stigma (associated with blame).
Geographic studies of HIV stigma provide a first step in generating hypotheses regarding potential community-level causes of stigma and may lead to more effective community-level interventions.
Spatial regression using generalized additive models and point pattern analyses using K-functions were used to assess the spatial scale(s) at which each form of HIV stigma clusters, and to assess whether the spatial clustering of each stigma indicator was present after adjustment for individual-level characteristics.
There was evidence that externalized stigma (blame) was geographically heterogeneous across the study area, even after controlling for individual-level factors (P=0.01). In contrast, there was less evidence (P=0.70) of spatial trend or clustering of internalized stigma (shame).
Our results may point to differences in the underlying social processes motivating each form of HIV stigma. Externalized stigma may be driven more by cultural beliefs disseminated within communities, whereas internalized stigma may be the result of individual-level characteristics outside the domain of community influence. These data may inform community-level interventions to decrease HIV-related stigma, and thus impact the HIV epidemic.