News

Smarter healthcare starts with better connections, better data and faster and more detailed analysis. That’s exactly where the Belgian Extreme Blue students have been working on with their Rare Disease Diagnosis Project.
In the EU, a rare disease is one which affects fewer than 5 people in 10 000. This number actually translates into an estimated population of 29 million European people affected by rare diseases, a situation that enforced European Institutions to recognize the rare disease issue as a key area for growth in Healthy Life Years.
Because of the low frequency of these pathologies, there are a lot of challenges to address. First of all, the quality of life of the patient really depends on how early a diagnosis is made. Currently, this is one of the major challenges within this field; patients are often diagnosed too late, which has a big impact on their quality of life, their life expectancy and the treatment costs. This diagnosis delay is, among others, caused by a worldwide scattering of information and expertise combined with a lack of appropriate communication tools and by the difficulty of tracking patient’s history.

To address these challenges, IBM’s Extreme Blue student team and UZA joined forces and came up with a reasoning engine, to improve the knowledge and diagnostic capability of these rare diseases. Within what’s called the Rare Disease Diagnosis project, scientific knowledge and technology became intertwined as never before and showed how ICT can support smarter diagnoses and in that way improve healthcare quality and cost-effectiveness.

In order to diagnose rare diseases, single access to information coming from various sources like clinical data, medical records, pharmacy and labs is necessary. Therefore the students introduced the semantic data warehouse, a new layer of intelligence, built on top of the current separated databases. The layer translates data into a uniform and standard language based on healthcare industry standards and integrates in that way systems inside the hospital. It’s possible to see even further and also integrate databases across different hospitals. The possibilities of exploiting this amount of structured data are endless. Indeed, this system allows for medical history retrieval across different departments, tracking of treatment outcomes, inter-hospital collaboration and data exchange and … smarter diagnoses. The reasoning engine is a unique combination of a rule engine and a statistical analysis tool and is build on top of such a layer.

The specialists of the UZA were already experimenting with rules to identify patients by screening patient’s records (test results, symptoms …) looking for the ones who satisfy the rules. Nevertheless, it was difficult and time consuming to document and execute them and the results were with limited success. The new rule engine is able to detect several gaps and conflicts in the rules written by specialists and to present them in a user-friendly way, so that specialists can easily analyze and improve them.

Yet, this was not sufficient enough to get satisfying results. The students also introduced statistical analysis on the data. The statistical modeling tools analyze data from normal and sick patients and derive additional new rules which can be presented to the specialists. The tool is also able to identify patients in a large population via anomaly detection. Both techniques of rules and statistical analysis were combined in the reasoning engine. The results of the rules generated by the statistical tools and executed on the rules engine on existing data were exceeding all expectations. Indeed, the engine was able to identify patients in a large data sample with both a higher sensitivity and specificity.

The unique combination of specialist’s knowledge, a rules engine and statistical analysis was the decisive factor in this pilot. The project proved that medical knowledge combined with modern IT tools can support an earlier, quicker and more accurate diagnosis. This reasoning engine can serve as an intelligent and dynamic knowledgebase on rare diseases who can mean the difference in the lives of these rare disease patients.

Contact
U.Z.A

Ann Segers,

Communicatieverantwoordelijke UZA

Mobile: +32 472 93 00 30

Email: ann.segers@uza.be

IBM

Yves Van Seters,

Media Relations IBM Belux

Mobile: +32 478 27 10 33

Email: yvanseters@be.ibm.com

Twitter: @YvesVS