Synthetic Samples Research Articles

Developing a novel ultraselective and ultrasensitive label-free direct spectrofluorimetric nanobiosensor for direct highly fast field detection of explosive triacetone triperoxide

BackgroundDirect detection of the notorious explosive triacetone triperoxide (TATP) is very difficult because it lacks facile ionization and UV absorbance or fluorescence. Besides, the current indirect methods are time-consuming and need a pre-step for TATP cleavage to hydrogen peroxide. Moreover, they commonly show significant false-positive results in the presence of some camouflage which limits their field applications. Herein, for the first time, a novel label-free field-applicable spectrofluorimetric nanobiosensor was developed for direct TATP detection using a novel activated-protein protected gold nanocluster (ABSA-AuNCs; QY = 28.3 %) synthesized by a combined protein-assisted-ultrasonication procedure. ResultsThe ABSA-AuNCs revealed a fluorescence spectrum centered at 330.0 nm which was significantly quenched by TATP (binding constant = 154.06 M-1; ΔG = −12.5 kJ mol−1; E(%) = 88.5 %). This phenomenon was used as a basis for direct TATP quantification, providing a working range of 0.01–40.0 mg L−1 and a detection limit of 6.7 μg L−1 which is the lowest LOD provided for TATP detection up to now. A %RSD of 0.9 % and 1.56 % was obtained for repeatability and inter-day reproducibility, respectively. The selectivity was checked against a variety of camouflages, revealing ultra-selectivity. Several synthetic samples prepared by several camouflages and real samples (clay soil and real water media) were analyzed, revealing quantitative recoveries of TATP. SignificanceDuring the production of the notorious explosive TATP, it can be discharged into water and soil. This novel method eliminated the false-positive results of traditional methods and is applicable for direct quantitative detection of camouflaged TATP and its residues in real soil and water samples in a highly short response time (2 min). The camouflaged TATP analysis is important for tracking the terrorist attacks in field conditions and analysis of soil and water can provide a first indication of the location of the production site.

Ceftriaxone photodegradation in wastewater using AgCuFe2O4/ZnO Almond-like heterogeneous nanocatalyst anchored on multi walled carbon nanotubes: Synthesis, characterization, mechanism study, and bioassay effluent

This study investigated the Ceftriaxone (CFT) degradation by photocatalytic process using AgCuFe2O4@MWCNT/ZnO as novel nanophotocatalyst. Firstly, the catalyst was prepared by microwave-assisted chemical coprecipitation method as a simple, fast, and green procedure with high efficiency. FESEM, EDS, Mapping, Line scan, FTIR, XRD, BET, DRS, PL, TGA, and VSM analyzes was carried out to identify the characteristics of the AgCuFe2O4@MWCNT/ZnO. The effect of key operational parameters for CFT photodegradation including solution pH, catalyst dosage, CFT concentration, and irradiation time was elaborated. Under optimal photocatalytic conditions (pH 7, 5 mg/L of CFT concentration, catalyst dosage of 0.24 g/L, and 60 minutes of irradiation) removal efficiency reached 90.1 % in synthetic samples and 75 % in real wastewater samples. •OH and •O2– are the dominant generated reactive oxygen species in the process. The kinetic study of the process revealed that the CFT degradation process followed the pseudo-first-order kinetic and Langmuir-Hinshelwood models with kc = 0.412 mg/L.min and kL-H =0.053 L/mg. Photocatalytic process indicated a powerful ability for mineralization of CFT (85.31 % of COD degradation). After four catalyst recovery cycles, the CFT degradation efficiency was achieved by 58.14 %, demonstrating the catalyst's recovery capability and chemical stability. The results of toxicity assessment using lettuce and basil seeds germination indicated a significant detoxification of CFT-containing effluent compared to the untreated CFT-containing wastewater. This study offers AgCuFe2O4@MWCNT/ZnO as effective, stable, and competitive catalyst for hastening and enhancing the photocatalytic process to mitigate environmentally related pollutants of high concern.

