Robust Modeling Approach Research Articles

Machine Learning Readmission Risk Modeling: A Pediatric Case Study.

Background Hospital readmission prediction in pediatric hospitals has received little attention. Studies have focused on the readmission frequency analysis stratified by disease and demographic/geographic characteristics but there are no predictive modeling approaches, which may be useful to identify preventable readmissions that constitute a major portion of the cost attributed to readmissions. Objective To assess the all-cause readmission predictive performance achieved by machine learning techniques in the emergency department of a pediatric hospital in Santiago, Chile. Materials An all-cause admissions dataset has been collected along six consecutive years in a pediatric hospital in Santiago, Chile. The variables collected are the same used for the determination of the child's treatment administrative cost. Methods Retrospective predictive analysis of 30-day readmission was formulated as a binary classification problem. We report classification results achieved with various model building approaches after data curation and preprocessing for correction of class imbalance. We compute repeated cross-validation (RCV) with decreasing number of folders to assess performance and sensitivity to effect of imbalance in the test set and training set size. Results Increase in recall due to SMOTE class imbalance correction is large and statistically significant. The Naive Bayes (NB) approach achieves the best AUC (0.65); however the shallow multilayer perceptron has the best PPV and f-score (5.6 and 10.2, resp.). The NB and support vector machines (SVM) give comparable results if we consider AUC, PPV, and f-score ranking for all RCV experiments. High recall of deep multilayer perceptron is due to high false positive ratio. There is no detectable effect of the number of folds in the RCV on the predictive performance of the algorithms. Conclusions We recommend the use of Naive Bayes (NB) with Gaussian distribution model as the most robust modeling approach for pediatric readmission prediction, achieving the best results across all training dataset sizes. The results show that the approach could be applied to detect preventable readmissions.

Modelling complex investment decisions in Germany for renewables with different machine learning algorithms

Investment decisions in renewable energies are known to be influenced by many diverse drivers, e.g. social, political, geographic, economic and psychological. Non-comprehensive models are problematic since missed interactions might introduce bias.We implement a robust modelling approach by (1) using a large data set with 1.4 million solar installations and (2) three different machine learning algorithms (deep neural networks, gradient boosting, random forests). Generalized linear models serve as baseline and comparison.A high prediction accuracy can be achieved on the county level with deep neural networks (adjusted R2 = 0.86) and gradient boosting (adjusted R2 = 0.87). The most important drivers are population per county, followed by type of urbanisation and social variables like unemployment, with varying degree of importance for the different machine-learning algorithms. Our approach points out both differences and agreements across methods and therefore a higher confidence in their interpretation.

A 5-gene prognostic nomogram predicting survival probability of glioblastoma patients.

BackgroundGlioblastoma (GBM) remains the most biologically aggressive subtype of gliomas with an average survival of 10 to 12 months. Considering that the overall survival (OS) of each GBM patient is a key factor in the treatment of individuals, it is meaningful to predict the survival probability for GBM patients newly diagnosed in clinical practice.Material and MethodsUsing the TCGA dataset and two independent GEO datasets, we identified genes that are associated with the OS and differentially expressed between GBM tissues and the adjacent normal tissues. A robust likelihood‐based survival modeling approach was applied to select the best genes for modeling. After the prognostic nomogram was generated, an independent dataset on different platform was used to evaluate its effectiveness.ResultsWe identified 168 differentially expressed genes associated with the OS. Five of these genes were selected to generate a gene prognostic nomogram. The external validation demonstrated that 5‐gene prognostic nomogram has the capability of predicting the OS of GBM patients.ConclusionWe developed a novel and convenient prognostic tool based on five genes that exhibited clinical value in predicting the survival probability for newly diagnosed GBM patients, and all of these five genes could represent potential target genes for the treatment of GBM. The development of this model will provide a good reference for cancer researchers.

