Christian Heumann

Coverage optimization in wireless sensor networks (WSNs) is a fundamental yet NP-hard problem that directly affects monitoring quality and efficiency. Existing solutions mainly rely on meta-heuristic algorithms that use fitness-based evaluations, which often incur high computational overhead, slow convergence, and limited scalability, particularly in real-time or high-precision monitoring scenarios. In this article, we examine the relationship between effective coverage area and redundant distances in an analytical manner. We then propose reformulating WSN coverage optimization as a distance-gradient-based convex optimization problem, which can be subsequently solved using the first-order gradient descent (GD) algorithm or the second-order quasi-Newton algorithm. Extensive comparative experiments against five representative meta-heuristic methods, the virtual force algorithm (VFA) and a general convex optimization algorithm (CVX), demonstrate that our approach achieves near-optimal coverage while preserving network connectivity within milliseconds, highlighting its advantages over existing methods for WSNs coverage optimization.

We evaluate the performance of targeted maximum likelihood estimation (TMLE) for estimating the average treatment effect in missing data scenarios under varying levels of positivity violations. We employ model- and design-based simulations, with the latter using undersmoothed highly adaptive lasso on the 'WASH Benefits Bangladesh' dataset to mimic real-world complexities. Five missingness-directed acyclic graphs are considered, capturing common missing data mechanisms in epidemiological research, particularly in one-point exposure studies. These mechanisms include also not-at-random missingness in the exposure, outcome, and confounders. We compare eight missing data methods in conjunction with TMLE as the analysis method, distinguishing between non-multiple imputation (non-MI) and multiple imputation (MI) approaches. The MI approaches use both parametric and machine-learning models. Results show that non-MI methods, particularly complete cases with TMLE incorporating an outcome-missingness model, exhibit lower bias compared to all other evaluated missing data methods and greater robustness against positivity violations across. In Comparison MI with classification and regression trees (CART) achieve lower root mean squared error, while often maintaining nominal coverage rates. Our findings highlight the trade-offs between bias and coverage, and we recommend using complete cases with TMLE incorporating an outcome-missingness model for bias reduction and MI CART when accurate confidence intervals are the priority.

Colorectal adenocarcinoma is caused in part by widespread epigenetic deregulation, yet the analysis of genome-wide DNA methylation of colorectal adenocarcinoma is complicated due to spatial correlation among CpG, multiscale patterns of differential methylation, and confounding cellular heterogeneity in bulk tissue. This study develops a simple yet effective framework that combines rigorous statistical modelling with modern machine learning-based prediction, on the Illumina 450K data from The Cancer Genome Atlas (TCGA) colorectal adenocarcinoma cohort. In our framework, differentially methylated regions (DMRs) were first detected using functional smoothing and permutation-based bump hunting, revealing both focal CpG island hypermethylation and broad hypomethylated domains spanning hundreds of kilobases. Next, we performed reference-based cell-type deconvolution and surrogate variable analysis (SVA) controlled immune/stromal admixture and hidden confounding effects, yielding well-calibrated single-site and region-level inference. Then, for tumour-status prediction, we empirically study the performance of classical Logistic Regression model against Random Forest, Gradient Boosting (XGBoost), and Feed-forward neural network; the results show that Logistic Regression achieves the lowest root-mean-square error and Brier score, reflecting its superior probability calibration. In general, our integrated framework provides biologically interpretable and highly predictive methylation signatures of colorectal cancer and offers a transferable baseline for future large-scale cancer epigenomics studies. Our code is publicly available at https://github.com/matekum/tcga-coad-methylation-ekum-2025a . Author summary: We present a reproducible framework for analyzing genome-wide DNA methylation in colorectal cancer using public TCGA data. Our aim is twofold. We started by making single-site and regional findings statistically reliable by accounting for spatial correlation, hidden confounding, and cell-type mixtures; and second, we evaluate practical classifiers for tumour vs. normal prediction. We smooth effect sizes across neighbouring CpG sites and use permutation tests to detect differentially methylated regions, while surrogate variable analysis and reference-based cell-fraction estimates reduce unwanted variation. We then compare logistic regression with random forests, gradient boosting, and a neural network using principal component features. Across models, we observe near-perfect ranking of samples and strong probability calibration, with logistic regression performing best on calibration. All code, parameters, and figure scripts are openly available so others can reproduce and adapt the framework. Beyond colorectal cancer, the approach provides a template for robust methylome analysis and predictive modelling in other large-scale epigenomic studies.

