mirugwe alex | Makerere University School of Public Health (original) (raw)

Papers by mirugwe alex

The Ebola virus disease (EVD) is an extremely contagious and fatal illness caused by the Ebola vi... more The Ebola virus disease (EVD) is an extremely contagious and fatal illness caused by the Ebola virus. Recently, Uganda witnessed an outbreak of EVD, which generated a lot of attention on various social media platforms. To ensure effective communication and implementation of targeted health interventions, it is crucial for stakeholders to comprehend the sentiments expressed in the posts and discussions on these online platforms. In this study, we used deep learning techniques to analyze the sentiments expressed in Ebola-related tweets during the outbreak. We explored the application of three deep learning techniques to classify the sentiments in 8,395 tweets as positive, neutral, or negative. The techniques examined included a 6-layer convolutional neural network (CNN), a 6-layer long-short-term memory model (LSTM), and an 8-layer Bidirectional Encoder Representations from Transformers (BERT) model. The study found that the BERT model outperformed both the CNN and LSTM-based models across all the evaluation metrics, achieving a remarkable classification accuracy of 95%. These findings confirm the reported effectiveness of transformer-based architectures in tasks related to natural language processing, such as sentiment analysis.

Preprints.org, 2024

Translational Medicine, Dec 18, 2023

Introduction: Cervical cancer is the fourth most prevalent cancer among women worldwide and is a ... more Introduction: Cervical cancer is the fourth most prevalent cancer among women worldwide and is a significant contributor to cancer-related deaths, with an estimated 300,000 women losing their lives to the disease annually. Most of these fatalities occur in Low and Middle-Income Countries (LMICs), such as Uganda, where access to screening and treatment options is limited. Early detection of cervical cancer is crucial to improve the chances of survival for patients. Currently, cervical cancer screening is typically performed through pap smears, which involve manual examination of cervical samples for abnormalities by medical experts. This process is
costly, time-consuming and prone to errors, leading to inaccurate diagnoses. Therefore, it is essential to find more effective and efficient alternative methods for cervical cancer screening to improve access in LMICs and alleviate the burden of cervical cancer.

Objective: The purpose of this study is to develop an automated pre-cervical cancer screening algorithm to detect precancerous cervical lesions.

Materials and methods: We developed a cancer screening algorithm using a 21-layer deep-learning convolution neural network trained on a dataset of 2300 images collected from local sources and some obtained from Kaggle.

Results: The best-performing classifier had an Area Under Curve (AUC) of the accuracy of 91.37%, a precision of 88.80%, a recall of 94.69%, an F1 score of 91.65% and an AUC of 96.0%.

Conclusion: The development and implementation of automated pre-cervical cancer screening algorithms have the potential to revolutionize cervical cancer detection and contribute significantly to reducing the burden of the disease, particularly in resource-limited settings.

This paper presents a comprehensive literature review on the application of artificial intelligen... more This paper presents a comprehensive literature review on the application of artificial intelligence techniques in Ugandan healthcare and the medical industry. Recently, the data generated in the health domain has exceeded the human cognitive capacity to analyze it effectively. Several approaches have been suggested to address this problem but in several studies, Artificial Intelligence has been found to be the best and the most effective solution as far as speed, accuracy, robustness, and reliability are concerned. We searched and reviewed AI health-related peer-reviewed articles in ScienceDirect, Springer, PubMed, arXiv, IEEE Xplore, medRxiv, PLOS, Wiley Online Library, BioMed Central, bioRxiv, and Scopus published between 2012 and 2022. This literature survey covered 38 research papers, and the review showed that the most applied AI subfields are statistical learning, machine learning, and deep learning. The paper highlights the challenges, gaps, and opportunities required to impr...

American Journal of Preventive Medicine and Public Health, 2023

There is a growing global interest in adopting digital tools during public health emergencies. Se... more There is a growing global interest in adopting digital tools during public health emergencies. Several tools have been developed to improve pandemic management activities, such as case investigations, contact tracing and follow-up of contacts' health status, data management and analysis, giving epidemiologists rapid warnings of potential virus exposure, and evaluation of different public health response measures. In an effort to control the September 2022 Ebola Virus Disease outbreak in Uganda, the ministry of health adopted the Go.Data tool to manage the outbreak data. In this paper, we examine the effectiveness of the Go.Data tool as a digital outbreak management system. We discuss the experiences, challenges, best practices, and recommendations.

