Comparison of machine learning algorithms and multiple linear regression for live weight estimation of Akkaraman lambs

This study was designed to predict the post-weaning weights of Akkaraman lambs reared on different farms using multiple linear regression and machine learning algorithms. The effect of factors the age of the dam, gender, type of lambing, enterprise, type of flock, birth weight, and weaning weight was analyzed. The data was collected from a total of 25,316 Akkaraman lambs raised at multiple farms in the Çiftlik District of Niğde province. Comparative analysis was conducted by using multiple linear regression, Random Forest, Support Vector Machines (and Support Vector Regression), Extreme Gradient Boosting (XGBoost) (and Gradient Boosting), Bayesian Regularized Neural Network, Radial Basis Function Neural Network, Classification and Regression Trees, Exhaustive Chi-squared Automatic Interaction Detection (and Chi-squared Automatic Interaction Detection), and Multivariate Adaptive Regression Splines algorithms. In this study, the test dataset was divided into five layers using the K-fold cross-validation method. The performance of models was compared using performance criteria such as Adjusted R-squared (Adj-[Formula: see text]), Root Mean Square Error (RMSE), Mean Absolute Deviation (MAD), and Mean Absolute Percentage Error (MAPE) by utilizing test populations in the predicted models. Additionally, the presence of low standard deviations for these criteria indicates the absence of an overfitting problem. [Formula: see text]The comparison results showed the Random Forest algorithm had the best predictive performance compared to other algorithms with Adj-[Formula: see text], RMSE, MAD, and MAPE values of 0.75, 3.683, 2.876, and 10.112, respectively. In conclusion, the results obtained through Multiple Linear Regression for the live weights of Akkaraman lambs were less accurate than the results obtained through artificial neural network analysis.

Modelling bluetongue and African horse sickness vector (Culicoides spp.) distribution in the Western Cape in South Africa using random forest machine learning

BACKGROUND

Culicoides biting midges exhibit a global spatial distribution and are the main vectors of several viruses of veterinary importance, including bluetongue (BT) and African horse sickness (AHS). Many environmental and anthropological factors contribute to their ability to live in a variety of habitats, which have the potential to change over the years as the climate changes. Therefore, as new habitats emerge, the risk for new introductions of these diseases of interest to occur increases. The aim of this study was to model distributions for two primary vectors for BT and AHS (Culicoides imicola and Culicoides bolitinos) using random forest (RF) machine learning and explore the relative importance of environmental and anthropological factors in a region of South Africa with frequent AHS and BT outbreaks.

METHODS

Culicoides capture data were collected between 1996 and 2022 across 171 different capture locations in the Western Cape. Predictor variables included climate-related variables (temperature, precipitation, humidity), environment-related variables (normalised difference vegetation index-NDVI, soil moisture) and farm-related variables (livestock densities). Random forest (RF) models were developed to explore the spatial distributions of C. imicola, C. bolitinos and a merged species map, where both competent vectors were combined. The maps were then compared to interpolation maps using the same capture data as well as historical locations of BT and AHS outbreaks.

RESULTS

Overall, the RF models performed well with 75.02%, 61.6% and 74.01% variance explained for C. imicola, C. bolitinos and merged species models respectively. Cattle density was the most important predictor for C. imicola and water vapour pressure the most important for C. bolitinos. Compared to interpolation maps, the RF models had higher predictive power throughout most of the year when species were modelled individually; however, when merged, the interpolation maps performed better in all seasons except winter. Finally, midge densities did not show any conclusive correlation with BT or AHS outbreaks.

CONCLUSION

This study yielded novel insight into the spatial abundance and drivers of abundance of competent vectors of BT and AHS. It also provided valuable data to inform mathematical models exploring disease outbreaks so that Culicoides-transmitted diseases in South Africa can be further analysed.