Measurement uncertainty is often treated as noise to be reduced, but in reality, it reflects a mismatch between point-based observations and the underlying stochastic nature of physical processes. In scenarios involving both regression targets and classification-based thresholds, this discrepancy can be systematically exploited for improved optimization. We propose a novel training approach based on a compound loss function that integrates two inputs describing the same target: a metric signal (e.g., a physical measurement) and a categorical estimate (e.g., derived from expert judgment, statistical aggregation, or repeated sampling). This dual-input strategy reflects how individual noisy observations differ from their expected behavior over time and uses this difference to optimize learning under aleatoric uncertainty. The approach is model-agnostic, scalable to multiple targets, and allows for tuning prior knowledge contributions via a weighting parameter. Using synthetic data derived from opto-semiconductor measurements, we demonstrate a +17.78% improvement in delivery conformity under realistic constraints, with only a marginal increase in normalized regression error (ΔMAE = +0.0073). These results highlight the method’s practical applicability and its scalability to complex, multi-target optimization scenarios.

Lifetime risks are a useful tool in quantifying health risks related to radiation exposure and play an important role in the radiation detriment and, in the case of radon, for radon dose conversion. This study considers the lifetime risk of dying from lung cancer related to occupational radon exposure. For this purpose, in addition to other risk measures, the lifetime excess absolute risk (LEAR) is mainly examined. Uncertainty intervals for such lifetime risk estimates and corresponding statistical methods are rarely presented in the radon literature. Based on previous work on LEAR estimates, the objective of this article is to introduce and discuss novel methods to derive uncertainty intervals for lifetime risk estimates for lung cancer related to occupational radon exposure. Uncertainties of two main components of lifetime risk calculations are modeled: uncertainties of risk model parameter estimates describing the excess relative risk for lung cancer and of baseline mortality rates. Approximate normality assumption (ANA) methods derived from likelihood theory and Bayesian techniques are employed to quantify uncertainty in risk model parameters. The derived methods are applied to risk models from the German “Wismut” uranium miners cohort study (full Wismut cohort with follow-up up to 2018 and sub-cohort with miners first hired in 1960 or later, designated as “1960+ sub-cohort”). Mortality rate uncertainty is assessed based on information from the WHO mortality database. All uncertainty assessment methods are realized with Monte Carlo simulations. Resulting uncertainty intervals for different lifetime risk measures are compared. Uncertainty from risk model parameters imposes the largest uncertainty on lifetime risks but baseline lung cancer mortality rate uncertainty is also substantial. Using the ANA method accounting for uncertainty in risk model parameter estimates, the LEAR in % for the 1960+ sub-cohort risk model was 6.70 with a 95% uncertainty interval of [3.26; 12.28] for the exposure scenario of 2 Working Level Months from age 18–64 years, compared to the full cohort risk model with a LEAR in % of 3.43 and narrower 95% uncertainty interval [2.06; 4.84]. ANA methods and Bayesian techniques with a non-informative prior yield similar results, whenever comparable. There are only minor differences across different lifetime risk measures. Based on the present results, risk model parameter uncertainty accounts for a substantial share of lifetime risk uncertainty for radon protection. ANA methods are the most practicable and should be employed in the majority of cases. The explicit choice of lifetime risk measures is negligible. The derived uncertainty intervals are comparable to the range of lifetime risk estimates from uranium miners studies in the literature. These findings should be accounted for when developing radiation protection policies, which are based on lifetime risks.