dc.contributor.author | Murungi, Robert M | |
dc.date.accessioned | 2013-02-26T09:37:33Z | |
dc.date.issued | 2011 | |
dc.identifier.citation | Masters of science in Chemistry | en |
dc.identifier.uri | http://erepository.uonbi.ac.ke:8080/xmlui/handle/123456789/11599 | |
dc.description.abstract | The goal of this project was to develop a functional livestock diagnostic prototype to be used by pastoralist
communities in arid and semi arid lands of Kenya. The prototype will be used by non-experts in the
veterinary domain to address pastoralist livestock health needs. The prototype has been implemented using
artificial neural network that learns on existing and emerging livestock diseases. Learning involves training
the network via a graphical user interface provided by the system. Our neural network uses supervised
learning, where a set of input and output values (i.e. symptoms and diseases) are supplied to the network.
Once the network has been trained, the system prompts the user with probable hypothesis regarding the
livestock disease symptoms. The user in turn confirms or denies the hypothesis through a web based
graphical interface. Finally the system diagnoses the disease that the livestock is suffering from and
provides appropriate medical recommendations.
For the system to associate a certain disease with specific symptoms, the symptoms were set to meet a
threshold of at least 59%. On average, the threshold level ranging above or equal to 59% presented to the
system during the testing process was 79.59%. The system associated the symptoms with their respective
diseases with 69.64% accuracy. On the other hand, the threshold level ranging below 59% presented to the
system during the testing process was 38.75% on average. In each case the system could not associate the
symptoms with any disease. During evaluation process, majority of the pastoralists and the veterinarians
were satisfied that the system was operating according to their expectations. | en |
dc.description.sponsorship | University of Nairobi | en |
dc.language.iso | en | en |
dc.publisher | University of Nairobi | en |
dc.subject | Livestock | en |
dc.subject | diagnostic support system | en |
dc.subject | rural pastoralist communities | en |
dc.subject | Kenya | en |
dc.subject | prototype | en |
dc.title | Livestock diagnostic support system for rural pastoralist communities in Kenya: a prototype | en |
dc.type | Thesis | en |
local.publisher | Chemistry | en |