Modelling African horse sickness emergence and transmission in the South African control area using a deterministic metapopulation approach

African horse sickness is an equine orbivirus transmitted by Culicoides Latreille biting midges. In the last 80 years, it has caused several devastating outbreaks in the equine population in Europe, the Far and Middle East, North Africa, South-East Asia, and sub-Saharan Africa. The disease is endemic in South Africa; however, a unique control area has been set up in the Western Cape where increased surveillance and control measures have been put in place. A deterministic metapopulation model was developed to explore if an outbreak might occur, and how it might develop, if a latently infected horse was to be imported into the control area, by varying the geographical location and months of import. To do this, a previously published ordinary differential equation model was developed with a metapopulation approach and included a vaccinated horse population. Outbreak length, time to peak infection, number of infected horses at the peak, number of horses overall affected (recovered or dead), re-emergence, and Rv (the basic reproduction number in the presence of vaccination) were recorded and displayed using GIS mapping. The model predictions were compared to previous outbreak data to ensure validity. The warmer months (November to March) had longer outbreaks than the colder months (May to September), took more time to reach the peak, and had a greater total outbreak size with more horses infected at the peak. Rv appeared to be a poor predictor of outbreak dynamics for this simulation. A sensitivity analysis indicated that control measures such as vaccination and vector control are potentially effective to manage the spread of an outbreak, and shortening the vaccination window to July to September may reduce the risk of vaccine-associated outbreaks.

Progressive trends on the application of artificial neural networks in animal sciences - A review

In recent years, artificial neural networks have become the subject of intensive research in a number of scientific areas. The high performance and operational speed of neural models open up a wide spectrum of applications in various areas of life sciences. Objectives pursued by many scientists, who use neural modelling in their research, focus - among others - on intensifying real-time calculations. This study shows the possibility of using Multilayer-Perceptron (MLP) and Radial Basis Function (RBF) models of artificial neural networks for the future development of new methods for animal science. The process should be explained explicitly to make the MLP and RBF models more readily accepted by more researchers. This study describes and recommends certain models as well as uses forecasting methods, which are represented by the chosen neural network topologies, in particular MLP and RBF models for more successful operations in the field of animals sciences.

A model for the assessment of bluetongue virus serotype 1 persistence in Spain

Bluetongue virus (BTV) is an arbovirus of ruminants that has been circulating in Europe continuously for more than two decades and has become endemic in some countries such as Spain. Spain is ideal for BTV epidemiological studies since BTV outbreaks from different sources and serotypes have occurred continuously there since 2000; BTV-1 has been reported there from 2007 to 2017. Here we develop a model for BTV-1 endemic scenario to estimate the risk of an area becoming endemic, as well as to identify the most influential factors for BTV-1 persistence. We created abundance maps at 1-km2 spatial resolution for the main vectors in Spain, Culicoides imicola and Obsoletus and Pulicaris complexes, by combining environmental satellite data with occurrence models and a random forest machine learning algorithm. The endemic model included vector abundance and host-related variables (farm density). The three most relevant variables in the endemic model were the abundance of C. imicola and Obsoletus complex and density of goat farms (AUC 0.86); this model suggests that BTV-1 is more likely to become endemic in central and southwestern regions of Spain. It only requires host- and vector-related variables to identify areas at greater risk of becoming endemic for bluetongue. Our results highlight the importance of suitable Culicoides spp. prediction maps for bluetongue epidemiological studies and decision-making about control and eradication measures.

Spatial distribution modelling of Culicoides (Diptera: Ceratopogonidae) biting midges, potential vectors of African horse sickness and bluetongue viruses in Senegal

Background

In Senegal, the last epidemic of African horse sickness (AHS) occurred in 2007. The western part of the country (the Niayes area) concentrates modern farms with exotic horses of high value and was highly affected during the 2007 outbreak that has started in the area. Several studies were initiated in the Niayes area in order to better characterize Culicoides diversity, ecology and the impact of environmental and climatic data on dynamics of proven and suspected vectors. The aims of this study are to better understand the spatial distribution and diversity of Culicoides in Senegal and to map their abundance throughout the country.

Methods

Culicoides data were obtained through a nationwide trapping campaign organized in 2012. Two successive collection nights were carried out in 96 sites in 12 (of 14) regions of Senegal at the end of the rainy season (between September and October) using OVI (Onderstepoort Veterinary Institute) light traps. Three different modeling approaches were compared: the first consists in a spatial interpolation by ordinary kriging of Culicoides abundance data. The two others consist in analyzing the relation between Culicoides abundance and environmental and climatic data to model abundance and investigate the environmental suitability; and were carried out by implementing generalized linear models and random forest models.