Real-time nondestructive monitoring of Common Carp Fish freshness using robust vision-based intelligent modeling approaches

In the current research, the potential of a novel method based on the artificial neural network was investigated to diagnose the freshness of common carp (Cyprinus carpio) during ice storage. Fish as an aquaculture product has high nutrients and low-fat content. So, people have consumed it as a safe and high-value foodstuff in their daily diet. Investigation of fish freshness is proposed as a significant issue in the aquaculture industry since fish spoils rapidly. The applied system of this study is comprised of the following steps: First, images of samples were captured and the pre-processing operation was done on the images. Then, particular channels including R, G, B, H, S, I, L*, a*, and b* were computed. Next, feature extraction was performed to obtain 6 types of texture features from each channel. Afterward, the hybrid Artificial Bee Colony-Artificial Neural Network (ABC-ANN) algorithm was applied to select the best features. Finally, the Support Vector Machine (SVM), K-Nearest Neighbor (K-NN) and Artificial Neural Network (ANN) algorithms as the most common methods were used to classify fish images. The best performance of the K-NN classifier was calculated in the k = 8 neighborhood size with the accuracy of 90.48. The best kernel function for the SVM algorithm was polynomial with C, sigma, and accuracy of 1, 2 and 91.52 percent, respectively. In this system, the input layer has consisted of 22 neurons based on the feature selection operation and 4 classes including most fresh, fresh, fairly fresh and spoiled have been used as the number of output layer. At the end, the best results of the MLP networks were achieved by LM learning algorithm and 6 neurons in the hidden layer with the 22–10–4 topology and accuracy of 93.01 percent. The achieved results demonstrate the high performance of the ANN classifier for evaluation of common carp freshness during ice storage as a rapid, accurate, non-destructive, real-time and automated method. It shows the potential of computer vision method in combination with artificial neural networks as an intelligent technique for evaluation of fish freshness.

Developing optimization & robust models for a mixed-model assembly line balancing problem with semi-automated operations

In this study, we present two new mathematical models for mixed-model assembly line balancing . The line is semi-automated and characterized by various alternative formats for the operations performed, which range from human operators and assisting robots to robotics solutions, where the operators may have different skill levels and operators on adjacent stations may cooperate. Given the cycle time, the first model minimizes the fixed and variable costs associated with setting up the assembly line. However, the second mathematical model involves a cycle time, which is variable and it changes between a lower bound and an upper bound. These bounds reflect the expected future changes in demand. The objective is to minimize both the costs and the cycle time. To obtain more accurate and reliable results, we present a robust solution for the different confidence levels ( α ) determined by managers. To verify the mathematical models, we considered different test problems and compared the robust solutions under three different conditions. The results showed that the robust modeling approach obtains a more reliable design than the traditional cost trade-off alone. The model allows decision makers to select assembly operations based on a better understanding of their decision impacts in both the short term and long term under conditions of demand uncertainty.

Reconstruction of Structural Flexibility And Acceptance Model (SFAM) in Green Marketing Mix Strategy

This research reconstructs the Structural Flexibility and Acceptance Model (SFAM), as a concept / theory / proposition step on statistical modeling. Existing and widely used statistical modeling, such as SEM, PLS, GSCA and WarpPLS, have flaws and still lack some flaws. SFAM reconstruction is done by reducing existing weaknesses and by adding some new and often needed statistical concepts to statistic users. Structural Flexibility and Acceptance Model (SFAM) is a very intense method of data analysis, especially in the field of social and humanities education. Research in the field in general is to investigate the system, in which one of the properties of the system is complex. The simplification solution can be done through statistical modeling. The reconstruction of Structural Flexibility and Acceptance Model (SFAM) will also be equipped with the concept of path analysis with category data, so it can be used to analyze category data. It is also integrated with a flexible and robust modeling approach based on nonparametric smoothing spline regression analysis (RNSS) capable of capturing the form of relationships that depend on empirical data, and the uniformity of free assumption-based models of assumptions and the assumption of homosexuality of variance errors.

