Clustering with missing and left‐censored data: A simulation study comparing multiple‐imputation‐based procedures

Applicability of machine learning models for the assessment of long-term pollutant leaching from solid waste materials

Column leaching tests are a common approach for evaluating the leaching behavior of contaminated soil and waste materials, which are often reused for various construction purposes. Standardized up-flow column leaching tests typically require about 7 days of laboratory work to evaluate long-term leaching behavior accurately. To reduce testing time, we developed linear and ensemble models based on parametric and non-parametric Machine Learning (ML) techniques. These models predict leachate concentrations of relevant chemical compounds at different Liquid-to-Solid ratios (LS) based on measurements at lower LS values. The ML models were trained using 82 column leaching test samples for Construction and Demolition Waste materials collected in Germany during the last two decades. R-Squared values measuring models' performance are as follows: Sulfate = 0.94, Vanadium = 0.97, Chromium = 0.82, Copper = 0.92, group of 15 (US-EPA) PAHs = 0.98 (values averaged over predictive models for LS 2 and 4). Sensitivity analysis utilizing the Shapley Additive Explanation value indicates that in addition to the concentrations of the considered compound at LS<=1, electrical conductivity and pH are the most critical features of each model, while concentrations of other compounds also play a minor role.

Multiobjective semisupervised learning with a right-censored endpoint adapted to the multiple imputation framework

Semisupervised learning aims to use additional knowledge in the search for data structure. In clinical applications, including predictive information in the construction of a data-driven classification is of major importance. This work was motivated by a study that aimed to identify different patterns of immune parameters that would be associated with relapse-free survival in a cohort of breast cancer patients. Supervised and unsupervised objectives can be concomitantly optimized using multiobjective optimization. We propose such a procedure that addresses two challenges in the semisupervised approach, that is, missing data and additional knowledge based on survival time. The former was handled by using multiple imputation and consensus clustering. Survival information was incorporated in the supervised objective through the estimation of a cross-validation error of a Cox regression. A simulation study was performed to assess the performance of the proposed procedure. On complete datasets, the performances were compared to those of an existing modified multiobjective semisupervised learning method. The added value of including the survival data in the learning process was assessed by comparing the procedure to unsupervised learning. The proposed procedure showed better performance than the existing method, notably in the selection of the number of clusters. On incomplete datasets, the procedure showed little sensitivity to most of its parameters, even though a high number of imputations and partition initialization seeds improved the performance. The performance was degraded with a high proportion of missing data (40%) and with more ambiguous data structures. Simulation results and application on real data support the conclusion that our procedure enables the construction of a classification associated with a right-censored endpoint on a possibly incomplete dataset.

Heterogeneity in Blood Biomarker Trajectories After Mild TBI Revealed by Unsupervised Learning

Concussions, also known as mild traumatic brain injury (mTBI), are a growing health challenge. Approximately four million concussions are diagnosed annually in the United States. Concussion is a heterogeneous disorder in causation, symptoms, and outcome making precision medicine approaches to this disorder important. Persistent disabling symptoms sometimes delay recovery in a difficult to predict subset of mTBI patients. Despite abundant data, clinicians need better tools to assess and predict recovery. Data-driven decision support holds promise for accurate clinical prediction tools for mTBI due to its ability to identify hidden correlations in complex datasets. We apply a Locality-Sensitive Hashing model enhanced by varied statistical methods to cluster blood biomarker level trajectories acquired over multiple time points. Additional features derived from demographics, injury context, neurocognitive assessment, and postural stability assessment are extracted using an autoencoder to augment the model. The data, obtained from FITBIR, consisted of 301 concussed subjects (athletes and cadets). Clustering identified 11 different biomarker trajectories. Two of the trajectories (rising GFAP and rising NF-L) were associated with a greater risk of loss of consciousness or post-traumatic amnesia at onset. The ability to cluster blood biomarker trajectories enhances the possibilities for precision medicine approaches to mTBI.

SARS-CoV-2 Antibodies in Hemodialysis Patients Six Months after Infection Compared to Healthcare Workers

Background

The humoral response to SARS-CoV-2 infection in hemodialysis patients needs to be clarified.

Methods

In this retrospective study performed in two dialysis facilities, we measured the circulating levels of SARS-CoV-2 antibodies in patients who were on maintenance hemodialysis during the first wave of the epidemic in March and April 2020 and were still alive 6 months later. We also investigated associations between the patients diagnosed as infected during the first wave and several clinical, biological, and radiological parameters of COVID-19. Finally, we compared these circulating levels of SARS-CoV-2 antibodies with those of a control group of healthcare workers infected during the same period.

Results

Of the 299 hemodialysis patients who recovered from the first wave of the epidemic 6 months before, 59 had a positive SARS-CoV-2 antibody whereas only 45 patients were diagnosed as infected during the first wave of the epidemic. All infected hemodialysis patients developed circulating antibodies. Using a clustering method, a significant correlation was identified between the cluster with the lowest circulating levels of SARS-CoV-2 antibodies and the severity of COVID-19 based on several parameters including CRP, BNP, lymphocyte count, neutrophil-lymphocyte ratio, and oxygen requirements, as well as pulmonary involvement on chest scan. Moreover, the circulating levels of the SARS-CoV-2 antibodies in surviving hemodialysis patients (n = 59) were similar to those of the control group (n = 17).

Conclusion

The main finding of this study is that all of the surviving hemodialysis patients who were diagnosed with SARS-CoV-2 infection from March to April 2020 developed a persistent humoral response with significant circulating levels of SARS-CoV-2 antibodies, 6 months later. Another important finding is that surviving hemodialysis patients who had more severe disease had lower circulating levels of SARS-CoV-2 antibodies. Finally, circulating levels of SARS-CoV-2 antibodies were similar in surviving hemodialysis patients and healthcare workers without kidney disease.