Research Square, 2023

This paper presents a comprehensive literature review on the application of artificial intelligen... more This paper presents a comprehensive literature review on the application of artificial intelligence techniques in Ugandan healthcare and the medical industry. Recently, the data generated in the health domain has exceeded the human cognitive capacity to analyze it effectively. Several approaches have been suggested to address this problem but in several studies, Artificial Intelligence has been found to be the best and the most effective solution as far as speed, accuracy, robustness, and reliability are concerned. We searched and reviewed AI healthrelated peer-reviewed articles in ScienceDirect, Springer, PubMed, arXiv, IEEE Xplore, medRxiv, PLOS, Wiley Online Library, BioMed Central, bioRxiv, and Scopus published between 2012 and 2022. This literature survey covered 38 research papers, and the review showed that the most applied AI subfields are statistical learning, machine learning, and deep learning. The paper highlights the challenges, gaps, and opportunities required to improve and advance the application of AI in the Ugandan healthcare industry. We believe this study will help researchers and policymakers to foster AI innovations better.

Proceedings of 43rd Conference of the South African Institute of Computer Scientists and Information Technologists, 2022

One of the most challenging problems faced by ecologists and other biological re- searchers today... more One of the most challenging problems faced by ecologists and other biological re- searchers today is to analyze the massive amounts of data being collected by advanced monitoring systems like camera traps, wireless sensor networks, high-frequency radio track- ers, global positioning systems, and satellite tracking systems being used today. It has become expensive, laborious, and time-consuming to analyze this huge data using man- ual and traditional statistical techniques. Recent developments in the deep learning field are showing promising results towards automating the analysis of these extremely large datasets. The primary objective of this study was to test the capabilities of the state-of- the-art deep learning architectures to detect birds in the webcam captured images. A total of 10592 images were collected for this study from the Cornell Lab of Ornithology live stream feeds situated in six unique locations in United States, Ecuador, New Zealand, and Panama. To achieve the main objective of the study, we studied and evaluated two con- volutional neural network object detection meta-architectures, single-shot detector (SSD) and Faster R-CNN in combination with MobileNet-V2, ResNet50, ResNet101, ResNet152, and Inception ResNet-V2 feature extractors. Through transfer learning, all the models were initialized using weights pre-trained on the MS COCO (Microsoft Common Objects in Context) dataset provided by TensorFlow 2 object detection API. The Faster R-CNN model coupled with ResNet152 outperformed all other models with a mean average preci- sion of 92.3%. However, the SSD model with the MobileNet-V2 feature extraction network achieved the lowest inference time (110ms) and the smallest memory capacity (30.5MB) compared to its counterparts. The outstanding results achieved in this study confirm that deep learning-based algorithms are capable of detecting birds of different sizes in differ- ent environments and the best model could potentially help ecologists in monitoring and identifying birds from other species.

International Journal of Data Science and Big Data Analytics, 2021

The objective of this paper is to build a linear model for predicting the average amount of tip i... more The objective of this paper is to build a linear model for predicting the average amount of tip in dollars a waiter is expected to earn from the restaurant given the predictor variables, i.e., total bill paid, day, the gender of the customer (sex) time of the party, smoker and size of the party. The model was based on the data created by one waiter at a certain restaurant in California who recorded information about each tip he received. This model can be applied at any restaurant with similar predictor variables to determine the amount of tip. The final result from this analysis proved a regression model with a minimum prediction Root Mean Square Error (RMSE) of 1.1815.

