The Centers for Disease Control and Prevention (CDC) coordinates a labor-intensive process to measure the prevalence of autism spectrum disorder (ASD) among children in the United States. Random ...forests methods have shown promise in speeding up this process, but they lag behind human classification accuracy by about 5%. We explore whether more recently available document classification algorithms can close this gap.
Using data gathered from a single surveillance site, we applied 8 supervised learning algorithms to predict whether children meet the case definition for ASD based solely on the words in their evaluations. We compared the algorithms' performance across 10 random train-test splits of the data, using classification accuracy, F1 score, and number of positive calls to evaluate their potential use for surveillance.
Across the 10 train-test cycles, the random forest and support vector machine with Naive Bayes features (NB-SVM) each achieved slightly more than 87% mean accuracy. The NB-SVM produced significantly more false negatives than false positives (P = 0.027), but the random forest did not, making its prevalence estimates very close to the true prevalence in the data. The best-performing neural network performed similarly to the random forest on both measures.
The random forest performed as well as more recently available models like the NB-SVM and the neural network, and it also produced good prevalence estimates. NB-SVM may not be a good candidate for use in a fully-automated surveillance workflow due to increased false negatives. More sophisticated algorithms, like hierarchical convolutional neural networks, may not be feasible to train due to characteristics of the data. Current algorithms might perform better if the data are abstracted and processed differently and if they take into account information about the children in addition to their evaluations.
Deep learning models performed similarly to traditional machine learning methods at predicting the clinician-assigned case status for CDC's autism surveillance system. While deep learning methods had limited benefit in this task, they may have applications in other surveillance systems.
Celotno besedilo
Dostopno za:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
Summary Background We did a meta-analysis to assess factors associated with disparities in HIV infection in black men who have sex with men (MSM) in Canada, the UK, and the USA. Methods We searched ...Embase, Medline, Google Scholar, and online conference proceedings from Jan 1, 1981, to Dec 31, 2011, for racial comparative studies with quantitative outcomes associated with HIV risk or HIV infection. Key words and Medical Subject Headings (US National Library of Medicine) relevant to race were cross-referenced with citations pertinent to homosexuality in Canada, the UK, and the USA. Data were aggregated across studies for every outcome of interest to estimate overall effect sizes, which were converted into summary ORs for 106 148 black MSM relative to 581 577 other MSM. Findings We analysed seven studies from Canada, 13 from the UK, and 174 from the USA. In every country, black MSM were as likely to engage similarly in serodiscordant unprotected sex as other MSM. Black MSM in Canada and the USA were less likely than other MSM to have a history of substance use (odds ratio, OR, 0·53, 95% CI 0·38–0·75, for Canada and 0·67, 0·50–0·92, for the USA). Black MSM in the UK (1·86, 1·58–2·18) and the USA (3·00, 2·06–4·40) were more likely to be HIV positive than were other MSM, but HIV-positive black MSM in each country were less likely (22% in the UK and 60% in the USA) to initiate combination antiretroviral therapy (cART) than other HIV-positive MSM. US HIV-positive black MSM were also less likely to have health insurance, have a high CD4 count, adhere to cART, or be virally suppressed than were other US HIV-positive MSM. Notably, despite a two-fold greater odds of having any structural barrier that increases HIV risk (eg, unemployment, low income, previous incarceration, or less education) compared with other US MSM, US black MSM were more likely to report any preventive behaviour against HIV infection (1·39, 1·23–1·57). For outcomes associated with HIV infection, disparities were greatest for US black MSM versus other MSM for structural barriers, sex partner demographics (eg, age, race), and HIV care outcomes, whereas disparities were least for sexual risk outcomes. Interpretation Similar racial disparities in HIV and sexually transmitted infections and cART initiation are seen in MSM in the UK and the USA. Elimination of disparities in HIV infection in black MSM cannot be accomplished without addressing structural barriers or differences in HIV clinical care access and outcomes. Funding None.
The World Health Organization recommends the screening of all people living with HIV for tuberculosis (TB) disease, followed by TB treatment, or isoniazid preventive therapy (IPT) when TB is ...excluded. However, the difficulty of reliably excluding TB disease has severely limited TB screening and IPT uptake in resource-limited settings. We conducted an individual participant data meta-analysis of primary studies, aiming to identify a sensitive TB screening rule.
We identified 12 studies that had systematically collected sputum specimens regardless of signs or symptoms, at least one mycobacterial culture, clinical symptoms, and HIV and TB disease status. Bivariate random-effects meta-analysis and the hierarchical summary relative operating characteristic curves were used to evaluate the screening performance of all combinations of variables of interest. TB disease was diagnosed in 557 (5.8%) of 9,626 people living with HIV. The primary analysis included 8,148 people living with HIV who could be evaluated on five symptoms from nine of the 12 studies. The median age was 34 years. The best performing rule was the presence of any one of: current cough (any duration), fever, night sweats, or weight loss. The overall sensitivity of this rule was 78.9% (95% confidence interval CI 58.3%-90.9%) and specificity was 49.6% (95% CI 29.2%-70.1%). Its sensitivity increased to 90.1% (95% CI 76.3%-96.2%) among participants selected from clinical settings and to 88.0% (95% CI 76.1%-94.4%) among those who were not previously screened for TB. Negative predictive value was 97.7% (95% CI 97.4%-98.0%) and 90.0% (95% CI 88.6%-91.3%) at 5% and 20% prevalence of TB among people living with HIV, respectively. Abnormal chest radiographic findings increased the sensitivity of the rule by 11.7% (90.6% versus 78.9%) with a reduction of specificity by 10.7% (49.6% versus 38.9%).
