School of Computing and Informatics
http://ir-library.mmust.ac.ke:8080/xmlui/handle/123456789/43
2024-03-29T11:33:42ZAN INTEGRATED MODEL FOR E-HEALTH IMPLEMENTATION IN KENYA
http://ir-library.mmust.ac.ke:8080/xmlui/handle/123456789/2696
AN INTEGRATED MODEL FOR E-HEALTH IMPLEMENTATION IN KENYA
Shirandula, Ayub Hussein
numerous challenges which have forced developing countries to focus on digital
interventions known as e-health Technologies. Kenya, as a developing country, is not left
behind regarding the challenges of e-health. With the vision of efficient implementation and
use of e-health, there is need to develop a model that will enable the improvement of
effective e-health. The purpose of the study was to investigate the factors influencing
adoption and implementation of e-health in Kenya. The objectives of the study were to
determine the current status of e-health implementation in Kenya, determining critical
factors that affects the implementation of e-health Technologies in health sector in Kenya
and to develop a model for e-health implementation in Kenya. The Study used Normalization
Process Theory (NPT), Actor Network Theory (ANT) and Technology Organization and
Environmental Framework (TOE) to underpin the study. Data was collected using structured
questionnaire and interview of key informants. The quantitative data was then coded and
analyzed using both descriptive and inferential statistics and the qualitative data was
analyzed using thematic analysis. Pragmatist research philosophy was adopted given the
multiple realities and the fact that it fitted well with deductive and inductive approaches. The
study population included 1243 healthcare workers from which a sample of 303 respondents
were obtained using stratified sampling. The findings of the research indicated that social
factors were significant predictor of e-health implementation where (p=0.005<0.05),
organizational factors were significant predictor of e-health implementation where
(p=0.002<0.05), technological factors were significant predictor of e-health implementation
where (p=0.000<0.05) and environmental factors were significant predictor of e-health
implementation where (p=0.048<0.05). From the outcomes of this study, a model of e-health
implementation was realized to guide the process of effective e-health development and
used. The study concluded that there was need for various stakeholders to reflect on
organizational, social and environmental relationship and interaction with technical aspect
as technological factor was the major factor that affect e-health implementation in Kenya.
The model developed is a basis upon which future implementation of e-health can be based.
It was recommended that for effective implementation, there is need for a well-defined
implementation plan for e-health. The study further recommends for research on e-health
adoption.
2023-09-01T00:00:00ZALOCAL DIRECTIONAL TERNARY PATTERN TEXTURE DESCRIPTOR FOR MAMMOGRAPHIC BREAST CANCER CLASSIFICATION
http://ir-library.mmust.ac.ke:8080/xmlui/handle/123456789/1418
ALOCAL DIRECTIONAL TERNARY PATTERN TEXTURE DESCRIPTOR FOR MAMMOGRAPHIC BREAST CANCER CLASSIFICATION
MWADULO, MARY WALOWE
Breast cancer is a top killer illness for women globally, but early and effective screening can increase their survival rate. Mammography is the tool used by a radiologist to screen for breast cancer, however,a radiologist is susceptible to human observer variability, and therefore, reading and interpretation of mammography test results depend on the expertise of the radiologist administeringthe test. To improvethe reading and interpretation accuracy of the test, researchers’ developed computer-aided extraction descriptors thatextract discriminant features.These descriptors include the Local Binary Patterns (LBP), the Local Ternary Patterns (LTP), and the Local Directional Patterns (LDP), however,they have not yet yielded satisfactory results in differentiating breast cancer tumor types. The LBP descriptor is inadequately dependable in capturing breast cancer discriminant features because it is easily affected by noise. TheLTP descriptor uses a fixed threshold value for all images in a dataset, making it not invariant to pixel value transformation. It is also not practically easy to select an optimum threshold value in real applicationdomains. The LDPdescriptor relies on top k significant directional responses and ignores the remaining 8-kdirectional responses. Disregarding the remaining directional responses reduces the computation efficiency since each pixel in an image carries subtle information. Given the limitations identified among the mentioned local texture descriptors, developing an effective texture descriptor becomes a viable and challenging research problem. Therefore, this study seeks to develop an improved local texture descriptor that considers all directional responses and applies an adaptive threshold in encoding image gradient. The new Local Directional Ternary Pattern (LDTP) texture descriptor calculates the absolute difference between the value of the center pixel and the values of its local neighboring pixels for a 3x3 image region. To get edge responses in eight directions, the absolute differences are convolved with a kirsch mask, then the pixels are transformed into zeros and ones using mini-max normalization. We then passed the normalized values through a soft-max function to get the probability of an edge ina certain direction. Then, two threshold values are calculated and used to split the probability space into three parts for -1, 0, +1 bits to generate a ternary pattern. The resultant Local Directional Ternary Pattern (LDTP) code is then split into a positive and negative LDTP code. Histograms of negative and positive LDTP encoded images are fused to get texture features. We validated the LDTP texture descriptor on the Mammographic Image Analysis Society (MIAS) breast cancer dataset using Support Vector Machine (SVM) and Artificial Neural Network (ANN) classifiers for normal/abnormal and benign/malignant classes. When the LDTP texture descriptor was compared against LDP, LTP, and other existing texture descriptors, it showed robustness and reliability in encoding an image gradient. The highest classification accuracy was attained by the SVM classifier, with 97.32% and 93.93% for normal/abnormal and benign/malignant classes, respectively.