SSRN Electronic Journal

Wireless communication is rapidly growing, making it possible to design wireless network systems ... more Wireless communication is rapidly growing, making it possible to design wireless network systems that can constantly collect, analyze, evaluate, and validate our environment to get more control of the factors that influence it. With over a decade of intensive research and development, wireless sensor network technology has been emerging as a viable solution to many innovative applications. Various audio wireless consumer devices have been developed over the years. But these wireless TV headphones use Bluetooth technology which comes with a number of drawbacks; high power consumption, high cost, short distance coverage, and a limited number of users at a time. In this project, we have developed a wireless TV audio transceiver (transmitter to multiple receivers) using Arduino and nRF24L01 module. The nRF24L01 transceiver module uses the 2.4 GHz band and it can operate with band rates from 250 kbps up to 2 Mbps. If used in closed space and with lower band rate its range can reach up to 100 meters. The Wireless audio system operates at Radio Frequency (RF) signals. Specifically, it utilizes the IEEE802.15.4 standard to transmit the audio signals. The system is designed to transmit and receive the audio signal about 2.4Ghz frequencies. The system is powered using a 9Vdc battery. The Wireless audio system utilizes the IEEE802.15.4 Radio Frequency (RF) standard to transmit the audio signals. IEEE standard 802.15.4 offers the fundamental lower network layers of a Wireless Personal Area Network (WPAN) and focuses on low-cost, low-power communication between devices. The system will be designed to transmit and receive the audio signal using the 2.4Ghz band. The transmitter converts the input analog signal from the TV audio socket to a digital signal using the microcontroller. The digital signal will then be sent to the nRF24L01 module which modulates it using Gaussian Frequency Shift Keying (GFSK) modulation scheme and transmits it at 2.4GHz. The receivers use GFSK modulation to demodulate the digital signal received and convert it to an analog signal using the microcontroller. The analog signal is amplified by the LM386 circuit where users can individually modulate the volume of sound of their preferences. LM386 is a low voltage audio amplifier and frequently used in battery-powered music devices.

SSRN Electronic Journal

The goal of this Project is to build a linear model for predicting the average amount of tip in d... more The goal of this Project is to build a linear model for predicting the average amount of tip in dollars a waiter is expected to earn from the restaurant given the predictor variables i.e. total bill paid, day, the gender of the customer (sex) time of the party, smoker, and size of the party. And this was achieved through the use of the Linear Regression method. The dataset of 200 observations and 7 variables was divided into training and testing sets in a ratio of 8:2 respectively. The model was fitted using the lm() function of R on the train set and tested on the testing set using predict() function. And the model fitness was deeply analyzed to understand how well it fits the data. Using Lasso regularization approach, the model was improved and this helped to identify the most important predictors in estimating the amount of tip received by the waiter. And also an interaction of size and smoker was included in the final model which greatly improved its data fitness.

Zenodo, 2021

We provide a dataset of images(.jpeg) with their corresponding annotations files(.xml) used to tr... more We provide a dataset of images(.jpeg) with their corresponding annotations files(.xml) used to train a bird detection deep learning model. These images were collected from the live stream feeds of Cornell Lab of Ornithology (https://www.allaboutbirds.org/cams/) situated in 6 unique locations around the world as follows:
-Treman bird feeding garden at the Cornell Ornithology Laboratory in Ithaca, New York. At this station, Axis P11448-LE cameras are used to capture the recordings from feeders perched on the edge of both sapsucker woods and its 10-acre ponds. This site mainly attracts forest species like chickadees (Poecile atricapillus), red-winged blackbirds (Agelaius phoeniceus), and woodpeckers (Picidae). A total of 2065 images were captured from this location.
- Fort Davis in Western Texas, USA. At this site, a total of 30 hummingbird feeder cams are hosted at an elevation of over 5500 feet. From this site, 1440 images were captured.
-Sachatamia Lodge in Mindo, Ecuador. This site has a live hummingbird feed watcher that attracts over 132 species of hummingbirds including:
Fawn-breasted Brilliant, White-necked Jacobin, Purple-bibbed Whitetip, Violet-tailed Sylph, Velvet-purple Coronet, and many others. A total of 2063 images were captured from this location.
-Morris County, New Jersey, USA. Feeders at this location attract over 39 species including Red-bellied Woodpecker, Red-winged Blackbird, Purple Finch, Blue Jay, Pine Siskin, Hairy Woodpecker, and others. Footage at this site is captured by an Axis P1448-LE Camera and Axis T8351 Microphone. A total of 1876 images were recorded from this site.
-Canopy Lodge in El Valle de Anton, Panama. Over 158 bird species visit this location annually and these include Gray-headed Chachalaca, Ruddy Ground-Dove, White-tipped Dove, Green Hermit, and others. A total of 1600 images were captured.
-Southeast tip of South Island, New Zealand. At this site, nearly 10000 seabirds visit this location annually and a total of 1548 images were captured.
The Cornell Lab of Ornithology is an institute dedicated to biodiversity conversation with the main focus on birds through research, citizen science, and education. The autoscreen software was used to capture the images from the live feeds and images of approximately 1 Megapixel (Joint Photographic Experts Group) JPEG coloured images of resolution 1366times768times31366\times 768 \times 31366times768times3 pixels were collected (https://sourceforge.net/projects/autoscreen/). The software was taking a new image every 30 seconds and were captured during different times of the day in order to avoid a sample biased dataset. In total, 10592 images were collected for this study.