Absence of all of current cough, fever, night sweats, and weight loss can identify a subset of people living with HIV who have a very low probability of having TB disease. A simplified screening rule using any one of these symptoms can be used in resource-constrained settings to identify people living with HIV in need of further diagnostic assessment for TB. Use of this algorithm should result in earlier TB diagnosis and treatment, and should allow for substantial scale-up of IPT.
Celotno besedilo
Dostopno za:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
Public health surveillance by necessity occurs without explicit patient consent. There is strong legal and scientific support for maintaining name-based reporting of infectious diseases and other ...types of public health surveillance. We present conditions under which surveillance without explicit patient consent is ethically justifiable using principles of contemporary clinical and public health ethics. Overriding individual autonomy must be justified in terms of the obligation of public health to improve population health, reduce inequities, attend to the health of vulnerable and systematically disadvantaged persons, and prevent harm. In addition, data elements collected without consent must represent the minimal necessary interference, lead to effective public health action, and be maintained securely.
Abstract
Point-of-care antigen tests are an important tool for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) detection, yet are less clinically sensitive than real-time ...reverse-transcription polymerase chain reaction (RT-PCR), affecting their efficacy as screening procedures. Our goal in this analysis was to see whether we could improve this sensitivity by considering antigen test results in combination with other relevant information, namely exposure status and reported symptoms. In November 2020, we collected 3,419 paired upper respiratory specimens tested by RT-PCR and the Abbott BinaxNOW (Abbott Laboratories, Abbott Park, Illinois) antigen test at 2 community testing sites in Pima County, Arizona. We used symptom, exposure, and antigen-testing data to evaluate the sensitivity and specificity of various symptom definitions in predicting RT-PCR positivity. Our analysis yielded 6 novel multisymptom case definitions with and without antigen test results, the best of which overall achieved a Youden’s J index of 0.66, as compared with 0.53 for antigen testing alone. Using a random forest as a guide, we show that this definition, along with our others, does not lose the ability to generalize well to new data despite achieving optimal performance in our sample. Our methodology is broadly applicable, and our code is publicly available to aid public health practitioners in developing or fine-tuning their own case definitions.
Pooled estimates from across the African diaspora show that black men who have sex with men (MSM) are 15 times more likely to be HIV positive compared with general populations and 8·5 times more ...likely compared with black populations. Disparities in the prevalence of HIV infection are greater in African and Caribbean countries that criminalise homosexual activity than in those that do not criminalise such behaviour. With the exception of US and African epidemiological studies, most studies of black MSM mainly focus on outcomes associated with HIV behavioural risk rather than on prevalence, incidence, or undiagnosed infection. Nevertheless, black MSM across the African diaspora share common experiences such as discrimination, cultural norms valuing masculinity, concerns about confidentiality during HIV testing or treatment, low access to HIV drugs, threats of violence or incarceration, and few targeted HIV prevention resources.
The early identification of clusters of persons with tuberculosis (TB) that will grow to become outbreaks creates an opportunity for intervention in preventing future TB cases. We used surveillance ...data (2009-2018) from the United States, statistically derived definitions of unexpected growth, and machine-learning techniques to predict which clusters of genotype-matched TB cases are most likely to continue accumulating cases above expected growth within a 1-year follow-up period. We developed a model to predict which clusters are likely to grow on a training and testing data set that was generalizable to a validation data set. Our model showed that characteristics of clusters were more important than the social, demographic, and clinical characteristics of the patients in those clusters. For instance, the time between cases before unexpected growth was identified as the most important of our predictors. A faster accumulation of cases increased the probability of excess growth being predicted during the follow-up period. We have demonstrated that combining the characteristics of clusters and cases with machine learning can add to existing tools to help prioritize which clusters may benefit most from public health interventions. For example, consideration of an entire cluster, not only an individual patient, may assist in interrupting ongoing transmission.
Background. Although seasonal variation in tuberculosis incidence has been described in several recent studies, the mechanism underlying this seasonality remains unknown. Seasonality of tuberculosis ...disease may indicate the presence of season-specific risk factors that could potentially be controlled if they were better understood. We conducted this study to determine whether tuberculosis is seasonal in the United States and to describe patterns of seasonality in specific populations. Methods. We performed a time series decomposition analysis of tuberculosis cases reported to the Centers for Disease Control and Prevention from 1993 through 2008. Seasonal amplitude of tuberculosis disease (the difference between the months with the highest and lowest mean case counts), was calculated for the population as a whole and for populations with select demographic, clinical, and epidemiologic characteristics. Results. A total of 243 432 laboratory-confirmed tuberculosis cases were reported over a period of 16 years. A mean of 21.4% more cases were diagnosed in March, the peak month, compared with November, the trough month. The magnitude of seasonality did not vary with latitude. The greatest seasonal amplitude was found among children aged <5 years and in cases associated with disease clusters. Conclusions. Tuberculosis is a seasonal disease in the United States, with a peak in spring and trough in late fall. The latitude independence of seasonality suggests that reduced winter sunlight exposure may not be a strong contributor to tuberculosis risk. Increased seasonality among young children and clustered cases suggests that disease that is the result of recent transmission is more influenced by season than disease resulting from activation of latent infection.