A simple recipe for designing multicomponent ultra-high temperature ceramic classes by using structure maps coupled with machine learning

Successful synthesis of novel high entropy ceramic (HEC) for ultra-high temperature application classes, namely, borides, carbides, and nitrides, has been experiencing a bottleneck in having a suitable design and successful synthesis strategy. Producing high-entropy ultra-high-temperature ceramics from their oxides offers a major processing benefit, while employing a design approach using machine learning enhances the efficiency of the formation of single-phase HECs. In this regard, we propose a generalized strategy to generate a semi-synthetic database for each of these classes using literature data and atomic environment mapping-based structure plots, which can further be used to build machine learning models. The imbalance of the dataset was addressed using adaptive synthetic sampling and the edited nearest neighbors technique. The trained models are able to accurately predict over 90% of the single-phase chemistry for each of the classes. Furthermore, a few compositions representing these classes were successfully synthesized from the corresponding oxide mixture to validate the effectiveness of the proposed strategy.

Association of rare earth elements with secondary mineral phases formed during alkalinization of acid mine drainage

Rare Earth Elements (REE) are present in acid mine drainage (AMD) in micromolar concentrations and AMD discharge may lead to an environmental risk. Alkaline Passive Treatment Systems (PTS) are often used to treat AMD and trap toxic trace elements. This study was set up to identify mechanisms by which REE are trapped in or on secondary phases formed in a PTS. Batch alkalinization experiments were performed to simulate PTS by sequentially increasing the pH of AMD collected from the Tharsis mining area inside the Iberian Pyrite Belt and synthetic AMD water samples via CaCO3 addition. The solids that precipitated up to pH ~4 and between pH 4–6 were collected and characterized by Laser Ablation Inductively Coupled Plasma Mass Spectrometry (LA-ICP-MS) in combination with Scanning Electron Microscope/Energy Dispersive X-ray spectroscopy (SEM/EDX) and synchrotron-based X-ray Absorption Spectroscopy (XAS) and synchrotron-based Micro-X-ray Fluorescence (μ-XRF). Results reveal that REE are mostly scavenged between pH 4–6 in association with Al and Fe phases, whereas a smaller fraction is scavenged at pH ~4 by association with gypsum. Synchrotron-based analysis evidences the incorporation of La3+ into the gypsum structure by substituting Ca2+, indicating a co-precipitation mechanism with gypsum occurring mainly at low pH. Results from parallel adsorption and co-precipitation tests suggest that the REE scavenging between pH 4–6 could be due to a combination of adsorption and co-precipitation on Al(OH)3 and ferrihydrite. This implies that in PTS, REE would be mainly found in Al- (and Fe-) oxyhydroxides occurring in deeper layers of the PTS, i.e., where higher pH-values occur, though a small fraction, especially the light REE, could also be found incorporated into gypsum in the upper layers.

A new chemometrically optimised HPLC method for the simultaneous determination of imatinib mesylate and capecitabine, the active medicinal components

ABSTRACT Anticancer drugs such as imatinib mesylate (IMP) and capecitabine (CAP) and their metabolites enter the water cycle through wastewater treatment plant effluents or directly into the water cycle due to their limited biodegradability in typical wastewater treatment plants. However, their long-term effects on water resources are unknown due to their toxicity. The hazards concerned with chronic exposure to these substances raise concerns about their impact on organisms, environmental safety and biodiversity conservation. Therefore, in this study, a chemometrically optimised HPLC method was developed for the simultaneous determination of trace amounts of cytotoxic agents IMP and CAP used in cancer treatment in environmental water, synthetic urine samples and drug samples. Using the Box-Behnken design, optimum conditions discovered using the traditional method were optimised. The effect of each element was evaluated in the design by four factors based on HPLC determination. For the examined levels, mobile phase 1, mobile phase 2 ratio, flow rate, and column temperature parameters for the HPLC method were determined. Using regression analysis on the experimental results, the factors influencing resolution were identified. To achieve the reliability of the results, the Analysis of Variance (ANOVA) test was used. The ANOVA test was employed to verify the consistency of the data. With the information gathered, a model was developed. HPLC was used to examine the target chemicals. A good linearity (R2 > 0.99) was established for the IMP and CAP in the concentration ranges of 1.18–23.59 µg mL−1 and 0.76–15.09 µg mL−1, respectively, under optimum circumstances. The recoveries for spiked samples at three concentration levels ranged from 98.31 to 101.11. The created method has been applied to synthetic urine samples, regional environmental water samples, and pharmaceutical samples with success.

