The Brainteaser Ontology: a Progression and Monitoring of Brain-related Diseases

The Brainteaser Ontology: Progression and Monitoring of Brain-Related Diseases
The manuscript provides an overview of the BrainTeaser Ontology (BTO), an event based ontology that represents anamnestic history of patients and clinical course of patients with Amyotrophic Lateral Sclerosis (ALS) and Multiple Sclerosis (MS) diseases.

This manuscript seems like it was written by ChatGPT or some other generative AI!
This manuscript does not begin like a scholarly work. The first section seems mostly based on some throw-away commercial website that is intended to explain ALS to the general public.

The authors use words like ‘nerves’ to describe central nervous system disease. The terminology appears to be based on a sophomoric understanding of the disease, bringing into question whether their ontology is similar. This is simply unacceptable.

Specific problems:
First paragraphs are supposedly based on only one reference (1997 medical textbook) and a bunch of general-public targeting websites. Please update this section to write it using a list of academic references such as original work and academic reviews.

Statements such as “BTO will open up unprecedented opportunities to build data-driven, knowledge-informed Artifical Intelligence (AI) tools for diagnosis and/or progression prediction of ALS and MS diseases” which are grand claims with no backing.

“Compared to previous efforts, BTO overcomes state-of-the-art by jointly modeling two brain-related diseases and focusing on patients and their clinical history and lifestyle – rather than on biochemical processes associated with MS and ALS as related works do” this statement also has no reference. They previously cite two conference challenges, but nothing can be discerned about this. Is it that each team in each of the conferences only chose to model one disease MS or ALS and that this team takes on both? I do not see evidence that these authors read the work from the previous challenges or at least they do not say anything reasonable about how others have solved this problem in the past or how their efforts are different or better.
It is simply unacceptable to explain an ontology as being “BTO is a domain ontology that focuses on ALS and MS disease modeling and is based on several foundation ontologies (and, in general, semantic resources) to grant homogeneity and compatibility with already existing resources.” This is the entire paragraph. Which ontologies and other semantic resources is it based on? Is it the FoF ontology, referenced above?

The second section is a laundry list where authors go through what looks like a literature review of other ontologies, listing each one with a table as well and a cursory pointer to why BTO is better, at least it goes through a lot of the previous work, but it is quite superficial and misses the most relevant work, the STMSO, which must be considered here as it is most relevant and at least partially adheres to the BFO standard (which is not referenced here by the way), unlike the BTO. I am not necessarily saying that BFO is the end all of modeling, but it is a major standard that should be addressed.
Authors also do not state how they chose to create an ontology for the signs and symptoms of only two diseases without referring to the symptoms ontology http://purl.obolibrary.org/obo/symp.owl

Some other bits stand out:
Scope of the BTO includes:
“Anamnestic” origin of both MS and ALS? Can authors please explain how neurodegeneration in ALS relates to the immune system? Although the etiology looks similar, as far as I recall, it is very specifically not the same root cause, thus it would be expected that any similarities in disease progression would follow a path that many diseases take, not just these two, so the symptoms ontology or other broader ontology efforts are more appropriate for this modeling than this effort.
BTO is FAIR?
This group purports to have created a FAIR resource, but FAIR means Finable (first letter by the way). Ontologies are typically referenced by their three letter abbreviations, BTO in this case. However, BTO is a well-established ontology called the BRENDA Tissue Ontology. This is well used, included in BIOPORTAL, and properly formatted. The point of FAIR is to properly name research resources to reduce things like name collisions. https://bioportal.bioontology.org/ontologies/BTO/?p=classes&conceptid=root
Other problems with FAIR,
Authors state that the ontology uses SNOMED and NCIT however I do not see any mapping to any of these ontologies in their examples.

Query 1 shows things like
BTO:Patient - where is the mapping to all other ontologies? BTO is not the first ontology to define a patient.
BTO:?sex - What is this? Why is this here, it is completely inaccurate for patient data. If the authors wish to genotype the patients or have genotype information, which I do not see, then they might use this label, but the appropriate attribute for human patients should be gender.

There are already common data elements (CDEs) in place for these sorts of basic attributes of patients. How is BTO better than well-defined and community standard common data elements?

This paper presents the development of BrainTeaser Ontology (BTO) which is an ontology that models various aspects involving clinical events of Amyotrophic Lateral Sclerosis (ALS) and Multiple Sclerosis (MS) diseases. BTO has been developed with OBO design principles and FAIR standards taken into consideration. Some automated tools have been leveraged to ensure the consistency and syntactical accuracy of BTO. The ontology itself has been well-designed with 8 different semantic areas that correspond to ALS and MS disease progression. Though it has been developed for ALS and MS, it seems there is potential in generalizing it to other diseases which highlights the advantages of the modeling strategy employed by the authors. The paper also contains a discussion about a few downstream applications that use BTO. Furthermore, some example queries on BTO are given.

The paper is well-written discussing all the aspects of the modelling of BTO in detail, also providing examples where necessary. The authors have provided access to BTO in Zenodo through a .ttl file which can be opened with Protégé ontology editor. A link to full documentation is also provided there.

My comments on this paper are as follows.

