Initiatives

Initiative: State of Machine Learning and Healthcare in Africa

Goal

Toward better understanding of the Machine Learning and Healthcare community, research, industry and interest in Africa


Projects

  • Machine Learning and Healthcare in Africa: Bibliometric Study and Systematic Review (active)


Roadmap

  • Complete initial project to make an African Health dataset accessible

  • Use initial project as template to make additional datasets accessible

Machine Learning and Healthcare in Africa: Bibliometric Study and Systematic Review

Initiative: Privacy Preserving AI for African Health Datasets

Goal

Use privacy preserving AI tools to help make African health datasets more accessible


Projects

  • Initial Privacy Preserving African Health Dataset (active)


Roadmap

  • Complete initial project to make an African Health dataset accessible

  • Use initial project as template to make additional datasets accessible

Privacy Preserving AI for African Health Datasets
GoalUse privacy preserving AI tools to help make an initial African health dataset accessible

Project leadArchie Arakkal (contact on Discord @ Archie#9168)
Description TODO

Entrypoints
  • Help with code base
  • Help with academic paper writing
  • Looking for help with process of ethical approval around Privacy Preserving AI

Initiative: Semantic Applications for Biomedical Data Science

Goal

Developing approaches to use semantic resources, particularly open and collaborative knowledge graphs like Wikidata, for driving knowledge-based systems for clinical decision support and biomedical informatics


Projects

  • MeSH2Matrix: Machine learning-driven biomedical relation classification based on the MeSH keywords of PubMed scholarly publications (Accepted to BIR@ECIR)

  • Data models for annotating biomedical texts: the case of CORD-19 (Accepted to Sci-K@WWW)

  • Recommending scholarly articles to monitor COVID-19 trends in social media based on low-cost topic modeling (active)

  • MeSH2Ontology: Machine learning-driven biomedical ontology creation based on the MeSH keywords of PubMed scholarly publications (active)


Roadmap

  • Identify gaps towards the use of knowledge resources in biomedical applications

  • Develop algorithms to enhance the development of knowledge-based systems in biomedicine

  • Create methods and datasets to evaluate and adjust knowledge-based approaches

  • Validate proposed approaches for using knowledge resources in biomedical informatics

Project: Semantic Applications for Biomedical Data Science
GoalDevelop the use of knowledge resources in biomedical data science

Project leadsHoucemeddine Turki (contact on Discord @ csisc#7682)Bonaventure DossouChris Emezue
Description At the information age, many semantic resources are freely made available online. These resources include bibliographic databases (e.g., PubMed), taxonomies (e.g., MeSH), ontologies (e.g., Disease Ontology), and knowledge graphs (e.g., Wikidata). However, scientists tend to exclusively use advanced machine learning techniques for developing computer applications in biomedical data science. These techniques require a lot of human capacities and funding to work. So, they are not well adapted to the African context characterized by the scarcity of its support to research. Given that the datasets that are used for training machine learning models generally have semantic values (e.g., Electronic Health Records, Biomedical Publications), semantic resources can be embedded to algorithms allowing knowledge-aware and explainable machine learning with a limited complexity. This will allow users to achieve interesting accuracy for biomedical informatics applications at a low cost.

Entrypoints
  • Help with code base
  • Help with dataset creation
  • Help with academic paper writing
  • Propose biomedical informatics approaches based on knowledge resouces