Task 2, Unit 1 – What’s the Problem?

Unit 1, Section 2: How Do YOU Solve a Problem?

Q1: Identify an appropriate research philosophy for your project idea/research area.

Suitable research philosophies for training deep learning models in the field of Bone Marrow Transplantation could be a combination of two: positivism (ontology [1]) and pragmatism (epistemology [2]).

We know for example, based on previous research, results, and experience, that person A is a probable donor match for person B. This has been defined as a static and codified algorithm but are we missing something? In attempting to use AI to equivalence these algorithms, will it make the same assumptions or bring new insight and better matches.

Unit 1, Section 3: Known Knowns.

Q2: Select appropriate databases to identify research related to your problem investigation and use search strategies to identify relevant research related to your project area using e.g., Boolean searches.

Documentation of relevant research, theory and practice in Bone Marrow Transplantation is widely available. Possible databases (sources of information), more confined to the focus of HLA matching include:

  • World Marrow Donor Association (WMDA) resources [3]
  • European Marrow Donor Information System (EMDIS) databases and documentation [4]
  • Relevant medical journals and research.
  • Doctors, scientists, technicians, and researchers in the field.

Further to BMT, the topic of AI/Deep learning will also need to be researched:

  • University library [5]
  • Books [6], technical articles [7] and journals
  • Study courses, of note is Practical Deep Learning for Coders by Jeremy Howard, recorded at the University of Queensland [8].

Unit 1, Section 4: The Problem Is.

Q3: Select research strategies you would use to collect data for your project. And acquire feedback to improve the focus of your problem statement/research question.

Source and study historical matching data. This is available from several International sources as well as local databases such as the Welsh Bone Marrow Donor Registry (WBMDR).

References:

[1] Smith B, “Ontology”, [Online] Available: https://philpapers.org/archive/SMIO-11 [Accessed 06/06/2023]
[2] Goertz G, Mahoney J, “Concepts and measurement: Ontology and epistemology”, [Online] Available: https://libkey.io/libraries/1422/openurl?sid=google&auinit=G&aulast=Goertz&atitle=Concepts+and+measurement:+Ontology+and+epistemology&id=doi:10.1177/0539018412437108&title=Information+sur+les+sciences+sociales&volume=51&issue=2&date=2012&spage=205&issn=0539-0184 [Accessed 06/06/2023]
[3] WMDA, [Online] Available: https://wmda.info/ [Accessed 10/06/2023]
[4] PubMed, “EMDIS”, [Online] Available: https://pubmed.ncbi.nlm.nih.gov/7889762/ [Accessed 10/06/2023]
[5] UWTSD Library Resources, [Online] Available: https://librarysearch.uwtsd.ac.uk [Accessed 12/06/2023]
[6] Howard J, Gugger S, “Deep Learning for Coders with Fastai and PyTorch”, 1st Edition, O’Reilly, 2020
[7] Borisov V, Leemann T, Seßler K, Haug J, Pawelczyk M, Kasneci G, “Deep Neural Networks and Tabular Data: A Survey”, [Online] Available: https://arxiv.org/pdf/2110.01889.pdf [Accessed 30/05/2023]
[8] FastAI, “Practical Deep Learning”, [Online] Available: https://course.fast.ai/ [Accessed 21/05/2023]