Analytics of Heterogeneous Breast Cancer Data Using Neuroevolution

Breast cancer prognostic modeling is difficult since it is governed by many diverse factors. Given the low median survival and large scale breast cancer data, which comes from high throughput technology, the accurate and reliable prognosis of breast cancer is becoming increasingly difficult. While accurate and timely prognosis may save many patients from going through painful and expensive treatments, it may also help oncologists in managing the disease more efficiently and effectively. Data analytics augmented by machine-learning algorithms have been proposed in past for breast cancer prognosis; and however, most of these could not perform well owing to the heterogeneous nature of available data and model interpretability related issues. A robust prognostic modeling approach is proposed here whereby a Pareto optimal set of deep neural networks (DNNs) exhibiting equally good performance metrics is obtained. The set of DNNs is initialized and their hyperparameters are optimized using the evolutionary algorithm, NSGAIII. The final DNN model is selected from the Pareto optimal set of many DNNs using a fuzzy inferencing approach. Contrary to using DNNs as the black box, the proposed scheme allows understanding how various performance metrics (such as accuracy, sensitivity, F1, and so on) change with changes in hyper-parameters. This enhanced interpretability can be further used to improve or modify the behavior of DNNs. The heterogeneous breast cancer database requires preprocessing for better interpretation of categorical variables in order to improve prognosis from classifiers. Furthermore, we propose to use a neural network-based entity-embedding method for categorical features with high cardinality. This approach can provide a vector representation of categorical features in multidimensional space with enhanced interpretability. It is shown with evidence that DNNs optimized using evolutionary algorithms exhibit improved performance over other classifiers mentioned in this paper.

Computational Tools for Quantifying Concordance in Single-Cell Fate.

Cells are dynamic biological systems that interact with each other and their surrounding environment. Understanding how cell extrinsic and intrinsic factors control cell fate is fundamental to many biological experiments. However, due to transcriptional heterogeneity or microenvironmental fluctuations, cell fates appear to be random. Individual cells within well-defined subpopulations vary with respect to their proliferative potential, survival, and lineage potency. Therefore, methods to quantify fate outcomes for heterogeneous populations that consider both the stochastic and deterministic features of single-cell dynamics are required to develop accurate models of cell growth and differentiation. To study random versus deterministic cell behavior, one requires a probabilistic modelling approach to estimate cumulative incidence functions relating the probability of a cell's fate to its lifetime and to model the deterministic effect of cell environment and inheritance, i.e., nature versus nurture. We have applied competing risks statistics, a branch of survival statistics, to quantify cell fate concordance from cell lifetime data. Competing risks modelling of cell fate concordance provides an unbiased, robust statistical modelling approach to model cell growth and differentiation by estimating the effect of cell extrinsic and heritable factors on the cause-specific cumulative incidence function.

Temporal estimation of hydrodynamic parameter variability in stormwater constructed wetlands – The hysteresis effect during multi-rainfall events

Constructed wetland (CW) systems are increasingly implemented as appropriate water treatment facilities, the efficiency of which varies with time due to possible malfunctions such as clogging related to excessive loads of suspended solids or inadequate CW sizing. Numerical flow modeling was used to evaluate filtering performance and to improve the design of CWs. Such modeling usually involves the so-called van Genuchten-Mualem (vGM) soil hydrodynamic parameters which are difficult to measure. This paper focuses on the sensitivity of the model with respect to the vGM parameters revealing a predominant influence of the shape parameters and the saturated conductivity of the filter on the piezometric heads, during feeding and drainage. A simple, robust and low-cost inverse modeling approach, deterministic and stochastic, for the identification of the soil hydrodynamic properties from piezometric heads measured during successive storm events is also presented. The temporal variability of hydrodynamic parameters was assessed and analyzed with regards to modeling efficiency. Principal component analysis showed that the estimated hydrodynamic parameters from the feeding and drainage sub-periods were significantly different.

A Hybrid Approach for Model Order Reduction of Barotropic Quasi-Geostrophic Turbulence

We put forth a robust reduced-order modeling approach for near real-time prediction of mesoscale flows. In our hybrid-modeling framework, we combine physics-based projection methods with neural network closures to account for truncated modes. We introduce a weighting parameter between the Galerkin projection and extreme learning machine models and explore its effectiveness, accuracy and generalizability. To illustrate the success of the proposed modeling paradigm, we predict both the mean flow pattern and the time series response of a single-layer quasi-geostrophic ocean model, which is a simplified prototype for wind-driven general circulation models. We demonstrate that our approach yields significant improvements over both the standard Galerkin projection and fully non-intrusive neural network methods with a negligible computational overhead.

