Machine learning models for predicting the use of different animal breeding services in smallholder dairy farms in Sub-Saharan Africa.

This study is concerned with developing predictive models using machine learning techniques to be used in identifying factors that influence farmers' decisions, predict farmers' decisions, and forecast farmers' demands relating to breeding service. The data used to develop the models comes from a survey of small-scale dairy farmers from Tanzania (n = 3500 farmers), Kenya (n = 6190 farmers), Ethiopia (n = 4920 farmers), and Uganda (n = 5390 farmers) and more than 120 variables were identified to influence breeding decisions. Feature engineering process was used to reduce the number of variables to a practical level and to identify the most influential ones. Three algorithms were used for feature selection, namely: logistic regression, random forest, and Boruta. Subsequently, six predictive models, using features selected by feature selection method, were tested for each country-neural network, logistic regression, K-nearest neighbor, decision tree, random forest, and Gaussian mixture model. A combination of decision tree and random forest algorithms was used to develop the final models. Each country model showed high predictive power (up to 93%) and are ready for practical use. The use of ML techniques assisted in identifying the key factors that influence the adoption of breeding method that can be taken and prioritized to improve the dairy sector among countries. Moreover, it provided various alternatives for policymakers to compare the consequences of different courses of action which can assist in determining which alternative at any particular choice point had a high probability to succeed, given the information and alternatives pertinent to the breeding decision. Also, through the use of ML, results to the identification of different clusters of farmers, who were classified based on their farm, and farmers' characteristics, i.e., farm location, feeding system, animal husbandry practices, etc. This information had significant value to decision-makers in finding the appropriate intervention for a particular cluster of farmers. In the future, such predictive models will assist decision-makers in planning and managing resources by allocating breeding services and capabilities where they would be most in demand.

Genetic evaluation of test-day milk yields from smallholder dairy production systems in Kenya using genomic relationships.

Efforts to improve dairy production in smallholder farming systems of East Africa over the past decade have had limited impact because of the lack of records on performance to guide targeted breeding programs. Estimates of genetic parameters in these systems are lacking. Using data generated through a project ("Germplasm for Dairy Development in East Africa") in Kenya and a genomic relationship matrix from genotypic records, we examined the potential impact of different models handling contemporary groups or herd effects on estimates of genetic parameters using a fixed regression model (FRM) for test-day (TD) milk yields, and the covariance structure for TD milk yield at various stages of lactation for animals using a random regression model (RRM). Models in which herd groups were defined using production levels derived from the data fitted the data better than those in which herds were grouped depending on management practices or were random. Lactation curves obtained for animals under different production categories did not display the typical peak yield characteristic of improved dairy systems in developed countries. Heritability estimates for TD milk yields using the FRM varied greatly with the definition of contemporary herd groups, ranging from 0.05 ± 0.03 to 0.27 ± 0.05 (mean ± standard error). The analysis using the RRM fitted the data better than the FRM. The heritability estimates for specific TD yields obtained by the RRM were higher than those obtained by the FRM. Genetic correlations between TD yields were high and positive for measures within short consecutive intervals but decreased as the intervals between TD increased beyond 60 d and became negative with intervals of more than 5 mo. The magnitude of the genetic correlation estimates among TD records indicates that using TD milk records beyond a 60-d interval as repeated measures of the same trait for genetic evaluation of animals on smallholder farms would not be optimal. Although each individual smallholder farmer retains only a few animals, using the genomic relationship between animals to link the large number of farmers operating under specified environments provides a sufficiently large herd-group for which a breeding program could be developed.

Double-blind, vehicle-controlled randomized twelve-month neurodevelopmental toxicity study of common aluminum salts in the rat.

This good laboratory practice (GLP) study of aluminum salts in Sprague-Dawley rats was conducted according to double-blind, vehicle-controlled randomized design by exposing offspring to aluminum citrate in-utero, through lactation, and then in drinking water post-weaning. Three dose levels were used: 30, 100, 300 mg Al/kg bw/day, in addition to control groups that received either water or a sodium citrate solution (27.2 g/L). Endpoints were assessed in both female and male pups: behavioral (motor activity, T-maze, auditory startle, the Functional Observational Battery (FOB) with domains targeting autonomic function, activity, neuromuscular function, sensimotor function, and physiological function), cognitive function (Morris swim maze), brain weight, clinical chemistry, hematology, tissue/blood levels of aluminum and neuropathology. The most notable treatment-related effect observed in the offspring was renal pathology, most prominently in the male pups. Higher mortality and significant morbidity were observed in the male pups in the high Al-citrate dose group; leading to euthanization of this group at day 89. There was evidence for dose-response relationships between neuromuscular measurements-hind-limb and fore-limb grip strength-and Al-treatment in both males and females, although some of the effects may be secondary to body weight changes. No consistent treatment-related effects were observed in ambulatory counts (motor activity) in the different cohorts. No significant effects were observed for the auditory startle response, T-maze tests (pre-weaning day 23 cohort) or the Morris water maze test (day 120 cohort). None of the lesions seen on histopathological examination of brain tissues of the day 364 group was reported as treatment-related and, as these were also seen in the control group, were likely due to aging. In conclusion, these results indicate that concentrations of aluminum in the drinking water that are required to produce minimally detectable neurobiological effects in the rat are about 10,000 times higher than what is typically found in potable drinking water.