Precision health – giving patients a voice How Luxembourg is supporting the fight against pandemics and chronic diseases, one patient at a time

Guy Fagherazzi has his work cut out. Recruited in 2019 as a researcher at the Luxembourg Institute for Health (LIH) for his expertise in epidemiology and cohort studies, the scientist found himself at the epicentre of Luxembourg's response to COVID-19 in 2020. With the support of the Government, the LIH and the University, he leads the Predi-COVID study, which gathers data from COVID-19 patients. The aim of the study is to shorten the response time in the event of a future pandemic. But Guy Fagherazzi's main work is on precision health, in other words the way in which treatments can be adapted to each patient and their individual needs, beyond the clinical environment.

Dr Fagherazzi, we have just mentioned personalised medicine and precision health. What do these terms actually mean?

Personalised medicine and precision health are two closely related concepts. Personalised medicine focuses on treating, diagnosing and monitoring patients, while precision health is a broader concept that goes beyond personalised medicine and also covers aspects of prevention and precision public health. The basic idea is to move away from standardised medicine for everyone, in which the same approaches are used regardless of the patient, and towards much more individualised treatments. We try to identify the best approach in line with each individual's profile so that we can meet the patient's needs as closely as possible.

In your personal statement on the LIH website, you indicate that you want to 'pave the way for modern real-life health research studies'. What is so revolutionary about this new approach to medical research?

The goal of this approach is to incorporate new technologies into clinical research and also, in the future, to apply these technologies to patient monitoring as part of standard clinical practice. One aim is to identify patient groups or profiles that can serve as a basis to adapt the treatment of individuals with chronic diseases.

We are also trying to develop the future of research, with vast quantities of data collected thanks to new technologies, for example via patients' smartphones. These methods fall within the realm of big data and artificial intelligence and are applied to address specific clinical needs – that is our aim in the department.

What technologies are you developing in this area?

Within the department, we are working on smartphone applications and connected medical devices. We are also working on two digital information sources that may help us to adapt our treatments.

The first is social media posts by patients themselves. By gathering these data – we currently have access to around a billion data points from social media –, we are able to identify the main concerns of people suffering from chronic diseases like diabetes or cancer. In a traditional clinical research context, we would not have access to this information.

The second source of information comes from our research on vocal biomarkers. In other words, we are exploring subtle changes to people's voices. These changes, which are not always perceived by the human ear, are signals that can help to diagnose an individual's state of health using artificial intelligence. Tiny variations in someone's voice can indicate whether the person is ill or not.

Can you give us some examples of specific applications?

In the field of oncology, mental health problems such as fatigue, stress, anxiety and depression are very common during cancer treatment, especially chemotherapy, and we can monitor these by analysing the voice. In the long term this should help us offer more personalised treatment. This is our vision as we try to develop our work.

When it comes to people with diabetes, we can use a screening tool to detect diabetes from a person's voice, for example via an audio recording. We have already tested this approach for COVID-19 detection.


Dr Guy Fagherazzi is Director of the Department of Precision Health at the Luxembourg Institute of Health (LIH) and Head of the Deep Digital Phenotyping Research Unit. He trained as a mathematician and statistician and is an expert in cohort studies. He leads the Predi-COVID study, which monitors thousands of patients affected by the virus during the pandemic. He champions the use of digital tools like artificial intelligence and big data in the field of medicine, especially in diabetes research. In this context, he is a keen advocate of precision health.

How do you extract this data?

We mainly work with public data from social media platforms like Twitter, via a user licence that you can sign with Twitter. This offers access to data for marketing and promotional purposes and also for the purposes of health research. For the past few years we have been using a data collection tool for Twitter that enables us to enter keywords of interest to us about specific diseases that we want to study.

Based on this data, we are establishing a global online health observatory where we use aggregated monitoring to identify messages from our target population and adapt our research priorities. We don't try to identify individuals; instead we try to identify signals.

In what ways are these 'new' technologies necessary for your work as Research Leader of the research unit?

The research team bases its work on the development of these technologies to gather relevant data. One of the aims is to develop screening tools to facilitate the detection of diseases at an early stage, unlike invasive blood tests. In some cases, a few seconds of voice recording are enough to provide a first indication of risk. For people with a chronic disease, the idea is to make it easier to monitor the disease on a daily basis (stress, depression, anxiety, pain, fatigue), while personalising treatment based on an objective vocal biomarker.

What are the specialisations of the researchers working with you on these projects?

The team is very multidisciplinary, with a strong focus on new technologies for health. We have doctors, epidemiologists, data scientists and bioinformaticians in the team, and we also plan to incorporate aspects of psychology and sociology to help guide us in how best to integrate these technologies in daily practice. The multidisciplinary aspect brings real added value. Rather than having a very specific, narrow view, we try to approach the question from several complementary angles.

Do you find the talent you need in Luxembourg?