Evaluating the monetary values of greenhouse gases emissions in life cycle impact assessment

It is commonly acknowledged that greenhouse gas (GHG) emissions from anthropogenic sources accelerate climate change impacts. Efforts are made by governments and companies to reduce GHG emissions via policies and actions. In order to determine which actions to prioritize among many options, benefits of emission reductions are often monetized, to compare with the costs of action or with benefits that can be obtained from other actions. Life cycle assessment (LCA) is a commonly used tool to assess the amount of GHGs emitted over the life cycle of a service, policy or product system. However, the damage modelling of GHGs in life cycle impact assessment (LCIA) and its monetary values have not been separately evaluated. This hinders the application of LCA in relevant decision contexts. This study evaluates the cause-effect chains and associated monetary values of GHG in three LCIA methods LIME2, EPS2015 and ReCiPe2016. Among these three, EPS2015 covers most damage categories, including the ones on human health, ecosystem and social assets. ReCiPe2016 does not include social assets damages and LIME2 does not consider ecosystem damages in climate change impact. Human health damages are well estimated in all three methods, contributing to 70–97% of the GHG monetary values. The lack of data is a clear obstacle across methods. Further research is needed to develop comprehensive and robust modelling approach for ecosystem damages, which are not well covered in current LCIA methods. Moreover, due to the scope of environmental LCA, there is a lack of consideration on socio-economic consequences, which may not be negligible for climate change. The resulting monetary value of GHG, expressed in per tonne CO2-eq are 16, 160 and 140 US$2017 respectively in LIME2, EPS2015 and ReCiPe 2016. These monetary values are reasonable for use in decision contexts where LCA is applied. Further research is, however, needed to reduce the current uncertainty of at least 1–2 orders of magnitude.

Characterising and modelling free ammonia and ammonium inhibition in anaerobic systems

Inhibition by ammoniacal nitrogen, consisting of free ammonia (NH3) and ammonium ion (NH4+), has been widely investigated for anaerobic digestion. However, despite the large amount of research on the subject, ammoniacal nitrogen inhibition still threatens many anaerobic digesters. This paper presents (i) a method to reliably characterise ammoniacal nitrogen inhibition and (ii) a robust inhibition modelling approach. Results showed that NH3 and NH4+ inhibition need to be jointly determined, which can only be done by performing inhibition tests at various total ammoniacal nitrogen (TAN) concentrations and pH values. These test conditions were reliably achieved using the salts NH4HCO3 and NH4Cl without pH adjustment, rather than by using NH4Cl with pH adjustment. The use of only salts showed a lower pH change during the inhibition test (∼1.5 days), thereby decreasing the uncertainty in TAN speciation and strengthening the test and model outputs. A threshold inhibition function satisfactorily described (R2 > 0.99) the joint inhibition of NH3 and NH4+ on three distinct inocula, and provided a better description of the inhibition testing results than a non-competitive inhibition function (R2 ∼0.70). The key advantage of the proposed threshold inhibition function is its capacity to identify the inhibition lower limit (concentration where inhibition starts; KImin) and upper limit (concentration where inhibition is complete; KImax). The threshold inhibition function also identifies the 50% inhibition concentration (KI50) at the midpoint of KImin and KImax. Finally, experimental and model results show that at pH 7.3–7.7 and TAN concentrations above 2000 mgN·L−1, both NH3 and NH4+ contribute significantly to overall inhibition.

Supercritical carbon dioxide-based technologies for the production of drug nanoparticles/nanocrystals - A comprehensive review.

