The gap shouldn’t be that wide between US and EU clinical trials
Human beings, and to a greater extent scientists, try to order items for a purpose. In biomedical sciences, probably no other subject is more diversely ordered or classified as human conditions or diseases – hereinafter abbreviated as ‘diseases’. None of these classifications claims to be the one and only truth or is able to serve all purposes. Instead, many classifications co-exist and are widely accepted – or enforced – by the various stakeholders in life sciences: lab scientists, clinicians, pharmaceutical and biotech firms, regulatory bodies, governments, etc. With different classifications come different definitions of terms even if they are highly similar or have exactly the same meaning.
But what about searching, comparing and analyzing similar data where different diseases classifications are applied? Or what about compiling data about diseases that are produced and maintained for a specific purpose?
Let’s take clinical trial registry data as an example. The central clinical trial repository of the United States, ClinicalTrials.gov, uses the Medical Subject Headings (MeSH) to encode the human condition under study while the other big repository, the EU Clinical Trials Register, uses the Medical Dictionary for Regulatory Activities (MedDRA). Because both registries make use of another disease classification, they can’t be merged automatically on the level of diseases.
MeSH is maintained by the National Library of Medicine (NLM) and is well known among scientists as it’s used, among other things, to classify publications from PubMed, the generally accepted ‘golden source’ of biomedical literature. MedDRA is less well-known in the fundamental scientific community but has greater adoption in a clinical, healthcare and regulatory setting like health economics or pharmacovigilance.
We, at ONTOFORCE, try to overcome this issue of unmergable clinical trial data by using as much as possible the Unified Medical Language System (UMLS), also managed by the NLM, to map the disease terminologies drawn from many different sources. In our search engine DISQOVER we bring UMLS, MeSH, MedDRA and many other classifications like International Classifications of Diseases (ICD versions 9 and 10), Orphanet Rare Disease Ontology (ORDO), Human Disease Ontology (HDO), and Human Phenotype Ontology (HPO) together (see Fig 1).
Figure 1: Screenshot from filter widgets for diseases in DISQOVER. The general term ‘diabetes mellitus’ is selected in the MeSH classification tree. This automatically responds in an update of the other classification trees. The trees are partially expanded on lower levels via the V-shaped icon and scrolling can be done with the up and down arrows.
Our semantic technology and unique user interface makes the mappings of these diseases meaningful and useful. Search for and browse through diseases while using the filter widgets that show the disease classification as a list or a structured tree. If you’re used to working with a classification, you can use it to make specific filterings – even if the related disease data wasn’t encoded with this classification – thanks to the `universal mapping’ feature. At the same time, you can see how the filtered diseases are ordered in the other classifications as you can see the classifications all together.
To conclude, the gap shouldn’t be that wide between US and EU clinical trials anymore. From now on, you can search and analyze data in both registries automatically. And we continue to further extend and improve your search for disease information.
More about this in my next blog.