Results

A total of 1,373,929 specimens of the genus Culicoides belonging to at least 32 different species were collected in 96 sites during the survey. According to the RF (random forest) models which provided better estimates of abundances than Generalized Linear Models (GLM) models, environmental and climatic variables that influence species abundance were identified. Culicoides imicola, C. enderleini and C. miombo were mostly driven by average rainfall and minimum and maximum normalized difference vegetation index. Abundance of C. oxystoma was mostly determined by average rainfall and day temperature. Culicoides bolitinos had a particular trend; the environmental and climatic variables above had a lesser impact on its abundance. RF model prediction maps for the first four species showed high abundance in southern Senegal and in the groundnut basin area, whereas C. bolitinos was present in southern Senegal, but in much lower abundance.

Conclusions

Environmental and climatic variables of importance that influence the spatial distribution of species abundance were identified. It is now crucial to evaluate the vector competence of major species and then combine the vector densities with densities of horses to quantify the risk of transmission of AHS virus across the country.

Electronic supplementary material

The online version of this article (10.1186/s13071-018-2920-7) contains supplementary material, which is available to authorized users.

Comparing the effect of modeled climatic variables on the distribution of African horse sickness in South Africa and Namibia

Africa horse sickness (AHS) is a lethal disease of horses with a seasonal occurrence that is influenced by environmental conditions that favor the development of Culicoides midges (Diptera: Ceratopogonidae). This study compared and evaluated the relationship of various modeled climatic variables with the distribution and abundance of AHS in South Africa and Namibia. A comprehensive literature review of the historical AHS reported data collected from the Windhoek archives as well as annual reports from the Directorate of Veterinary services in Namibia were conducted. South African AHS reported data were collected from the South African Department of Agriculture, Forestry, and Fisheries. Daily climatic data were extracted for the time period 1993-2011 from the ERA-interim re-analysis dataset. The principal component analysis of the complete dataset indicated a significant statistical difference between Namibia and South Africa for the various climate variables and the outbreaks of AHS. The most influential parameters in the distribution of AHS included humidity, precipitation, evaporation, and minimum temperature. In South Africa, temperature had the most significant effect on the outbreaks of AHS, whereas in Namibia, humidity and precipitation were the main drivers. The maximum AHS cases in South Africa occurred at temperatures of 20-22° C and relative humidity between 50-70%. Furthermore, anthropogenic effects must be taken into account when trying to understand the distribution of AHS.

Identification of Suitable Areas for African Horse Sickness Virus Infections in Spanish Equine Populations

African horse sickness (AHS) is one of the most important vector-borne viral infectious diseases of equines, transmitted mainly by Culicoides spp. The re-emergence of Culicoides-borne diseases in Europe, such as the recent bluetongue (BT) or Schmallenberg outbreaks, has raised concern about the potential re-introduction and further spread of AHS virus (AHSV) in Europe. Spain has one of the largest European equine populations. In addition, its geographical, environmental and entomological conditions favour AHSV infections, as shown by the historical outbreaks in the 1990s. The establishment of risk-based surveillance strategies would allow the early detection and rapid control of any potential AHSV outbreak. This study aimed to identify the areas and time periods that are suitable or at high risk for AHS occurrence in Spain using a GIS-based multicriteria decision framework. Specifically risk maps for AHS occurrence were produced using a weighted linear combination of the main risk factors of disease, namely extrinsic incubation period, equine density and distribution of competent Culicoides populations. Model results revealed that the south-western and north-central areas of Spain and the Balearic Islands are the areas at the highest risk for AHSV infections, particularly in late summer months. Conversely, Galicia, Castile and Leon and La Rioja can be considered as low-risk regions. This result was validated with historical AHS and BT outbreaks in Spain, and with the Culicoides vector distribution area. The model results, together with current Spanish equine production features, should provide the foundations to design risk-based and more cost-effective surveillance strategies for the early detection and rapid control potential of AHS outbreaks in Spain.