Inexpensive and colorimetric RNA detection at ambient temperature with a cell-free protein synthesis platform

While many proposals of paper-based diagnostics utilize cell-free gene expression systems, these assays oftentimes suffer from the need for temperature cycling and high operational costs, particularly for developing countries. Here, we explore and report the experimental conditions for the colorimetric detection of viral RNA with an in vitro transcription/translation assay that uses crude E. coli extracts at room temperature where the signal amplification is aided by body heat. Clinically-relevant concentrations of RNA (ca. 600 copies/test) were detected from synthetic RNA samples. The activation of cell-free gene expression was achieved using toehold-switch-mediated riboregulatory elements that are specific to RNA sequences. The colorimetric output was generated by the α-complementation of β-galactosidase ω-fragment (LacZω) with cell-free expressed LacZα, using an X-gal analogue as a substrate. The estimated cost of a single reaction is as low as ~ 0.26 euro/test, which may help to facilitate the accessibility of the diagnostic kit in developing countries. With future optimizations and bacterial strain engineering, production costs can be even further brought down, and the test times can be shortened.Graphical

Mitigation of urban road collapses based on machine learning via integrating susceptibility assessment and geophysical detection validation

Road collapse is a frequent and damaging disaster in cities. The complexity and uncertainty inherent in urban environments pose significant challenges to mitigating road collapses. This paper presents a novel framework integrating machine learning-based susceptibility assessment and geophysical detection validation for urban road collapse risk reduction. Three oversampling techniques, random oversampling, synthetic minority oversampling technique for nominal and continuous features (SMOTENC), and adaptive synthetic sampling (ADASYN), are first utilized to implement data augmentation on urban road collapse accident samples. Subsequently, three machine learning models, support vector machine (SVM), random forest (RF), and extreme gradient boosting (XGBoost), are developed to evaluate road collapse susceptibility by extracting collapse-inducing patterns from historical accident data. Particularly, on-site geophysical hazard detection is conducted to validate the assessment results. The results demonstrate that XGBoost with SMOTENC achieves satisfactory performance in identifying road collapses with accuracy (0.9608) and AUC (0.9796). The spatial distribution of road collapse susceptibility in Shanghai central area follows a high-moderate-low pattern from northwest to southeast. The geophysical detection reveals a correlation between higher road collapse susceptibility and increased severity of underground diseases, validating the generalization capacity of XGBoost in actual operational environments. Additionally, the structural problems of underground pipelines are identified as the most influential factors for urban road collapse. This research offers valuable insights for urban road collapse mitigation and resilience improvement of transportation infrastructure.

Sustainable Processes of Biosorption of Pb(II) Ions from Synthetic Wastewater Using Waste Biomass from Mullein Leaves

The aim of this study is to evaluate mullein (Verbascum thapsus) as a cost-effective and sustainable adsorbent for the biosorption of Pb(II) ions from synthetic wastewater samples. Biosorption of mullein was investigated as a function of initial Pb(II) concentration (25–400 mg L−1), biosorbent dosage (2–20 g L−1), solution pH (3–7), and contact time (10–120 min). Mullein as a material with a high affinity for Pb(II) ions had a biosorption efficiency of 98.56% under the optimal conditions: pH 6, initial concentration of Pb(II) at 100 mg L−1, contact time of 90 min, and biosorbent dosage of 20 g L−1. The FTIR spectra of mullein leaves showed that oxygen-containing functional groups on the surface are potentially active sites for the biosorption of Pb(II) ions. EDS analysis and the pHPZC value confirmed the adsorption of Pb(II) ions at the active sites of the mullein. Kinetic and isotherm data enabled insights into the modes of Pb(II) biosorption on the mullein surface which were best explained with the pseudo-second-order kinetic model and the Freundlich adsorption isotherm. Biosorption occurs on the mullein surface via multilayer adsorption. The reusability of mullein showed that the native biosorbents can be reused five times, showing the economic and sustainable benefit of this low-cost biosorbent material.