Low drug bioavailability, which is mostly a result of poor aqueous drug solubilities and of inadequate drug dissolution rates, is one of the most significant challenges that pharmaceutical companies are currently facing, since this may limit the therapeutic efficacy of marketed drugs, or even result in the discard of potential highly effective drug candidates during developmental stages. Two of the main approaches that have been implemented in recent years to overcome poor drug solubility/dissolution issues have frequently involved drug particle size reduction (i.e., micronization/nanonization) and/or the modification of some of the physicochemical and structural properties of poorly water soluble drugs. A large number of particle engineering methodologies have been developed, tested, and applied in the synthesis and control of particle size/particle-size distributions, crystallinities, and polymorphic purities of drug micro- and nano-particles/crystals. In recent years pharmaceutical processing using supercritical fluids (SCF), in general, and supercritical carbon dioxide (scCO2), in particular, have attracted a great attention from the pharmaceutical industry. This is mostly due to the several well-known advantageous technical features of these processes, as well as to other increasingly important subjects for the pharmaceutical industry, namely their "green", sustainable, safe and "environmentally-friendly" intrinsic characteristics. In this work, it is presented a comprehensive state-of-the-art review on scCO2-based processes focused on the formation and on the control of the physicochemical, structural and morphological properties of amorphous/crystalline pure drug nanoparticles. It is presented and discussed the most relevant scCO2, scCO2-based fluids and drug physicochemical properties that are pertinent for the development of successful pharmaceutical products, namely those that are critical in the selection of an adequate scCO2-based method to produce pure drug nanoparticles/nanocrystals. scCO2-based nanoparticle formation methodologies are classified in three main families, and in terms of the most important role played by scCO2 in particle formation processes: as a solvent; as an antisolvent or a co-antisolvent; and as a "high mobility" additive (a solute, a co-solute, or a co-solvent). Specific particle formation methods belonging to each one of these families are presented, discussed and compared. Some selected amorphous/crystalline drug nanoparticles that were prepared by these methods are compiled and presented, namely those studied in the last 10-15 years. A special emphasis is given to the formation of drug cocrystals. It is also discussed the fundamental knowledge and the main mechanisms in which the scCO2-based particle formation methods rely on, as well as the current status and urgent needs in terms of reliable experimental data and of robust modeling approaches. Other addressed and discussed topics include the currently available and the most adequate physicochemical, morphological and biological characterization methods required for pure drug nanoparticles/nanocrystals, some of the current nanometrology and regulatory issues associated to the use of these methods, as well as some scale-up, post-processing and pharmaceutical regulatory subjects related to the industrial implementation of these scCO2-based processes. Finally, it is also discussed the current status of these techniques, as well as their future major perspectives and opportunities for industrial implementation in the upcoming years.

Sub-basalt Imaging of Hydrocarbon-Bearing Mesozoic Sediments Using Ray-Trace Inversion of First-Arrival Seismic Data and Elastic Finite-Difference Full-Wave Modeling Along Sinor\u2013Valod Profile of Deccan Syneclise, India

Imaging below the basalt for hydrocarbon exploration is a global problem because of poor penetration and significant loss of seismic energy due to scattering, attenuation, absorption and mode-conversion when the seismic waves encounter a highly heterogeneous and rugose basalt layer. The conventional (short offset) seismic data acquisition, processing and modeling techniques adopted by the oil industry generally fails to image hydrocarbon-bearing sub-trappean Mesozoic sediments hidden below the basalt and is considered as a serious problem for hydrocarbon exploration in the world. To overcome this difficulty of sub-basalt imaging, we have generated dense synthetic seismic data with the help of elastic finite-difference full-wave modeling using staggered-grid scheme for the model derived from ray-trace inversion using sparse wide-angle seismic data acquired along Sinor–Valod profile in the Deccan Volcanic Province of India. The full-wave synthetic seismic data generated have been processed and imaged using conventional seismic data processing technique with Kirchhoff pre-stack time and depth migrations. The seismic image obtained correlates with all the structural features of the model obtained through ray-trace inversion of wide-angle seismic data, validating the effectiveness of robust elastic finite-difference full-wave modeling approach for imaging below thick basalts. Using the full-wave modeling also allows us to decipher small-scale heterogeneities imposed in the model as a measure of the rugose basalt interfaces, which could not be dealt with ray-trace inversion. Furthermore, we were able to accurately image thin low-velocity hydrocarbon-bearing Mesozoic sediments sandwiched between and hidden below two thick sequences of high-velocity basalt layers lying above the basement.

Time series sightability modeling of animal populations.

