https://github.com/cwwhitney/kenya_fruit_trees

Decision models for fruit tree interventions in Kenya - built in AgenaRisk

https://github.com/cwwhitney/kenya_fruit_trees

Repository

Decision models for fruit tree interventions in Kenya - built in AgenaRisk

README.Rmd

---
output: github_document
---

Supplementary: Integration of fruit trees for better nutrition – Case study in Kenya

Cory Whitney and Eike Luedeling
University of Bonn

## Bayesian Network models 

The Bayesian Network models are given below. The green nodes within the ‘Production’ sub-module illustrate the computation used to arrive at the volume, quality and nutrient content of the available fruit. The purple nodes within the ‘Household’ sub-module represent household composition. Together these illustrate the computation used to arrive at the household fruit consumption potential, labor availability, and nutrient needs per year and per farm. The values generated for nutrient intakes and demands are generated based on the types of individuals represented in the ‘Household Composition (nutrient demand groups)’ node. The node states are ‘Healthy pregnant woman’, ‘Healthy lactating woman’, ‘Healthy not reproducing woman’, ‘Unhealthy pregnant woman’, ‘Unhealthy lactating woman’, ‘Unhealthy not reproducing woman’, ‘Healthy man’, ‘Unhealthy man’, ‘Healthy adolescent’, ‘Unhealthy adolescent’, ‘Healthy child’, ‘Unhealthy child’, ‘Healthy infant’, ‘Unhealthy infant’, ‘Healthy elderly’, and ‘Unhealthy elderly. Specific information about the needs of these is shown in the Supplementary (Luedeling and Whitney 2018). 

In the following models the grey outline circles represent sub-modules of related nodes used to generate probabilities of the respective model parts. Green arrows indicate positive effects and red arrows indicate negative effects. Model built in the AgenaRisk software (Fenton and Neil 2017).

Under the proposed policy, both papaya and mango trees are expected to be introduced into and managed within smallholder household farms and gardens. Papaya is typically managed as a sparsely-branched tree, typically quite tall. It can take from six to twelve months from seed to fruit. A serving of papaya of 200 grams contains roughly 4% of the daily recommended calories, 12% of vitamin A, and 1% of iron and zinc. Mango is typically managed as a large tree. It can take five to eight years from seed to fruit, nursery saplings can take about four years. A serving of mango of 200 grams contains roughly 5% of the daily recommended calories and vitamin A, 1% of iron, and zinc. We worked together with stakeholders to gather knowledge about the possible costs and risks of the introduction of these trees to farming systems and the expected benefits to the nutrition of farming households. 

### Energy

![Bayesian Network model of the household nutritional energy impacts of planting fruit trees on farms in Kenya. The model calculates probabilities for the expected impact on the annual energy intake gap among rural farming households given the intervention to provide planting materials for fruit trees.](Figures/AgenaRisk_Energy.png) 

For the results of the Value of information analysis see the full list of AgenaRisk generated [EV(P)PI and EV|PI values](https://htmlpreview.github.io/?raw.githubusercontent.com/CWWhitney/kenya_fruit_trees/main/data/EVPI_Minimized/EVPI_Energy.html). We are mainly interested in the EVPI (last collection of values in the display). For more information, and for running locally, see [the full model in AgenaRisk](./Models/BN_Model_Energy_170613.cmp). This includes all respective tables of probability.

### Iron

![Bayesian Network model of the household nutritional impacts in terms of dietary iron of planting fruit trees on farms in Kenya. The model calculates probabilities for the expected impact on the annual iron intake gap among rural farming households given the intervention to provide planting materials for fruit trees.](Figures/AgenaRisk_Iron.png) 

For the results of the Value of information analysis see the full list of AgenaRisk generated [EV(P)PI and EV|PI values](https://htmlpreview.github.io/?raw.githubusercontent.com/CWWhitney/kenya_fruit_trees/main/data/EVPI_Minimized/EVPI_Iron.html). We are mainly interested in the EVPI (last collection of values in the display). For more information, and for running locally, see [the full model in AgenaRisk](./Models/BN_Model_Iron_170613.cmp). This includes all respective tables of probability.

### Vitamin A

![Bayesian Network model of the household nutritional Vitamin A impacts of planting fruit trees on farms in Kenya. The model calculates probabilities for the expected change in the annual gap of Vitamin A intake  among rural farming households given the intervention to provide planting materials for fruit trees.](Figures/AgenaRisk_VitA.png) 

For more information, and for running locally, see [the full model in AgenaRisk](./Models/BN_Model_Vit_A_170613.cmp). This includes all respective tables of probability. For the results of the Value of information analysis see the full list of AgenaRisk generated [EV(P)PI and EV|PI values](https://htmlpreview.github.io/?raw.githubusercontent.com/CWWhitney/kenya_fruit_trees/main/data/EVPI_Minimized/EVPI_Vit_A.html). We are mainly interested in the EVPI (last collection of values in the display). 

### Zinc

![Bayesian Network model of the household nutritional Zinc impacts of planting fruit trees on farms in Kenya. The model calculates probabilities for the expected change in the annual gap of Zinc intake  among rural farming households given the intervention to provide planting materials for fruit trees.](Figures/AgenaRisk_Zinc.png) 

For more information on household nutritional Zinc impacts, and for running locally, see [the full model in AgenaRisk](./Models/BN_Model_Zinc_170415.cmp). This includes all respective tables of probability. For the results of the Value of information analysis see the full list of AgenaRisk generated [EV(P)PI and EV|PI values](https://htmlpreview.github.io/?raw.githubusercontent.com/CWWhitney/kenya_fruit_trees/main/data/EVPI_Minimized/EVPI_Zinc.html). We are mainly interested in the EVPI (last collection of values in the display). 

# References

Fenton, N., and M. Neil. “AgenaRisk Professional Version 7.0.” Revision 3451 VOI, 2017.

Luedeling, E., and Cory Whitney. “Probabilistic Causal Models for Nutrition Outcomes of Agricultural Actions - Kenya Model.” Probabilistic Causal Models for Nutrition Outcomes of Agricultural Actions, 2018.

Owner

GitHub Events

Issues and Pull Requests

Last synced: about 1 year ago