Generally, yes. Luxembourg has the advantage of good working conditions and more importantly a very dynamic research environment.

The country is not yet necessarily very well known at European level, but people are often surprised by the quality, relevance and dynamism of the Luxembourg research environment when they first come into contact with it. That's what happened to me.

And what was it that attracted you to Luxembourg personally?

What attracted me personally was the fact that the research priorities were completely in line with what I wanted to develop. The alignment between the national research priorities on precision medicine, digital health – Luxembourg is strongly committed to these areas.

Deep Digital Phenotyping (DDP)

The Deep Digital Phenotyping team uses digital data, often from smartphones, to understand and analyse behaviour and health. Applications include detecting disease outbreaks and monitoring chronic diseases and mental health.

Your own career path took you from your initial studies in mathematics and statistics to medical research. How did you reach this point?

I have always loved maths, so much so that I wanted to pursue a career in numbers, in modelling. During my specialisation in biostatistics, I decided to focus on medical applications in the field of health. I then continued with a Master's in Epidemiology, a field that combines a statistics/mathematics approach with specific public health problems.

What I find fascinating is precisely this ability to build bridges between different disciplines by bringing together experts in different areas. My hope is that by working together with medical professionals, I can bring an understanding of medical issues and combine this with aspects of data modelling and research.

COVID-19 must have been the ideal opportunity to put all the theories to the test.

COVID-19 was a big shock for the entire world, but for research it was also a catalyst and an accelerator. At the LIH we immediately organised our efforts under the umbrella of Research Luxembourg and we developed studies at an incredible pace with the resources that were very quickly made available to us. I had never witnessed that in my life before. In the space of a few weeks, we set up a large-scale cohort study, Predi-COVID, in which we collected data and samples to get a better idea of the profiles of people with COVID-19.

We were also able to collect voice samples from study participants, and this represented a huge source of very valuable, highly relevant information that boosted research on this disease.

Now, with all that COVID-19 has taught us, we can transfer everything that we have learned about methodology to other chronic diseases like diabetes, cancer and mental health disorders.

How did the LIH support you during this period?

At the very beginning of the pandemic, we grouped ourselves into a Task Force in which all the research institutes and the University of Luxembourg worked in coordination to develop different research projects. We were given the task of developing the cohort study and monitoring the development of patients and individuals with COVID-19. I was personally in charge of setting up the Predi-COVID study, which is still going today. We are now following a thousand people two years after they were infected with SARS-CoV-2.

At the time, we put other research projects on hold, and for a short but intense period we all concentrated on setting up these research projects. We received financial support from the FNR and the Fondation André Losch, which enabled us to develop the research and gave us the resources we needed to make full use of the data we had collected.

What is special about Luxembourg's research environment?

The dynamism, the research priorities and the commitment to developing digital technologies. Luxembourg has always had a history of positioning itself either in niche areas or in a few promising growth areas – and digital technologies are clearly one such area –, and that is ideal for researchers like me. We have broad support to develop innovative research in these fields.

Is there one achievement that you are particularly proud of?

The Predi-COVID study, which is the result of a remarkable team effort. All the LIH teams stepped up and we worked day and night. Even during the lockdown, when we had to work from home, we managed to set everything up in a very short space of time.

On a personal level it is very satisfying because it is the first cohort study that illustrates our philosophy of precision health, with a data collection that offers a general overview of the participants. This study enabled us to realise our vision, which we are now successfully applying to other contexts.

What are your plans for your future career?

I am currently at a stage where I need to show that my ideas and my research are going to have an impact. My aim is to bring vocal biomarkers to clinical practice so that we can demonstrate their potential and implement them with clinical research projects.

We are internationally recognised in this field, with international collaborations that are gaining pace and supervision for PhD students on the topic. In five years' time, I see myself continuing to develop this area, helping the department to grow and developing more innovative projects.

Is there a side of Luxembourg that you are less keen on?

Like all cross-border workers, working from home is still complicated and traffic jams are an issue. But you get used to it. In terms of public health, one thing that is regrettable is that there is not yet a Public Health Act in Luxembourg that would facilitate access to existing data. We are sitting on a gold mine of data, but there is still no legal framework to make the most of this wealth of material. A law is under discussion, and we are all eagerly waiting for positive developments that will facilitate the use of data for research purposes.

Which three words would you use to characterise Luxembourgers?

When I arrived in Luxembourg, I was struck by the openness of the Luxembourg people. When you arrive here from another country, you realise that Luxembourg is European by definition – you need only look at its languages, culture and school system.

Luxembourgers are courteous and polite – maybe more so than some French people!

And since I have had the opportunity to work with colleagues in other institutions, I can say that they are very dynamic and forward-looking. There is a positivity in Luxembourg, a feeling that we have the resources, we have a strategy – how can we improve the situation? And that's something that is very gratifying.

Dr Fagherazzi, thank you for taking the time to talk to us.

This interview has been edited for the purposes of the article.