Predictive Modeling of COVID-19 Readmissions: Insights from Machine Learning and Deep Learning Approaches.

This project employs artificial intelligence, including machine learning and deep learning, to assess COVID-19 readmission risk in Malaysia. It offers tools to mitigate healthcare resource strain and enhance patient outcomes. This study outlines a methodology for classifying COVID-19 readmissions. It starts with dataset description and pre-processing, while the data balancing was computed through Random Oversampling, Borderline SMOTE, and Adaptive Synthetic Sampling. Nine machine learning and ten deep learning techniques are applied, with five-fold cross-validation for evaluation. Optuna is used for hyperparameter selection, while the consistency in training hyperparameters is maintained. Evaluation metrics encompass accuracy, AUC, and training/inference times. Results were based on stratified five-fold cross-validation and different data-balancing methods. Notably, CatBoost consistently excelled in accuracy and AUC across all tables. Using ROS, CatBoost achieved the highest accuracy (0.9882 ± 0.0020) with an AUC of 1.0000 ± 0.0000. CatBoost maintained its superiority in BSMOTE and ADASYN as well. Deep learning approaches performed well, with SAINT leading in ROS and TabNet leading in BSMOTE and ADASYN. Decision Tree ensembles like Random Forest and XGBoost consistently showed strong performance.

Enhancing local road pavement condition prediction using Bayesian-optimized ensemble machine learning and adaptive synthetic sampling technique

ABSTRACT Pavement condition prediction helps road agencies to schedule maintenance, rehabilitation, and reconstruction, and to allocate limited funds and resources to such activities. Compared to state highways, pavement performance prediction for local roads has received relatively little attention in the literature due to perceptions of low importance, low levels of investment in data collection, poor data quality, and high variation within the data elements. Additionally, local road pavement condition data may suffer from dataset imbalance, often leading to unreliable condition predictions. Hence, this paper introduces a methodology to predict local pavement condition using various single estimator and ensemble machine learning (ML) models along with the adaptive synthetic sampling method. The study develops nine (9) Bayesian-optimised ML models: category boosting (CatBoost), adaptive boosting, decision tree, extra trees, gradient boosting, light gradient-boosting machine, k-nearest neighbour, random forest, and artificial neural network. The ensemble ML and CatBoost were found to exhibit the best model performance, with an average testing accuracy of 0.82, specificity of 0.81, sensitivity of 0.63, and F-measure of 0.61. These results underscore the efficacy of ensemble ML models in pavement condition prediction. The proposed approach can be beneficial to local road agencies in their long-term planning, scheduling, and budgeting.

Luminescent carbon dots endowed with selective recognition of the carcinoid tumor biomarkers in biological fluids

