Linked Open Health-related Fitness Data System

Tracking #: 2092-3305

Roberto Reda
Filippo Piccinini
Antonella Carbonaro

Responsible editor: 
Guest Editors Sensors Observations 2018

Submission type: 
Full Paper
In recent years, the market for the Internet of Things (IoT) has seen a proliferation of health wearable devices such as smart watches, fitness bands, and wellness appliances continuously collecting and storing a huge amount of physiological parameters. These data can be potentially exploited by the research community in order to gain valuable insights into our health systems. However, IoT self-tracked health data come from a variety of different heterogeneous sources and in proprietary formats, which often lead them to remain confined into separate data silos. Thus, when it comes to analysing IoT health and fitness datasets, data collection and data integration have to be done manually by domain experts. This time consuming and prone to error process significantly hampers an efficient exploitation of the information available. Semantic Web technologies can be a viable and comprehensive solution for describing, integrating and sharing heterogeneous IoT datasets. The aim of this work is to propose a web platform for the standardisation of data collection and integration to permit users to get a common view of the available information. To achieve our purpose we designed the IoT Fitness Ontology and we leveraged Semantic Web technologies in order to make IoT health and fitness datasets freely available to the community in a shared, semantically meaningful, and reusable manner.
Full PDF Version: 


Solicited Reviews:
Click to Expand/Collapse
Review #1
By Jean Paul Calbimonte submitted on 01/Feb/2019
Review Comment:

This paper presents a system that relies on Linked Data technologies to publish IoT health data.

This work addresses the problem of heterogeneity in fitness and health-related data collected by IoT devices such as wearables or smartwatches, with the goal of integrating information, potentially for research purposes.
This motivation comes from a valid idea of reusing this data for the common good, and using SW as the cornerstone technology to achieve this goal. However, I detect several issues in this paper, related to the general idea, as well as to the technical details presented, and the scope of the paper.

First, regarding formalities of the submission, this was presented as a full research paper, although it turns out it is more of a system paper. The LOD system is the main contribution, and the level of novelty from the paper is not too high. It uses standard LOD technologies for representation and storage of RDF data, and well known technologies (R2RML based) for transformation from non-RDF to RDF data. Judging from these facts, the paper might be more relevant as a 'systems paper', although it would need to be evaluated according to different criteria.

Second, the motivation of the paper is well intentioned, although it appears to me that the privacy aspect has been neglected, while it should be of foremost importance for this type of data. Health-related data, and even more, personal health data is extremely sensitive, and even if it is for research purposes, its collection and acquisition often requires to pass through ethical committees before approval. Even outside the medical domain, and only at the fitness and wellbeing levels, in the GDPR era, privacy has to be considered and addressed form the beginning, and the architecture should explicitly integrate protection and security concerns. As acknowledged in the paper discussion, this is not yet the case. As mentioned in Section 5, the proposed system intends to provide this data 'freely available on the Web'. The authors may need to rethink these aspects.

Third, about the ontology. It is not clear to me if the IFO ontology is part of the contributions of this paper or not. If understood correctly, this is not the case, as the ontology seems to have been published in [38,39]. In any case, it is surprising that it does not reuse the SSN ontology, which has been widely adopted for sensor observations. Several of the concept related to an 'Episode' could in fact be represented with the SSN ontology (and related vocabularies). Also, there does not seem to be a formal evaluation of the ontology. The authors mention an assessment by some consultant, although without any further information this cannot be considered as an ontology evaluation according to existing methodologies.

Fourth, there is no real evaluation of the system, as the authors acknowledge in the limitations subsection. Given that the paper focuses on the system itself as the main contributions, then the evaluation would be expected to focus on the different criteria relative to the functioning of the system. For example, evaluate the scalability of the system when exposed to large volumes of data, or how it responds to higher degrees of heterogeneity in the data sources, or how it can be assessed in terms of user experience. The reported experiences with the system are not a real evaluation, they only provide an initial feasibility indication. As mention before, the privacy aspect is also of primary importance, and evaluation should consider this as well.

Fifth, technically, the choices need to be better motivated. For instance, why Fuseki, is it scalable enough for IoT scenarios? How about considering RDF stream processing solutions? are they in any way useful for this kind of very dynamic use case. This is not a trivial topic, as sensor data can quickly grow out of control in terms of volume and velocity.

Overall, the paper presents a system but lacks evaluation, motivation for some technical choices, and disregards important aspects such as velocity and privacy.

