Empowering Patient Similarity Networks through Innovative Data-Quality-Aware Federated Profiling

Continuous monitoring of patients involves collecting and analyzing sensory data from a multitude of sources. To overcome communication overhead, ensure data privacy and security, reduce data loss, and maintain efficient resource usage, the processing and analytics are moved close to where the data are located (e.g., the edge). However, data quality (DQ) can be degraded because of imprecise or malfunctioning sensors, dynamic changes in the environment, transmission failures, or delays. Therefore, it is crucial to keep an eye on data quality and spot problems as quickly as possible, so that they do not mislead clinical judgments and lead to the wrong course of action. In this article, a novel approach called federated data quality profiling (FDQP) is proposed to assess the quality of the data at the edge. FDQP is inspired by federated learning (FL) and serves as a condensed document or a guide for node data quality assurance. The FDQP formal model is developed to capture the quality dimensions specified in the data quality profile (DQP). The proposed approach uses federated feature selection to improve classifier precision and rank features based on criteria such as feature value, outlier percentage, and missing data percentage. Extensive experimentation using a fetal dataset split into different edge nodes and a set of scenarios were carefully chosen to evaluate the proposed FDQP model. The results of the experiments demonstrated that the proposed FDQP approach positively improved the DQ, and thus, impacted the accuracy of the federated patient similarity network (FPSN)-based machine learning models. The proposed data-quality-aware federated PSN architecture leveraging FDQP model with data collected from edge nodes can effectively improve the data quality and accuracy of the federated patient similarity network (FPSN)-based machine learning models. Our profiling algorithm used lightweight profile exchange instead of full data processing at the edge, which resulted in optimal data quality achievement, thus improving efficiency. Overall, FDQP is an effective method for assessing data quality in the edge computing environment, and we believe that the proposed approach can be applied to other scenarios beyond patient monitoring.

ANTi-Vax: a novel Twitter dataset for COVID-19 vaccine misinformation detection

OBJECTIVES

COVID-19 (SARS-CoV-2) pandemic has infected hundreds of millions and inflicted millions of deaths around the globe. Fortunately, the introduction of COVID-19 vaccines provided a glimmer of hope and a pathway to recovery. However, owing to misinformation being spread on social media and other platforms, there has been a rise in vaccine hesitancy which can lead to a negative impact on vaccine uptake in the population. The goal of this research is to introduce a novel machine learning-based COVID-19 vaccine misinformation detection framework.

STUDY DESIGN

We collected and annotated COVID-19 vaccine tweets and trained machine learning algorithms to classify vaccine misinformation.

METHODS

More than 15,000 tweets were annotated as misinformation or general vaccine tweets using reliable sources and validated by medical experts. The classification models explored were XGBoost, LSTM, and BERT transformer model.

RESULTS

The best classification performance was obtained using BERT, resulting in 0.98 F1-score on the test set. The precision and recall scores were 0.97 and 0.98, respectively.

CONCLUSION

Machine learning-based models are effective in detecting misinformation regarding COVID-19 vaccines on social media platforms.

P5419Cardiac reverse remodeling and recovery in dilated cardiomyopathy medication-naive patients requiring durable left ventricular assist device support

Background

Occasionally new onset cardiomyopathy patients (pts) present late and with such advanced disease stage that they cannot tolerate heart failure (HF) drug therapy. We sought to investigate the cardiac recovery (CR) potential following a combination of left ventricular assist device (LVAD) and guideline-directed HF drug therapy in this medication-naive population.

Methods

Chronic advanced HF requiring durable continuous-flow LVAD were prospectively evaluated. Patients with acute HF (myocarditis etc.) or post LVAD follow up <3 months were excluded. The “meds-treated” group (n=203) comprised patients treated adequately with at least one neurohormonal blocking agent during their HF history (b-blocker, ACEI/ARB, Aldosterone antagonist) and “meds-naive” group (n=8) comprised patients who were never before treated adequately with any HF medication. Baseline and follow up clinical, hemodynamic, imaging and laboratory data were analyzed. LVAD patients were phenotyped as CR responders or non responders, based on published predefined criteria.

