Toward a knowledge graph for medical diagnosis: issues and usage scenarios

Availability of a large amount of data in the medical sector and an increasing computational infrastructure to exploit them allow the definition of new advanced services for medical diagnosis to support the work of medical doctors and to increase the quality of life of patients. Telemedicine, ambient assisted living, and a better management of electronic health records are some of the new generation of services that need to be further developed in the next years. Most of the issues that are faced by medical doctors and ICT practitioners defining these services are related to their level of trust and reliability. How to guarantee them is still in an open issue to be addressed. To this regard, practitioners have defined a plethora of standards and specifications to be used for better management of information and an increased level of trust of the adopting services. For instance, SNOMED CT and ICD-10 are standardized specifications that are widely used in clinical contexts. However, further issues arise due to the lack of interoperability among these different standards, and specifications adopted in the medical domain. Ontologies are recognized as an effective instrument to deal with the interoperability problem and the human disease ontology (DOID) and the Infectious Disease Ontology (IDO) are only some of the many proposals in the area. However, existing ontologies cover different aspects of the medical domain and have different objectives spanning from clinical diagnosis to a better organization of electronic medical records. To deal with such issues, we propose a knowledge graph for medical diagnosis leveraging and aligning existing largely used standards and ontologies. In detail, we present some of the typical issues to be faced and aligned them by focusing on ICD-10, SNOMED CT, DOID, and SYMP ontology. Then we discuss some scenarios of usage for the envisioned knowledge graph. We address how a knowledge graph can benefit interoperability of electronic health records, can support clinicians in their work and patients in increasing the quality of their lives by means of a new generation of telemedicine services, and can be an added value for medical insurance services.