International Journal of Data Science and Big Data Analytics, 2021

University of Cape Town, 2020

This project aims at developing, validating, and testing several classification statistical model... more This project aims at developing, validating, and testing several classification statistical models that could predict whether or not an office room is occupied using several data features, namely temperature (◦C), light (lx), humidity (%), CO2 (ppm), and a humidity ratio. The data is modeled using classification techniques i.e. Logistic regression, Classification tree, Bagging-Random forest, and Gradient boosted trees.
These models were trained and then after evaluated against validation and test sets and using confusion matrices to obtain classification and misclassification rates. The logistic model was trained using glmnet R package, Tree package for classification tree model, randomForest for both Bagging and Random Forest Models, and gbm package for Gradient Boosted Model.
The best accuracy was obtained from the Random Forest Model with a classification rate of 93.21% when it was evaluated against the test set. Light sensor is also the most significant variable in predicting whether the office room is occupied or not, this was observed in all the five models.

Makerere University , 2018

4Wireless communication is rapidly growing, making it possible to design wireless network systems... more 4Wireless communication is rapidly growing, making it possible to design wireless network systems that
can constantly collect, analyze, evaluate and validate our environment to get more control of the factors
that influence it. With over a decade of intensive research and development, wireless sensor network
technology has been emerging as viable solution to many innovative applications. Various audio wireless
consumer devices have been developed over years. But these wireless TV headphones use Bluetooth
technology which comes with a number of drawbacks; high power consumption, high cost, short distance
coverage and limited number of users at time. In this project, we have developed a wireless TV audio
transceiver (transmitter to multiple receivers) using Arduino and nRF24L01 module. The nRF24L01
transceiver module uses the 2.4 GHz band and it can operate with band rates from 250 kbps up to 2 Mbps.
If used in closed space and with lower band rate its range can reach up to 100 meters.
The Wireless audio system operates at Radio Frequency (RF) signals. Specifically, it utilizes
IEEE802.15.4 standard to transmit the audio signals. The system is designed to transmit and receive the
audio signal about 2.4Ghz frequencies. The system is powered using a 9Vdc battery. The Wireless audio
system utilizes IEEE802.15.4 Radio Frequency (RF) standard to transmit the audio signals. IEEE
standard 802.15.4 offers the fundamental lower network layers of a Wireless Personal Area Network
(WPAN) and focuses on low-cost, low-power communication between devices. The system will be
designed to transmit and receive the audio signal using 2.4Ghz band.
The transmitter converts the input analog signal from the TV audio socket to digital signal using the
microcontroller. The digital signal will then be sent to the nRF24L01 module which modulates it using
Gaussian Frequency Shift Keying (GFSK) modulation scheme and transmits it at 2.4GHz.
The receivers use GFSK modulation to demodulate the digital signal received and convert it to an analog
signal using the microcontroller. The analog signal is amplified by LM386 circuit where users can
individually modulate the volume of sound of their preferences. LM386 is a low voltage audio
amplifier and frequently used in battery powered music devices.

Drafts by mirugwe alex

Research Square, 2023

Introduction: Cervical cancer is the fourth most prevalent cancer among women worldwide and is a ... more Introduction: Cervical cancer is the fourth most prevalent cancer among women worldwide and is a significant contributor to cancer-related deaths, with an estimated 300,000 women losing their lives to the disease annually. Most of these fatalities occur in low and middle- income countries (LMICs), such as Uganda, where access to screening and treatment options is limited. Early detection of cervical cancer is crucial to improve the chances of survival for patients. Currently, cervical cancer screening is typically performed through Pap smears, which involve manual examination of cervical samples for abnormalities by medical experts. This process is costly, time-consuming, and prone to errors, leading to inaccurate diagnoses. Therefore, it is essential to find more effective and efficient alternative methods for cervical cancer screening to improve access in LMICs and alleviate the burden of cervical cancer.