2020-10-11T00:00:00ZMACHINE LEARNING MODEL FOR PREDICTION OF STUDENTS’ ACADEMIC PERFORMANCE, KENYA
http://ir-library.mmust.ac.ke:8080/xmlui/handle/123456789/1413
MACHINE LEARNING MODEL FOR PREDICTION OF STUDENTS’ ACADEMIC PERFORMANCE, KENYA
Musau, Obadiah Matolo
Prediction of students’ academic performance with high accuracy is useful in many ways in academic institutions. Institutions would like to know which students are likely to have low academic achievements or need assistance in order to finish their studies. Successful students’ academic performance prediction at an early stage in learning depends on many factors. Machine learning techniques can be utilized to predict students’ future academic performance. The primary objective of this study was to develop a machine learning model for prediction of students’ academic performance. To achieve this objective, the study was guided by the following theoretical and empirical objectives: 1. To analyse existing studies on students’ academic performance prediction, 2. To find out the most significant factors that affect students’ academic performance, 3. To develop a model for students’ academic performance prediction in Kenya and, 4. To validate the students’ academic performance prediction model. Student data was collected from 1720 former secondary school students currently enrolled in tertiary institutions using questionnaires. The data included students’ academic performance, demographic features, social features and school related features. Naïve Bayes, Decision Trees and Neural Networks were used to predict students’ final examination grade. The performance of the prediction models was validated using 10-fold cross-validation method. J48 Decision Tree prediction model achieved 85.9 % prediction accuracy, Naïve Bayes prediction model achieved 78.96% prediction accuracy and Neural Networks Multi Perceptron prediction model achieved the lowest prediction accuracy of 73.73%. This work will help educational institutions, school managements, government ministries, parents, donors and other education stakeholders to predict students’ performance and identify nonperforming student that need assistance to finish their studies.
2020-10-05T00:00:00ZA SIZE METRIC-BASED EFFORT ESTIMATION METHOD FOR SERVICE ORIENTED ARCHITECTURE SYSTEMS
http://ir-library.mmust.ac.ke:8080/xmlui/handle/123456789/1393
A SIZE METRIC-BASED EFFORT ESTIMATION METHOD FOR SERVICE ORIENTED ARCHITECTURE SYSTEMS
SAMSON WANJALA MUNIALO, SAMSON WANJALA
Service Oriented Architecture (SOA) is one of the recent software development paradigms that enablealignment of business processes into integrated services within and outside organizationsregardless of the heterogeneity of technologies used. Determining the scope, effort and cost of SOA systems is important to facilitate theplanning and eventuallysuccessful implementation of software projects. A number of methodshave been proposedto estimate effort of building SOA projects.Despite the fact that these methods are promising, the problem of measuring SOA size and estimating SOA effort still remains largely unresolved mainly because there is limited attemptin using Unified Modeling Language (UML)size metrics todefine size-based attributes for estimatingSOA development effort. To address this problem, a set of size metrics were definedand effort estimation method that is based on the size metrics was developed.To automate the computation of the metric and the method, a static analysis tool that usesdeep learning techniques to detect UML arrows and recognize text was constructed. The automated tooldeep learning techniqueswere eachsubjected to validity checks based on datasetsof 100operation names and 100 arrow head images.Briand’s theoretical validationwas usedto test the validity of the designed size metricsand they were found to be mathematically sound. Experimental research design wasemployedto sampled SOA systems to test variables used in the study and the accuracy of the proposed effort estimation method and implementation automated tool. A survey involving experts from the industry was carried out to replicate and validatethe experimentdone by studentsand to determinethe appropriateness of the proposed size metrics, SOA development effort factors and the implementation automated tool. The experiment was based on a sample of 15students’ SOA projectsdeveloped by Meru University of Science and Technology studentswhilethe survey involved 20 programmers from the industry. Descriptivestatistics such as Mean magnitude of relative error (MMRE)and Magnitude of Error (MRE) were used to test SOA effort estimation accuracywhile linear regression analysistested relationshipamong variables identified in the study. Result from the experiment revealed that the proposed metrics and method are more accurate and there is a correlation between size attributes and SOA sizeand between SOA size and SOA development effort. Response from the survey showed that the proposed metrics and effort factors are valid and they have influence onsize and effort respectively. Findings from this study were meant toprovide a basis for future software engineering researchers to develop more effective and more accurate size metrics and effort estimation methods.
2020-09-25T00:00:00Z