Logistic regression models—or “sightability models”—fit to detection/non-detection data from marked individuals are often used to adjust for visibility bias in later detection-only surveys, with population abundance estimated using a modified Horvitz-Thompson (mHT) estimator. More recently, a model-based alternative for analyzing combined detection/non-detection and detection-only data was developed. This approach seemed promising, since it resulted in similar estimates as the mHT when applied to data from moose (Alces alces) surveys in Minnesota. More importantly, it provided a framework for developing flexible models for analyzing multiyear detection-only survey data in combination with detection/non-detection data. During initial attempts to extend the model-based approach to multiple years of detection-only data, we found that estimates of detection probabilities and population abundance were sensitive to the amount of detection-only data included in the combined (detection/non-detection and detection-only) analysis. Subsequently, we developed a robust hierarchical modeling approach where sightability model parameters are informed only by the detection/non-detection data, and we used this approach to fit a fixed-effects model (FE model) with year-specific parameters and a temporally-smoothed model (TS model) that shares information across years via random effects and a temporal spline. The abundance estimates from the TS model were more precise, with decreased interannual variability relative to the FE model and mHT abundance estimates, illustrating the potential benefits from model-based approaches that allow information to be shared across years.

On-line separation and sorting of chicken portions using a robust vision-based intelligent modelling approach

One of the major issues in food industry is automatic sorting of chicken portions. In the present study, we propose a new on-line method based on combined machine vision techniques and linear and nonlinear classifiers to categorise chicken portions automatically. Mechanical framework, conveyor belt, electrical and control units, lighting box, charge-coupled device (CCD) camera, separating unit, and air compressor are included in the study proposed system. Major classes of chicken portions can be categorised as breast, leg, fillet, wing, and drumstick. Imaging procedure in the study is carried out using CCD camera and a computer system. Geometrical aspects, colour, and textural features are extracted in the next step using the study dataset, so the best ones could be selected accordingly through Chi-Square methodology. Partial least squares regression (PLSR), linear discriminant analysis (LDA) and artificial neural network (ANN) respectively are employed to classify the data. Considering total accuracy level of PLSR, LDA and ANN obtained in the study, results indicated better performance level of ANN compared to linear models. The machine vision algorithm developed here together with the ANN classifier were evaluated on a sorting machine to separate test samples using separating units in the on-line mode. The processing time of proposed method is estimated as 15 ms for each image. The overall accuracy in maximum speed of conveyor, 0.2 m s−1, was obtained 93 percent that is appropriate in real-time applications. The total rate of processing and sorting chicken portions was also measured as approximately 2800 samples per hour.

A Systematic Literature Review and Critical Appraisal of Economic Modelling Approaches of Pharmacologic Interventions for Systemic Lupus Erythematosus (SLE)

Background: Systemic lupus erythematosus (SLE) is a complex, autoimmune disorder with long-term health consequences. Economic evaluation is used by health technology assessment (HTA) bodies to guide drug pricing and reimbursement decisions. Objective: To conduct a systematic review to identify published economic analyses of pharmacologic treatments for SLE and summaries and critically appraise the modelling techniques used. Methods: Electronic database searches were performed from January–February 2017. Supplementary searches of conference proceedings and HTA websites were conducted. Reference lists of included papers were examined for relevant studies. Result: Fifteen relevant economic analyses were identified; all compared belimumab plus standard of care (SOC) with SOC. Most were full economic evaluations (n=13), with two considering a budget impact analysis alone. Eleven used a simulation structure to model interdependencies between multiple dimensions of SLE; one original model and ten country-level adaptations of the same model. Key assumptions were associated with duration of treatment effect for belimumab plus SOC and extrapolated treatment duration. Conclusion: Limited economic evaluations of pharmacological treatments for SLE were identified. Future economic models in SLE should meet dual aims of transparency and robustness. Key recommendations include that a lifetime model is essential for handling short and long-term complications of SLE, including organ damage and flares, and whilst the simulation model has been shown to be acceptable, for transparency and ease of interpretation the Markov model could be an acceptable and robust modelling approach.

Translational model to predict pulmonary pharmacokinetics and efficacy in man for inhaled bronchodilators.