Objective: The purpose of this study is to develop an automated pre-cervical cancer screening algorithm to detect precancerous cervical lesions.
Methodology: We developed a cancer screening algorithm using a 21-layer deep-learning convolution neural network trained on a dataset of 2300 images collected from local sources and some obtained from Kaggle.

Results: The best-performing classifier had an AUC of the accuracy of 91.37%, a precision of 88.80%, a recall of 94.69%, an F1 score of 91.65%, and an AUC of 96.0%.

University of Cape Town

The dataset of 200 observations and 7 variables was divided into training and testing sets in a ratio of 8:2 respectively. The model was fitted using the lm() function of R on the train set and tested on the testing set using predict() function. And the model fitness was deeply analyzed to understand how well it fits the data.

Using Lasso regularization approach, the model was improved and this helped to identify the most important predictors in estimating the amount of tip received by the waiter. And also an interaction of size and smoker was included in the final model which greatly improved its data fitness.

Makerere , 2019

This manual gives a holistic approach to Microsoft Office suites i.e. Ms. Word, Excel, PowerPoint... more This manual gives a holistic approach to Microsoft Office suites i.e. Ms. Word, Excel, PowerPoint and Publisher. It also introduces elementary knowledge required in C programming language.
Even beginners will find this manual easy to use because its a self-teaching guide.

Renewables Vs Fossils, 2016

The expansion of non-fossil energy sources such as nuclear and hydroelectricity may be limited in... more The expansion of non-fossil energy sources such as nuclear and hydroelectricity may be limited in the future because of availability, affordability, and acceptability reasons. There are however other non-fossil energy sources, some of which have been used for millennia by humans, principally biomass (both traditional and modern), wind (for mechanical and electrical applications), and solar (for heating and electricity). The importance of biomass cannot be underemphasized: in 2014, it met 10% of the world’s primary energy demand and will increase to 11% in 2040 – in both cases, more than nuclear and hydroelectricity combined. Over the same period, wind and solar are expected to increase from about 1% to 6%.

Conference Presentations by mirugwe alex

Proceedings of 43rd Conference of the South African Institute of Computer Scientists and Information Technologists, 2022

One of the most challenging problems faced by ecologists and other biological researchers today i... more One of the most challenging problems faced by ecologists and other biological researchers today is to analyze the massive amounts of data being collected by advanced monitoring systems like camera traps, wireless sensor networks, high-frequency radio trackers, global positioning systems, and satellite tracking systems being used today. It has become expensive, laborious, and time-consuming to analyze this huge data using manual and traditional statistical techniques. Recent developments in the deep learning field are showing promising results towards automating the analysis of these extremely large datasets. The primary objective of this study was to test the capabilities of the state-ofthe-art deep learning architectures to detect birds in the webcam captured images. A total of 10592 images were collected for this study from the Cornell Lab of Ornithology live stream feeds situated in six unique locations in United States, Ecuador, New Zealand, and Panama. To achieve the main objective of the study, we studied and evaluated two convolutional neural network object detection meta-architectures, single-shot detector (SSD) and Faster R-CNN in combination with MobileNet-V2, ResNet50, ResNet101, ResNet152, and Inception ResNet-V2 feature extractors. Through transfer learning, all the models were initialized using weights pre-trained on the MS COCO (Microsoft Common Objects in Context) dataset provided by TensorFlow 2 object detection API. The Faster R-CNN model coupled with ResNet152 outperformed all other models with a mean average precision of 92.3%. However, the SSD model with the MobileNet-V2 feature extraction network achieved the lowest inference time (110ms) and the smallest memory capacity (30.5MB) compared to its counterparts. The outstanding results achieved in this study confirm that deep learning-based algorithms are capable of detecting birds of different sizes in different environments and the best model could potentially help ecologists in monitoring and identifying birds from other species.

Preprints.org, 2024

Translational Medicine, Dec 18, 2023

Objective: The purpose of this study is to develop an automated pre-cervical cancer screening algorithm to detect precancerous cervical lesions.

Results: The best-performing classifier had an Area Under Curve (AUC) of the accuracy of 91.37%, a precision of 88.80%, a recall of 94.69%, an F1 score of 91.65% and an AUC of 96.0%.