Biomarkers are commonly used in both basic research and clinical practice as diagnostic tools. Particularly, cancer biomarkers can be applied to the early detection and diagnosis or screening the tumors. Of special note, malignant tumors can release serotonin (5-hydroxytryptamine, 5-HT). In fact, 5-HT is an excellent biomarker for carcinoid tumors since its level in blood plasma is hardly affected by other parameters. Moreover, secretion of 5-HT gives rise to elevated levels of 5-hydroxyindole-3-acetic acid (HIAA), the major metabolite of 5-HT, in urine samples of the patients with carcinoid tumors. 5-HT and HIAA are regarded as eligible biomarkers, of which concentrations in blood plasma and urine samples can be utilized for prognosis, and early diagnosis of carcinoid syndrome. Here, we unveil the facile synthesis and characterization of new Tb(III)-doped carbon dots (CDs) in the present work. Strikingly, CDs exhibited superior features for sensing 5-HT and HIAA. It is of worth to note that the Tb(III)-doped CDs induce phosphorescent response to both 5-HT and HIAA in water. Limit of detections (LODs) for 5-HT and HIAA were found to be 0.44 nM and 0.49 nM, respectively. Remarkably, these are the lowest reported values for luminescent systems to date. Moreover, CDs were utilized as chemosensors for 5-HT and HIAA in simulated blood plasma and synthetic urine samples, respectively. To this purpose, extensive luminescence measurements were performed. Accordingly, the obtained results attested that CDs were promising chemosensors for the detection of both 5-HT and HIAA in simulated blood plasma and synthetic urine samples, respectively. Last but not least, we successfully demonstrate that Tb(III)-doped CDs can be used for the detection of 5-HT in artificial cerebrospinal fluid as well as in conditions mimicking the environment within secretory vesicles where 5-HT is stored. To the best of our knowledge, this is the first example of CDs which can induce remarkable phosphorescent response to 5-HT and HIAA in biological fluids.

Enhancement of Sustainable Recycling Systems for Industrial Waste in South Korea via Hazardous Characteristics Analysis

The South Korean government has implemented an acceptance system to promote the high-quality recycling of waste. Industrial waste generators must provide “hazardous characteristics data” to recycling operators. Nonetheless, ~80% of industrial safety accidents in South Korea occur during recycling, most involving fire or explosions. Moreover, a gap in safety management exists during ‘Circular Resource’ acceptance if the target substance is not regarded as waste. In this study collected data on hazardous waste characteristics. From 62 waste generators, 72 waste samples were collected, accounting for most of the resources accepted for recycling, including waste synthetic polymers, slag, dust, waste sand, and waste foundry sand. Then, the hazardous characteristics, as stated in the Ministry of Environment notifications, were assessed. Leaching toxicity was detected in one slag sample and six dust samples. The Cd, Cu, As, Pb, Zn, Ni, Hg, F, and CN levels dissatisfied the Soil Contamination Warning Standard in 31 samples. Explosivity was not detected in any sample, whereas flammability was detected in one waste synthetic polymer sample. The results revealed 15 cases of potential flammability. Flammability is legally defined as below the criteria if the combustion speed criterion is not met. However, in the case of flame ignition, which could cause large fires and safety accidents, the relevant notification should be revised. In this study, we aimed to improve the gap between the hazardous waste management systems and industrial fields through actual measurements of hazardous characteristics. By doing so, we seek to contribute to the prevention of environmental and safety accidents. By continuously accumulating data and utilizing actual measurements, we aim to revise and enhance relevant regulations, ultimately improving the hazardous characteristics of waste management systems.

CARBO: Clustering and rotation based oversampling for class imbalance learning

Class imbalance of a data set is a crucial problem in machine learning where one class significantly outnumbers others. In such a data set, classification is a troublesome task for the standard classification algorithms, leading to bias towards the majority class. Different methods have been developed so far, such as oversampling, undersampling, and cost-sensitive learning, to deal with class imbalance circumstances. Among these techniques, oversampling technique does not suffer from the information loss and critical cost selection challenges. However, appropriate synthetic sample generation can be challenging and vulnerable to privacy leakage. This research proposed an oversampling technique, called CARBO, using threshold-based geometric rotation and majority class influenced clustering. Unlike the existing resampling approaches to class imbalance problem, we contribute to consider the data privacy and optimal sample generation together for effective oversampling. The performance of CARBO is evaluated using 44 benchmark imbalanced data set. The empirical analysis elucidates that CARBO can make boosting-based C4.5 ensemble classifiers perform higher for 73% of the data set than six state-of-the-art approaches. In addition, the theoretical compatibility analysis of CARBO demonstrates its robustness.