Translational pharmacokinetic (PK) models are needed to describe and predict drug concentration‐time profiles in lung tissue at the site of action to enable animal‐to‐man translation and prediction of efficacy in humans for inhaled medicines. Current pulmonary PK models are generally descriptive rather than predictive, drug/compound specific, and fail to show successful cross‐species translation. The objective of this work was to develop a robust compartmental modeling approach that captures key features of lung and systemic PK after pulmonary administration of a set of 12 soluble drugs containing single basic, dibasic, or cationic functional groups. The model is shown to allow translation between animal species and predicts drug concentrations in human lungs that correlate with the forced expiratory volume for different classes of bronchodilators. Thus, the pulmonary modeling approach has potential to be a key component in the prediction of human PK, efficacy, and safety for future inhaled medicines.

Estimating the number of cases of podoconiosis in Ethiopia using geostatistical methods.

In 2011, the World Health Organization recognized podoconiosis as one of the neglected tropical diseases. Nonetheless, the number of people with podoconiosis and the geographical distribution of the disease is poorly understood. Based on a nationwide mapping survey and geostatistical modelling, we predict the prevalence of podoconiosis and estimate the number of cases across Ethiopia. We used nationwide data collected in Ethiopia between 2008 and 2013. Data were available for 141,238 individuals from 1,442 villages in 775 districts from all nine regional states and two city administrations. We developed a geostatistical model of podoconiosis prevalence among adults (individuals aged 15 years or above), by combining environmental factors. The number of people with podoconiosis was then estimated using a gridded map of adult population density for 2015. Podoconiosis is endemic in 345 districts in Ethiopia: 144 in Oromia, 128 in Southern Nations, Nationalities and People's [SNNP], 64 in Amhara, 4 in Benishangul Gumuz, 4 in Tigray and 1 in Somali Regional State. Nationally, our estimates suggest that 1,537,963 adults (95% confidence intervals, 290,923-4,577,031 adults) were living with podoconiosis in 2015. Three regions (SNNP, Oromia and Amhara) contributed 99% of the cases. The highest proportion of individuals with podoconiosis resided in the SNNP (39%), while 32% and 29% of people with podoconiosis resided in Oromia and Amhara Regional States, respectively. Tigray and Benishangul Gumuz Regional States bore lower burdens, and in the remaining regions, podoconiosis was almost non-existent. Discussion: The estimates of podoconiosis cases presented here based upon the combination of currently available epidemiological data and a robust modelling approach clearly show that podoconiosis is highly endemic in Ethiopia. Given the presence of low cost prevention, and morbidity management and disability prevention services, it is our collective responsibility to scale-up interventions rapidly.

Estimating the number of cases of podoconiosis in Ethiopia using geostatistical methods

Background: In 2011, the World Health Organization recognized podoconiosis as one of the neglected tropical diseases. Nonetheless, the magnitude of podoconiosis and the geographical distribution of the disease is poorly understood. Based on a nationwide mapping survey and geostatistical modelling, we predict the prevalence of podoconiosis and estimate the number of cases across Ethiopia. Methods: We used nationwide data collected in Ethiopia between 2008 and 2013. Data were available for 141,238 individuals from 1,442 communities in 775 districts from all nine regional states and two city administrations. We developed a geostatistical model of podoconiosis prevalence among adults (individuals aged 15 years or above), by combining environmental factors. The number of people with podoconiosis was then estimated using a gridded map of adult population density for 2015. Results: Podoconiosis is endemic in 345 districts in Ethiopia: 144 in Oromia, 128 in Southern Nations, Nationalities and People’s [SNNP], 64 in Amhara, 4 in Benishangul Gumuz, 4 in Tigray and 1 in Somali Regional State. Nationally, our estimates suggest that 1,537,963 adults (95% confidence intervals, 290,923-4,577,031 adults) were living with podoconiosis in 2015. Three regions (SNNP, Oromia and Amhara) contributed 99% of the cases. The highest proportion of individuals with podoconiosis resided in the SNNP (39%), while 32% and 29% of people with podoconiosis resided in Oromia and Amhara Regional States, respectively. Tigray and Benishangul Gumuz Regional States bore lower burdens, and in the remaining regions, podoconiosis was almost non-existent. Conclusions: The estimates of podoconiosis cases presented here based upon the combination of currently available epidemiological data and a robust modelling approach clearly show that podoconiosis is highly endemic in Ethiopia. Given the presence of low cost prevention, and morbidity management and disability prevention services, it is our collective responsibility to scale-up interventions rapidly.