American Journal of Preventive Medicine and Public Health, 2023

Research Square, 2023

This paper presents a comprehensive literature review on the application of artificial intelligen... more This paper presents a comprehensive literature review on the application of artificial intelligence techniques in Ugandan healthcare and the medical industry. Recently, the data generated in the health domain has exceeded the human cognitive capacity to analyze it effectively. Several approaches have been suggested to address this problem but in several studies, Artificial Intelligence has been found to be the best and the most effective solution as far as speed, accuracy, robustness, and reliability are concerned. We searched and reviewed AI healthrelated peer-reviewed articles in ScienceDirect, Springer, PubMed, arXiv, IEEE Xplore, medRxiv, PLOS, Wiley Online Library, BioMed Central, bioRxiv, and Scopus published between 2012 and 2022. This literature survey covered 38 research papers, and the review showed that the most applied AI subfields are statistical learning, machine learning, and deep learning. The paper highlights the challenges, gaps, and opportunities required to improve and advance the application of AI in the Ugandan healthcare industry. We believe this study will help researchers and policymakers to foster AI innovations better.

Proceedings of 43rd Conference of the South African Institute of Computer Scientists and Information Technologists, 2022

International Journal of Data Science and Big Data Analytics, 2021

SSRN Electronic Journal

Zenodo, 2021

International Journal of Data Science and Big Data Analytics, 2021

University of Cape Town, 2020

Makerere University , 2018

Research Square, 2023

Introduction: Cervical cancer is the fourth most prevalent cancer among women worldwide and is a ... more Introduction: Cervical cancer is the fourth most prevalent cancer among women worldwide and is a significant contributor to cancer-related deaths, with an estimated 300,000 women losing their lives to the disease annually. Most of these fatalities occur in low and middle- income countries (LMICs), such as Uganda, where access to screening and treatment options is limited. Early detection of cervical cancer is crucial to improve the chances of survival for patients. Currently, cervical cancer screening is typically performed through Pap smears, which involve manual examination of cervical samples for abnormalities by medical experts. This process is costly, time-consuming, and prone to errors, leading to inaccurate diagnoses. Therefore, it is essential to find more effective and efficient alternative methods for cervical cancer screening to improve access in LMICs and alleviate the burden of cervical cancer.

Results: The best-performing classifier had an AUC of the accuracy of 91.37%, a precision of 88.80%, a recall of 94.69%, an F1 score of 91.65%, and an AUC of 96.0%.

University of Cape Town

Makerere , 2019

Renewables Vs Fossils, 2016

Proceedings of 43rd Conference of the South African Institute of Computer Scientists and Information Technologists, 2022

University of cape Town, 2022

One of the most challenging problems faced by ecologists and other biological researchers today i... more One of the most challenging problems faced by ecologists and other biological researchers today is to analyze the massive amounts of data being collected by advanced monitoring systems such as camera traps, wireless sensor networks, high-frequency radio trackers, global positioning systems, and satellite tracking systems being used today. It has become expensive, laborious, and time-consuming to analyze the large datasets using manual and traditional statistical techniques. Recent developments in the field of deep learning are showing promising results towards automating the analysis of these extremely large datasets. The primary objective of this study is to test the capabilities of the state-of-the-art deep learning architectures to detect birds in the webcam captured images. A total of 10592 images were collected for this study from the Cornell Lab of Ornithology live stream feeds situated in six unique locations in United States, Ecuador, New Zealand, and Panama. To achieve the main objective of the study, two convolutional neural network object detection meta-architectures, single-shot detector (SSD) and Faster R-CNN in combination with MobileNet-V2, ResNet50, ResNet101, ResNet152, and Inception ResNet-V2 feature extractors were studied and evaluated. Through the use of transfer learning, all the models were initialized using weights pre-trained on the MS COCO (Microsoft Common Objects in Context) dataset provided by the TensorFlow 2 object detection API. The Faster R-CNN model coupled with ResNet152 outperformed all other models with a mean average precision of 92.3%. However, the SSD model with the MobileNet-V2 feature extraction network achieved the lowest inference time (110ms) and the smallest memory capacity (30.5MB) compared to its counterparts. The outstanding results achieved in this study confirm that deep learning-based algorithms are capable of detecting birds of different sizes in different environments and the best model could potentially help ecologists in monitoring and identifying birds from other species present in the environment.