With immune-related diseases and cancer as its main application areas, the Luxembourg Institute of Health specialises in the development of precision medicine. “Technological developments are making medical research more applied, and we feel that it is our obligation to make it more applicable to patients,” says CEO Ulf Nehrbass. “Our aim is to be able to define the right treatment track for the right patient.”

To reach this goal, the availability of clean, interoperable data is essential. The LIH is involved in several projects using data from patient cohorts. “However, biomedical data, such as blood and tissue samples, can have strong variations due to small differences in how biological samples were taken and manipulated, which make the follow-up analysis very complex. This is why we need to create strictly quality-controlled, standardised data.”

Generating interoperable data across borders

Ulf Nehrbass, CEO of the Luxembourg Institute of Health, speaks about data-driven reserach with patient impactA key project in this context is CLINNOVA, which involves clinicians and researchers from Luxembourg, France, Germany and Switzerland working on realising the benefits of precision medicine for treatment decisions through data federation, standardisation, and interoperability. “The clinical centres involved make huge efforts to ensure that samples are taken in exactly the same fashion,” explains Professor Nehrbass. “All samples are shipped to the Integrated BioBank of Luxembourg (IBBL) for processing, and then forwarded to the project partner in charge of that specific type of samples. The data generated is highly standardised and fed into a federated data space.”

I think that we will soon become an international go-to address for immune-related diseases.

The clinical centres involved in CLINNOVA work with large numbers of patients, which means that the initiative has already created substantial databases contributing to “clean” data lakes. “I think that we will soon become an international go-to address for immune-related diseases,” Professor Nehrbass points out, and refers to a recent collaboration with a Japanese research team that was able to validate a biomarker indicating the early onset of Parkinson’s diseases using Luxembourg’s remarkable Parkinson’s database.

Precision and prevention

In addition to biomedical data, the LIH has also ventured into the field of digitally captured data, notably voice recording. “Voice analysis is a very powerful tool for identifying the stage of the disease of a COVID patient,” says Professor Nehrbass. “We developed our own app, which was given in a clinical trial format to all the patients of a prospective COVID cohort. As we had other data, for example on how serious the infection was, we were able to train artificial intelligence (AI) algorithms able to recognise a COVID infection based on the voice alone.” The research centre is also looking at digital analyses of gait to identify Parkinson’s.

Creating digital avatars can provide a cheaper and faster ways to diagnose, define the state of the disease and look for a treatment predisposition.

Combining this wealth of data opens the door to unleashing the power of AI in healthcare. “Creating digital avatars can provide a cheaper and faster ways to diagnose, define the state of the disease and look for a treatment predisposition. With the help of AI, it is also possible to identify specific subgroups in patient cohorts that have something in common: a genetic disorder, a specific living environment or a type of diet, for example. Such analyses do not only enable precision medicine, but also new avenues for prevention.”

However, the availability of clean and structured data remains a necessity. “Training AI algorithms on retrospective, non-structured data has proven to be very difficult,” Professor Nehrbass points out. “They need to be developed based on clean and structured data sets. Once that is done, they become capable of providing relevant insights also from older, unstructured data. Clean data is clearly the value proposition that we can bring.”

Digital genotyping

The LIH has a good track record of cooperating with companies, and the centre is interested in extending collaboration into the data field. However, this is a delicate issue as health data is highly sensitive and protected by strict European regulations. “Before we go more into value creation around data, we need an environment that gives an interconnected, secure and legal data access,” says Professor Nehrbass. In this field, the LIH is working together with the Luxembourg National Data Service, a pioneering initiative developing solutions for sharing and reusing data for public and private data partners in a way that fully respects privacy and other regulations.

Professor Nehrbass is confident that in the next 5-10 years, the LIH will make a name for itself as a precision medicine institute. “I think that we will have a lead in digital genotyping, i.e. digital analyses of the genetic make-up of patients. Our high-quality databases will enable us to use AI in a fast, affordable way and identify new healthcare solutions that will fit the medical systems of the next 5-10 years.”

Photos: © Luxinnovation/Michel Brumat