Evaluation of parameters in the removal of azo Red 40 dye using electrocoagulation

The objective of the present investigation was to evaluate the parameters that affect the efficiency of electrocoagulation in a batch system to remove the azo-benzoic dye Red 40 present in synthetic wastewater by determining the effect of the operational parameters of voltage and treatment time. A 12×6 cm iron electrode holder was constructed with PVC pipes and screws separated by 2 cm, and a 1 L beaker was used as the electrocoagulation cell. Synthetic wastewater samples of red 40 were characterized, where UV–VIS was used to measure absorbance before and after being treated in the electrocoagulation cell. The effects of voltages (10, 15, and 20 V) and treatment times up to 60 min were evaluated. As a result, a higher removal of 93.85 % was obtained at 60 min when a voltage of 10 V was used with a lower change in temperature and energy consumption of only 14.08 kJ. The increase in conductivity and pH by values greater than 1600 µs/cm and 11, respectively, indicate the presence of diluted iron at the end of treatment; therefore, if complemented with other treatment methods such as filtration to remove excess iron produced, can be used in treatment plants and reduce their environmental impact, for this research the FTIR analysis showed that the sludge has a dye composition with hydroxides of metals along with a slight trace of dye degradation.

FedLAP-DP: Federated Learning by Sharing Differentially Private Loss Approximations

Conventional gradient-sharing approaches for federated learning (FL), such as FedAvg, rely on aggregation of local models and often face performance degradation under differential privacy (DP) mechanisms or data heterogeneity, which can be attributed to the inconsistency between the local and global objectives. To address this issue, we propose FedLAP-DP, a novel privacy-preserving approach for FL. Our formulation involves clients synthesizing a small set of samples that approximate local loss landscapes by simulating the gradients of real images within a local region. Acting as loss surrogates, these synthetic samples are aggregated on the server side to uncover the global loss landscape and enable global optimization. Building upon these insights, we offer a new perspective to enforce record-level differential privacy in FL. A formal privacy analysis demonstrates that FedLAP-DP incurs the same privacy costs as typical gradient-sharing schemes while achieving an improved trade-off between privacy and utility. Extensive experiments validate the superiority of our approach across various datasets with highly skewed distributions in both DP and non-DP settings. Beyond the promising performance, our approach presents a faster convergence speed compared to typical gradient-sharing methods and opens up the possibility of trading communication costs for better performance by sending a larger set of synthetic images. The source is available at https://github.com/a514514772/FedLAP-DP.

A comparative study on noise filtering of imbalanced data sets

Class imbalance in data sets used to train classifiers can negatively affect the performance of the resulting models. A commonly used solution to address this issue is to oversample the data sets and supplement them with synthetic samples. However, oversampling can increase the level of noise in the data sets. Several sampling methods attempt to prevent this negative effect in a variety of ways, but there are still many open questions about which solutions might be appropriate in different cases. In our study, we compared the impact of different noise filtering methods on relatively small, imbalanced synthetic data sets and then examined how noise filters perform as a preprocessing step of sampling methods following different sampling strategies. The success of the noise filtering was gauged through the performance of k-Nearest Neighbours models built on the balanced data sets. Our results highlight the importance of cleaning the minority class and provide insight into which noise filtering approaches might be useful as a first step to oversampling on imbalanced noisy data sets. Among the noise filtering methods included in our study, the GMM-based ones perform well on highly noisy imbalanced data sets. It is also worth highlighting some versions of ENN, which are very effective when the noise level is moderate.

Local strain heterogeneity associated with Al/Si ordering in anorthite, CaAl2Si2O8, with implications for thermodynamic mixing behavior and trace element partitioning in plagioclase feldspars