Review #2
By Utkarshani Jaimini submitted on 07/Feb/2019
Major Revision
Review Comment:

The manuscript raises a very important problem of fitness data collection and addresses. The authors created a semantic framework to integrate the fitness data from various devices for making it semantically more meaningful, sharable and reusable. The authors have provided an in-depth review of the related work which would help a novice reader to follow through the entire paper.
Abbreviations definition are missing before their first use such as in the case of IFO in section 2 and SQ in section 6.2. Explain the terms used in the ontology with examples such as Temporal Relationship, Body Posture etc. The authors did not provide a demo link to the ontology and the portal. The ontology is not available online for review. Please publish the ontology for the community to review and give feedback. Due to the unavailability of the ontology, it is difficult to assess and review the completeness of all the concepts in the ontology. The authors have categorized the episodes as physical activity and body measure. Sedentary minutes measured by the Fitbit is not a physical activity and is applicable to determine a person’s fitness goal etc. How do the authors plan to capture Sedentary minutes in the ontology? I suggest making the category generic as Activity. What is the time period for retrieving the data once permission is obtained? Why was the Apple Health data manually entered? What are the statistical operations mentioned in Section 6.1? Please provide the sample queries either in the manuscript or appendix or supplementary material. How do the authors plan to address the issue of data privacy in the future? Will they be incorporating HIPAA protocols in the portal. Please provide a scenario or a snapshot to describe the ontology’s functionality for use in real-life. The manuscript is well written with minor typing errors such as Section 4.2 hasTime Frame instead of hasTimeFrame, Section 5.1 I”n our system, once the IoT data is retrieved manually of automatically from remote servers. Use OR instead of OF, Section 5.4 Our system allows users access their personal data, USERS TO ACCESS.

Review #3
Anonymous submitted on 11/Mar/2019
Major Revision
Review Comment:

The paper presents and discusses a system about the consumption of open health related data.
The platform presented by the authors focuses on data provided by IoT devices.
The contribution of the paper is two-fold: the IFO ontology, a top-level ontology for the semantic representation of health data, and a LOD portal for consuming data collected.

The topic discussed in the paper is definitely timely and of interest for the community.
However, my feeling is that the paper is still incomplete for being accepted for publication.
I summarize below the two parts that the authors should consider to extend for the next round of review.

1) The IFO Ontology
The authors should provide a link to the ontology.
By considering that this is one of the contribution of the paper, the Reviewers should be able to check and evaluate it.
I invite the authors to check the papers of the Resource track published during the last two editions of ISWC for seeing how an ontology should be presented, discussed and evaluated.
Indeed, concerning the evaluation, the section provided by the authors is quite poor.
Please check the following papers for extending the methodology adopted for evaluating the proposed ontology:
- Obrst, L., Ceusters, W., Mani, I., Ray, S., Smith, B. In: The Evaluation of Ontologies. Springer US, Boston, MA (2007) 139–158
- Gangemi, A., Catenacci, C., Ciaramita, M., Lehmann, J.: Modelling ontology evaluation and validation. In Sure, Y., Domingue, J., eds.: The Semantic Web: Research and Applications, 3rd European Semantic Web Conference, ESWC 2006, Budva, Montenegro, June 11-14, 2006, Proceedings. Volume 4011 of Lecture Notes in Computer Science., Springer (2006) 140–154
- Gomez-Perez, A.: Ontology evaluation. In Staab, S., Studer, R., eds.: Handbook on Ontologies. International Handbooks on Information Systems. Springer (2004) 251–274
- Gruber, T.R.: Toward principles for the design of ontologies used for knowledge sharing? Int. J. Hum.-Comput. Stud. 43(5-6) (1995) 907–928
- Gruninger, M., Fox, M.: Methodology for the Design and Evaluation of Ontologies. In: IJCAI’95, Workshop on Basic Ontological Issues in Knowledge Sharing, April 13, 1995.

While, concerning the comparison with the state of the art and the link with existing resources, I suggest the authors to check the following paper:
- Mauro Dragoni, Tania Bailoni, Rosa Maimone, Claudio Eccher: HeLiS: An Ontology for Supporting Healthy Lifestyles. International Semantic Web Conference (2) 2018: 53-69
In particular, I invite the authors to:
- see if there is the possibility of reusing existing concepts (if the authors will share the ontology, the Reviewers will surely help them on this);
- check the literature for extending the state of the art of the manuscript, the paper contains some pointers to existing health-related ontologies.

2) The LOD portal
Also for the portal, a link should be provided.
Moreover, a user evaluation of the portal should be performed by the users.
Also here, I strongly invite the authors to check the papers from the In-Use track published during the last two editions of ISWC for seeing how a system should be discussed, presented and evaluated.
At least a usability study is mandatory for a contribution like the one presented by the authors.

Please check text for minor grammatical errors.
As example, page 6, row 16: "is build" -> "is built"