1. While the diagrams provided are easy to understand, I would suggest the authors to include a key describing each component of the diagram. e.g. arrows, lines, circles, triangles etc.
2. BTO has used NCIT as the primary external resource. One can assume that being the largest clinical terminology, SNOMED CT might have been the more suitable choice. NCIT is also more cancer-focused. What was the reason for the authors to go with NCIT instead of SNOMED CT?
3. Table 4 shows the annotations of a BTO concept. I was wondering why separate BTO concepts were created and mapped to NCIT with “isDefinedBy” instead of directly importing the relevant NCIT concepts to BTO.
4. Is “broaderTransitive” a “part-of” relation? It was not entirely clear the difference between this and “subclassOf”
5. Can the data include Persons who are not patients (maybe relatives)? Otherwise, I am not entirely sure the need to have the "Person" abstraction if all the persons are patients.
6. Please clarify the difference between “hasKinship” and “inKinshipWith.”
7. Is the “hasDisease” object property ever used to model patients’ diseases or are the diseases always associated with an event?
8. Section 4.3.1 mentions SNOMED CT was used to model “SurgicalType.” Did you use a specific subhierarchy of SNOMED CT like “Procedure”?
9. Are the queries SPARQL queries? Please mention the query language in the paper.
10. Table 3 has not been referenced in the text.
11. Page 6, line 42  missing comma after the word “possible”
12. Page 11, line 29  replace “…the query for return…” with “… the query to return …”
13. Page 18, line 40  repeated words “the risk of”

The Brainteaser Ontology: a Progression and Monitoring of Brain-related Diseases
Tracking #: 3454-4668
Decision: Minor revisions required

1. Quality and relevance of the described ontology: low-medium. While the ontology is relevant to specific clinical applications, it is not apparent that this ontology (in particular) would function better than a general disease ontology or clinical data ontology. Crafting an ontology for each disease seems overkill unless the authors can motivate the need for an ontology for a specific disease due to its rarity, uniqueness, or complexity. The authors do not provide such motivation, so the reader is left wondering why we need (yet another) disease/clinical data ontology. Regarding quality, I’ve made some specific comments below regarding the quality of the modeling and structure of the ontology and where improvements can be made. The ontology (in its current state) seems in need of curation, vetting, and development.
2. Illustration, clarity and readability of the describing paper, which shall convey to the reader the key aspects of the described ontology: medium. The clarity and readability of the paper is reduced due to grammatical errors, lack of clear reasons for explaining design decisions, and detailed descriptions of the classes and relations in the ontology. The data files are easily accessible, well-documented (including a README), and organized.

Overview:

This is a paper describing an ontology (the Brain Teaser Ontology (BTO)) that seeks to represent Amyotrophic Lateral Sclerosis (ALS) and Multiple Sclerosis (MS) – two types of neurological diseases. The paper and the ontology itself are interesting and it is apparent that much work has gone into each. I thank the authors for their thoroughness and insight. At this time, I recommend the publication for acceptance on condition of minor revisions required. I will provide non-critical and critical general comments below followed by some more detailed comments. When I am able, I will make specific recommendations on how to improve the paper and/or ontology. Additionally, there are a few grammatical errors that can be rectified to improve the readability of the manuscript. I recommend additional editing to fix these errors.

The ontology is easy to access, publicly available, and well-documented via (http://brainteaser.dei.unipd.it/ontology/). The authors make it clear that the focus of the ontology and its development is `clinical events` that are `associated` with ALS and MS. These include things like disease onset, symptoms, procedures, and relapses.

Detailed comments:

Section 1: none.

Section 2: This section provides a brief overview of previous or current ongoing efforts to model MS, ALS, or neurological diseases in ontologies.

Section 3: This section provides details about the design of BTO and the motivation of those design decisions. One phrase I’ve read a few times is that BTO utilizes an “event-based approach.” It would be helpful if the authors described (in detail) what this is and provided some references to its history. Many ontologies describe events (or equivalent), so it is a bit confusing what it means to use an “event-based approach.” Further, there are many classes in BTO that are not events, so this approach does not preclude the developers from modeling non-events in BTO.

Along these lines, it is touted that “event-based approaches” confer some serious advantages to other approaches (I am not sure what these other approaches are, nevertheless). Specifically, that this approach “allows to provide [sic] a unified model instead of using difference resources for each disease and it enhances ontology re-use as it is easier to extend BTO to represent other events or other diseases.” (section 3) This (or something similar to this) is present in the text a few times but it is unclear to me what exactly this means. Since the authors provide no further detail or evidence, I’m left wondering what this means. Some examples, elucidations, and descriptions would work well here.

The authors claim that BTO “complies with the Open Biological and Biomedical Ontology Foundry (OBO) and FAIR principles.” While the authors provide a list of ways that they judge BTO to be compliant with OBO principles, it should be made clear that BTO is not an OBO Foundry Ontology and that those principles have not yet been satisfied according to OBO Foundry (i.e., BTO has not applied for membership/inclusion in OBO Foundry).

Along these lines, OBO Foundry ontologies enforce re-use of ontologies through an import file mechanism wherein the imported/re-used terms retain their IDs (URIs) from the source ontologies. BTO re-mints or assigns new IDs using their own schema, which can cause interoperability obstacles with other ontologies. It would be good if the authors could justify their decision to mint new URIs/IDs for concepts in BTO rather than the standard practice of re-using existing URIs/IDs.

Section 4: minor grammatical errors

Section 5: minor grammatical errors

Section 6: minor grammatical errors

Section 7: none