Tuberculosis (TB) is the ninth leading cause of death worldwide. In 2016, an estimated 417,000 people died from Tuberculosis in Africa, and 1.7 million globally. In South Africa Tuberculosis is the leading cause of death with 450,000 people developing the disease every year and 89,000 people dying from it. That’s ten people every hour.
With over 800,000 confirmed cases and 18,000 deaths recorded in Africa and over 17 million confirmed cases and 670,000 recorded deaths globally as of August 2, 2020, the COVID-19 global pandemic continues to take a heavy toll around the world. In countries with a high prevalence level of TB, TB can create additional complexity to the COVID-19 response. And by the same token, COVID-19 is adding new complexity to the ongoing battle against TB.
For both TB and COVID-19 patients, medical imaging (Chest X-Rays and/or CT Scans) can sometimes be performed to identify and manage any chest abnormalities that may develop.
DISCLAMER
This challenge is solely for educational purposes. Chest X-Rays is one of various tools that can be used for triaging and screening for Tuberculosis. Furthermore, it is possible for an individual to develop different types of infections. This is to be treated purely as an exercise in tool development.
One potential follow up of this research is the detection of specific chest abnormalities associated with either COVID-19 or Tuberculosis, which could help in the development ofpersonalised treatment for patients. Promising projects are eligible, but not guaranteed for additional mentoring into a publication.
MinoHealth is a startup and multifaceted system with the objective of democratising Quality Healthcare with innovative and cutting edge technologies like Artificial Intelligence, (Big) Data Analytics and Cloud computing in Africa. minoHealth‘s AI Research lab, minoHealth AI Labs researches and experiments with Artificial Intelligence and ways it can be applied to Healthcare to make it faster, better and yet cheaper. They research and apply Artificial Intelligence to fields like Biotechnology, Neuroscience, Optometry, Epidemiology and Dietetics/Nutrition. Their research projects include ‘ScaffoldNet: Detecting and Classifying Biomedical Polymer-Based Scaffolds via a Convolutional Neural Network’, published in the Scientific Book, “Advances In Information and Communication” and “End-to-End Learning via a Convolutional Neural Network for Cancer Cell Line Classification” published in Journal of Industry - University Collaboration>. Their ongoing research projects also includes a collaboration with West African Centre for Cell Biology of Infectious Pathogens (WACCBIP) on the use of Machine Learning in the identification of a multivariate signature for Malaria Immunity. They are members of Massachusett Institute of Technology (MIT)'s Global BioSummit Community. They are also the lead for the Topic Group on Artificial Intelligence for Radiology, under the United Nations International Telecommunication Union (ITU) and World Health Organization (WHO) Focus Group on Artificial Intelligence for Health (FG-AI4H). minoHealth’s platform was listed in AppsAfrica’s ‘5 African innovations disrupting traditional sectors’.
reRunmila AI Institute is dedicated to preparing Ghana, and the rest of Africa for an Artificial Intelligence (AI) Future and the 4th Industrial Revolution. The world is drastically changing with all leading nations and blocs creating AI Plans & Strategies. Furthermore, plenty of these leading nations and blocs have already started implementing these strategies; Africa shouldn't be left behind. Through the skillful use of AI, we can leapfrog our development and solve many of Africa's grand challenges, especially connected to the 'Sustainable Development Goals (SDGs)'. Our goal is to train the talents that Africa requires to build this AI Future that the continent deserves and needs. We are doing this by equipping Africans, starting with Ghanaians, with practical skills in AI (Deep Learning & Machine Learning) and Data Science.
Our courses and workshops are practical, being Development-focused and designed around optimally teaching complex concepts in a fast and easy-to-grasp approach. Our students ?take away skills to develop their own AI projects that they can readily implement for their needs and demonstrate to would-be investors or employers. These courses and workshops are crafted by Artificial Intelligence Engineers and Researchers who are already working on real world applications of Artificial Intelligence in Africa.
The challenge asked to build an AI model that can classify Tuberculosis and Normal X-Ray results. With Tuberculosis infections still active as the COVID-19 pandemic continues, an automated tool to help identify TB has the potential to reduce hospital workload and optimize patient care during a time when hospitals are being overwhelmed by COVID-19 cases.
The evaluation metric for this competition is the Area Under the ROC curve (AUC). Sample submission should look like, where LABEL is the probabiltiy that the lung x-ray is TB-positive. kept values as probabilities.
| ID | LABEL |
|---|---|
| GTWSHFYQ | 0.45 |
| QTFSSMGD | 0.31 |
| TBLBHSYT | 0.90 |