Abstract Hard Mode IR powder absorption spectroscopy has been used to characterize local strain relaxation associated with Al/Si ordering in a suite of synthetic anorthite samples with structural states that vary from a high degree of Al/Si order through a metastable incommensurate structure at intermediate states of order to long-range order with I1 symmetry. The dominant feature accompanying the changing structural states is line broadening, which has been quantified by autocorrelation analysis and is attributed to local heterogeneous strain variations on a length scale of at least 1–5 unit cells. The autocorrelation results are consistent with contributions to the line broadening as being due to order parameters for both the C1 → I1 and I1 → P1 transitions, which couple biquadratically, λQod2Qdispl2. Close correlation with enthalpy variations from previously published calorimetric data indicates that the driving force for ordering can be understood in terms of elimination of strain fields arising from accommodating more or less rigid AlO4 and SiO4 tetrahedra in the feldspar framework. The metastable incommensurate structure of anorthite is closely analogous to the stable incommensurate structure that develops at intermediate compositions in the plagioclase solid solution, confirming that the same strain relaxation mechanism dominates the properties and behavior of all structural states across the solid solution. Elimination of strain heterogeneity by ordering on the basis of I1 symmetry determines the form of non-ideal mixing shown by the solid solution at high temperatures, and changes in elastic properties may contribute to a break in the slope of partitioning of trace elements between crystals and melt.

Deconvolution and Growth curve for Thermoluminescence glow curves

In this work, a new program is presented that, in addition to performing the deconvolution of Thermoluminescence glow curves, allows the user to store multiple deconvolution data. This data is stored in an internal database, which allows the fitted parameters to be compared. After these deconvolutions, the program allows the user to summarize the values of the selected parameters. These values can be fitted with some of the available functions to observe the behavior of each parameter for each isolated peak. The program was tested with data from the GLOCANIN project to observe the accuracy of the deconvolution in synthetic and real samples with different number of peaks. Natural quartz grains obtained from sediments and irradiated with different doses were used for the Growth curves. After deconvolution, five peaks were identified, which showed different behaviors; these new data were fitted with different functions using the same program.

Comparative analysis of machine learning algorithms for predicting diarrhea among under-five children in Ethiopia: Evidence from 2016 EDHS.

Background: Diarrhea is a major cause of mortality and morbidity in under-5 children globally, especially in developing countries like Ethiopia. Limited research has used machine learning to predict childhood diarrhea. This study aimed to compare the predictive performance of ML algorithms for diarrhea in under-5 children in Ethiopia. Methods: The study utilized a dataset of 9501 under-5 children from the Ethiopia Demographic and Health Survey 2016. Five ML algorithms were used to build and compare predictive models. The model performance was evaluated using various metrics in Python. Boruta feature selection was employed, and data balancing techniques such as under-sampling, over-sampling, adaptive synthetic sampling, and synthetic minority oversampling as well as hyper parameter tuning methods were explored. Association rule mining was conducted using the Apriori algorithm in R to determine relationships between independent and target variables. Results: 10.2% of children had diarrhea. The Random Forest model had the best performance with 93.2% accuracy, 98.4% sensitivity, 85.5% specificity, and 0.916 AUC. The top predictors were residence, wealth index, and child age, number of living children, deworming, wasting, mother's occupation, and education. Association rule mining identified the top 7 rules most associated with under-5 diarrhea in Ethiopia. Conclusion: The RF achieved the highest performance for predicting childhood diarrhea. Policymakers and healthcare providers can use these findings to develop targeted interventions to reduce diarrhea. Customizing strategies based on the identified association rules has the potential to improve child health and decrease the impact of diarrhea in Ethiopia.

Realistic morphology-preserving generative modelling of the brain

Medical imaging research is often limited by data scarcity and availability. Governance, privacy concerns and the cost of acquisition all restrict access to medical imaging data, which, compounded by the data-hungry nature of deep learning algorithms, limits progress in the field of healthcare AI. Generative models have recently been used to synthesize photorealistic natural images, presenting a potential solution to the data scarcity problem. But are current generative models synthesizing morphologically correct samples? In this work we present a three-dimensional generative model of the human brain that is trained at the necessary scale to generate diverse, realistic-looking, high-resolution and morphologically preserving samples and conditioned on patient characteristics (for example, age and pathology). We show that the synthetic samples generated by the model preserve biological and disease phenotypes and are realistic enough to permit use downstream in well-established image analysis tools. While the proposed model has broad future applicability, such as anomaly detection and learning under limited data, its generative capabilities can be used to directly mitigate data scarcity, limited data availability and algorithmic fairness.