Results

Univariate analysis showed that “med-naive” patients were younger, more likely to be on intravenous vasoactive agents, temporary mechanical support and with lower INTERMACS profile before LVAD implantation. Interestingly, no differences were seen in HF symptoms duration or other comorbidities. Baseline and follow up hemodynamics were similar in both groups, besides higher right atrial pressure pre-LVAD in the “meds-naive” group (16 vs 11 mmHg; p=0.04). Baseline echocardiographic (including LV dilation) and biochemical parameters revealed no differences between the groups, besides lower LVEF and higher BNP in the “meds-naive” group (14 vs 19%; p=0.03 and 2352 vs 1270; p=0.03, respectively). CR rates were significantly higher on “meds-naive” versus “meds-failed” group (50.0 vs 13.8%; p=0.005). Despite higher cardiac recovery rates in the “meds-naive” group the time course and magnitude of the favorable functional and structural response was similar among the CR responders of each of the 2 groups.

Conclusion

Young patients with new onset dilated cardiomyopathy sometimes present late, with advanced disease stage, unable to tolerate HF medications and requiring durable LVAD support. This patient population appears to have a potential for CR up to 50% and this could be factored in decisions surrounding their long-term therapeutic options.

P2628Predicting cardiac recovery before durable left ventricular assist device implantation in advanced heart failure patients

Background

Predicting cardiac recovery (CR) in advanced heart failure (HF) patients before left ventricular assist device (LVAD) implantation remains challenging. This study sought to investigate whether CR after LVAD unloading can be predicted by cardiac functional and structural parameters together with clinical characteristics.

Methods

From 2008 to 2016, consecutive advanced chronic HF patients (N=347) supported with durable continuous-flow LVADs were prospectively evaluated. Patients with acute HF etiologies or without adequate post-LVAD follow up (<3 months) were excluded. A great variety of clinical characteristics were evaluated in the remaining 285 subjects. LVAD patients were phenotyped while on support, as CR Responders or Non Responders, based on published predefined echocardiographic criteria. Multivariable logistic regression was used to form the model and the Utah Cardiac Recovery (UCAR) score was created from the regression beta coefficients of the final model.

Results

CR occurred in 13.7% of patients. Univariate analysis showed that responders were more likely to be young, female, non-ischemic cardiomyopathy, with shorter HF symptoms duration and no prior cardiac surgery. They had lower blood urea nitrogen and were more likely to be on temporary mechanical support before LVAD. The multivariable UCAR model (AUC=0.755; p<0.001) predicted CR using 3 clinical parameters – Figure.

Conclusion

Univariate and multivariable predictors of CR include both modifiable and non-modifiable patient characteristics that are known prior to LVAD implantation. The UCAR score can serve as a practical tool for targeted patient selection to implement protocols that facilitate CR in the advanced HF patient subpopulation that is most likely to respond.

Improvement of RNA-SIP by pyrosequencing to identify putative 4-n-nonylphenol degraders in activated sludge

Nonylphenols (NP) have estrogenic potential because of their phenolic ring, but the organisms involved in the degradation of this alkylated phenol remain unidentified. Using 16S ribosomal RNA (rRNA)-based stable isotope probing (SIP) and a new method based on pyrosequencing, we identified the bacteria involved in the degradation of the aromatic ring of [U-ring-(13)C] 4-n-NP in aerobic sludge. The first order degradation rate of 4-n-NP was 5.5d(-1). Single strand conformation polymorphism of density-separated labeled and unlabeled 16S rRNA showed significant differences and enabled selection of four representative fractions for pyrosequencing. Nineteen phylotypes showed a significant enrichment in the heavy fraction in the labeled pulse. The relative abundances of these phylotypes were combined with the RNA concentration of each fraction to yield a simple model of the distribution of each phylotype across the gradient. This model was used to estimate the percentage of labeling for each phylotype. The sequences showing the highest labeling (11%) were closely related to Afipia sp. but represented only 2 % of the RNA in the heavy fraction of the labeled pulse. The sequences representing the largest proportion of the RNA in the heavy fraction were related to Propionibacterium acnes and Frateuria aurantia, which are known to possess enzymes for phenol degradation. The model shows that despite Afipia having the highest (13)C enrichment, other species encoding phenol degradation pathways are responsible for more (13)C incorporation. Last, we showed that some species represent 12% of the total RNA but contain only 1% (13)C above natural abundance.

Cutaneous leishmaniasis in Al-Badarna: a prospective study among school children

This paper presents the results of a prospective study cutaneous liashmaniasis in 2 schools in Al-Badarna, during the period from July to December 1992. The pupils were interviewed and examined. Positive cases were referred to the Health Unit for treatment. Of 260 pupils 22 (8.46%) were found to be infected with active lesions.