6229 research outputs found
Sort by
A Transfer Learning and Two-Level Hyperparameter Optimization Based Model for Improved Classification of Diabetic Retinopathy
No full textMaster of Science in Information Technology, 2022.Automated diagnosis of disease from medical images using machine learning has been in rise in the recent past. One such case is the classification of diabetic retinopathy from fundus images. Diabetic Retinopathy is an eye disease that is a result of diabetes mellitus and it is major cause of blindness among people of the working age. Diabetic retinopathy has five main classes namely: No DR, Mild DR, Moderate DR, Severe DR, and Proliferative DR. Deep learning has been used previously in this field and it has proved to be better than conventional machine learning approaches. However, deep learning involves training a model from scratch thus making it to be data hungry, require high training cost, have poor generalizability, and they don’t deliver high performance. Meta-learning also known as learning-to-learn is a field of machine learning which aims at improving deep learning by enabling models to improve their performance capabilities and reduce training cost. Meta-learning techniques include multi-task learning, transfer learning, self-optimization, and few-shot learning. Several transfer learning architectures pre-trained on the ImageNet dataset have been used by different researchers and they have demonstrated superior performance over deep learning. However, domain-shift generalizability and optimal performance of pre-trained architectures are major challenges facing transfer learning. This so because these models are not properly tuned for cross-domain optimality. The aim of this study was to develop an improved model for classification of diabetic retinopathy into its five classes. To achieve this, the researcher used the following approach: A VGG16 network pre-trained in ImageNet was modified such that the top-layer was rebuilt and an attention model was added. Two-level optimization was used during training in which the model was allowed to self-tune its learning rate based on the training parameters. The EyePACS dataset obtained from Kaggle repository was used in training, validating, and testing the model. The model was developed in Google Collaboratory platform using python programming language, TensorFlow, and Keras. The study achieved the following results: Accuracy 89.06%, Precision 88.9%, Recall 89.2%, F1-Score 75%, Quadratic Cohen Kappa Metric 0.84, Area Under the Curve (AUC) 93.3%. The results of the study demonstrated improved performance compared to other existing models in literature such as Qummar et al (2019), Jinfeng et al (2020), Chilukoti et al (2022), that classify diabetic retinopathy into five classes. The study concluded that leveraging on previously acquired knowledge and efficient optimization of neural networks using data driven self-optimization delivers better performance than conventional machine learning and deep learning. In future researchers can consider using reinforcement learning and transfer learning in classification of diabetic retinopathy.Murang'a University of Technolog
Spatial Modelling of Malaria Prevalence in Kenya
No full textMaster of Science in Statistics, 2022Malaria is a leading cause of deaths in Kenya. A vector-borne disease caused by parasite of genus plasmodium; the disease is introduced into the human circulatory system from bites caused by infected female anopheles’ mosquitoes. A lot of effort and resources has been put in the fight against malaria, with large amount of national budget being used in the fight against malaria in developing countries which has led to underdevelopment, impoverished livelihoods and low human development index. Malaria burden affects the world’s poorest countries. About 90% of the malaria burden is reported in sub-Saharan Africa. Malaria cases are significantly high in countries of south-East Asia, Western Pacific region, Mediterranean and the Americas. As of 2017, five countries India, Uganda, DR Congo and Mozambique accounted for half of malaria cases reported around the world. In Kenya, the disease has led to impoverished livelihoods with the poorest communities of the country being the most affected. The disease has led to high mortality cases in children under five years and pregnant women. Loss of man hours and work days among adults in the country, leading to low productivity. Studies have shown that there has been a general lack of knowledge on how select demographic and social economic conditions risk factors affect the prevalence of malaria in Kenya. The method of the study involved performing the spatial models for malaria prevalence in Kenya while relaxing the assumptions of stationarity. The assumptions of linearity allowed some covariates like age to have a non-linear effect on prevalence of malaria. Using random walk model of 2nd order and the assumption of stationarity, it allowed covariates to vary spatially. Conditional autoregressive model was used. Data from malaria indicator survey of 2015 (KMIS2015) was used for the study. Both the social-economic and demographic variables were used as predictor variables. These included education level, wealth index, age, access to mosquito nets and place of residence. From the study, demographic and social-economic factors were found to have significant impact on Prevalence of malaria in Kenya. Most cases of malaria were reported in lake, western and coastal regions. The most prone areas were Kisumu, Homabay, Kakamega and Mombasa. There were less cases in central Kenya counties like Nyeri, Tharaka-Nithi with a significant number reported in arid and semi-arid regions of Northern-Kenya counties of Garissa, Mandera, Baringo. Rural population was more susceptible to malaria compared to those in urban areas. The odds of getting (verse not getting malaria) in urban places of residence increases by 0.84, which is estimated to .096, CIs 95% (0.70, 1.01), and a p-value .069. Malaria prevalence varied significantly from one region to another. The study established that Spatial autocorrelation exists among regions mostly due to weather patterns, geography, cultural practices and socio-economic factors.Murang'a University of Technolog