Display omitted
•Neural networks classify chief complaints better than two other methods.•Classification works best for conditions that are easier to identify during triage.•Nonspecific chief ...complaints are difficult for all classifiers assessed.
Syndromic surveillance detects and monitors individual and population health indicators through sources such as emergency department records. Automated classification of these records can improve outbreak detection speed and diagnosis accuracy. Current syndromic systems rely on hand-coded keyword-based methods to parse written fields and may benefit from the use of modern supervised-learning classifier models. In this paper, we implement two recurrent neural network models based on long short-term memory (LSTM) and gated recurrent unit (GRU) cells and compare them to two traditional bag-of-words classifiers: multinomial naïve Bayes (MNB) and a support vector machine (SVM). The MNB classifier is one of only two machine learning algorithms currently being used for syndromic surveillance. All four models are trained to predict diagnostic code groups as defined by Clinical Classification Software, first to predict from discharge diagnosis, and then from chief complaint fields. The classifiers are trained on 3.6 million de-identified emergency department records from a single United States jurisdiction. We compare performance of these models primarily using the F1 score, and we measure absolute model performance to determine which conditions are the most amenable to surveillance based on chief complaint alone. Using discharge diagnoses, the LSTM classifier performs best, though all models exhibit an F1 score above 96.00. Using chief complaints, the GRU performs best (F1 = 47.38), and MNB with bigrams performs worst (F1 = 39.40). We also note that certain syndrome types are easier to detect than others. For example, chief complaints using the GRU model predicts alcohol-related disorders well (F1 = 78.91) but predicts influenza poorly (F1 = 14.80). In all instances, the RNN models outperformed the bag-of-words classifiers suggesting deep learning models could substantially improve the automatic classification of unstructured text for syndromic surveillance.
IMPORTANCE: Criterion-standard specimens for tuberculosis diagnosis in young children, gastric aspirate (GA) and induced sputum, are invasive and rarely collected in resource-limited settings. A far ...less invasive approach to tuberculosis diagnostic testing in children younger than 5 years as sensitive as current reference standards is important to identify. OBJECTIVE: To characterize the sensitivity of preferably minimally invasive specimen and assay combinations relative to maximum observed yield from all specimens and assays combined. DESIGN, SETTING, AND PARTICIPANTS: In this prospective cross-sectional diagnostic study, the reference standard was a panel of up to 2 samples of each of 6 specimen types tested for Mycobacterium tuberculosis complex by Xpert MTB/RIF assay and mycobacteria growth indicator tube culture. Multiple different combinations of specimens and tests were evaluated as index tests. A consecutive series of children was recruited from inpatient and outpatient settings in Kisumu County, Kenya, between October 2013 and August 2015. Participants were children younger than 5 years who had symptoms of tuberculosis (unexplained cough, fever, malnutrition) and parenchymal abnormality on chest radiography or who had cervical lymphadenopathy. Children with 1 or more evaluable specimen for 4 or more primary study specimen types were included in the analysis. Data were analyzed from February 2015 to October 2020. MAIN OUTCOMES AND MEASURES: Cumulative and incremental diagnostic yield of combinations of specimen types and tests relative to the maximum observed yield. RESULTS: Of the 300 enrolled children, the median (interquartile range) age was 2.0 (1.0-3.6) years, and 151 (50.3%) were female. A total of 294 met criteria for analysis. Of 31 participants with confirmed tuberculosis (maximum observed yield), 24 (sensitivity, 77%; interdecile range, 68%-87%) had positive results on up to 2 GA samples and 20 (sensitivity, 64%; interdecile range, 53%-76%) had positive test results on up to 2 induced sputum samples. The yields of 2 nasopharyngeal aspirate (NPA) samples (23 of 31 sensitivity, 74%; interdecile range, 64%-84%), of 1 NPA sample and 1 stool sample (22 of 31 sensitivity, 71%; interdecile range, 60%-81%), or of 1 NPA sample and 1 urine sample (21.5 of 31 sensitivity, 69%; interdecile range, 58%-80%) were similar to reference-standard specimens. Combining up to 2 each of GA and NPA samples had an average yield of 90% (28 of 31). CONCLUSIONS AND RELEVANCE: NPA, in duplicate or in combination with stool or urine specimens, was readily obtainable and had diagnostic yield comparable with reference-standard specimens. This combination could improve tuberculosis diagnosis among children in resource-limited settings. Combining GA and NPA had greater yield than that of the current reference standards and may be useful in certain clinical and research settings.