Digital drugs, diagnosis, disease courses prediction: the unique opportunities offered by digital health solutions
Interview with Prof. Thomas Hügle, Director of the Rheumatology Department, CHUV
You are a strong believer in the healthcare digitalization. Why?
I believe healthcare digitalization will bring us in rheumatology to a second revolution. The first revolution took place around 15-20 years ago when biologics were discovered: a new form of treatment that completely changed patient care by strongly reducing deformation and disability caused by arthritis. Today we have other clinical needs to address: providing the right patient with the right treatment at the right time is one of them. I think this is where digital solutions and artificial intelligence will come in and deeply improve patient care.
In your opinion, which are the main healthcare areas that could benefit the most from the digitalization?
Digital drugs will start as self-management tools
helping patients affected by chronic diseases to better manage their disease and treatment autonomously. This is already happening in depression, for example, with available FDA-approved applications. Artificial intelligence will then come in in different ways, such as in diagnostics through the disease clustering
: what we currently identify as a single disease, like in fibromyalgia, arthritis, heart failure or lupus – in fact they include multiple subtypes into which we cannot properly cluster or classify patients yet. AI, with the unsupervised learning, will help us in this sense, telling us to which drug each patient could better respond.
Nowadays, people are still a bit afraid of AI, but I think it is not, I think it is a big opportunity to empower both patients and doctors.
In the end AI is learning on a large scale: we have huge databases whose proper analysis and interpretation could help us a lot. So why should we just rely on doctors’ experience, on cases that we may remember or not or on our gut feeling? Why not to use these new deep learning technologies to better learn from our experience?
Another field of AI activity is the prediction of disease courses
: as an example, we investigated a large cohort of ten thousand patients. For this cohort, we let the machine learning in a training set, we validate it in a test set and, in future, we’ll use the information to tell our patients how they will be in six months. You might ask why this is useful. First of all, in this way we provide patients with an additional “neutral” opinion regarding treatment decision. The machine will participate together with the doctor and the patient in the decision-making process. If the patient knows that in six months he will suffer from an active disease, he would be more prone to start or switch to the recommended treatment. If we are considering reducing or stopping a certain drug because the disease is quite stable and the machine confirms that this could be beneficial, we will be more confident in taking the right decision.
Another big issue we are facing in the healthcare sector is the shortage in nurses. As in the gastronomy sector where staff is not sufficient and digital applications are used for reservations and billing, the healthcare system could benefit from digital solutions to optimize consultations time, for example. Why measure blood pressure during a consultation when you can measure it at home? Or why fill patient reported outcomes during the consultation when patients could complete them at home and send them to the doctor in a digital format?
What is your perspective in terms of healthcare costs?
I see a massive opportunity to save costs. The truth is that right now we may do x-ray to a patient who has done them two weeks ago. With the implementation of the electronical medical record we could avoid this. Even though there are some threats, we need it. It’s a no-brainer, I believe. Together with this, if we have the right opportunity to find the good treatment earlier, patients will have to see the doctor less and will have shorter sick leaves. When we think about healthcare costs all these aspects should be taken into consideration.
Do you think both patients and doctors are ready for this?
I think it is very individual but in general yes, both patients and doctors are ready for it. It is mostly about user-friendliness
. If we develop something which is easy to handle, then people will use it. I personally see the willingness of patients in using health applications on their mobiles; maybe not 80 years old people, but the ones around 60-70 and younger will certainly use them.
We should not forget that there are also threats: we need to ensure that those applications are efficient; academic centers have to test them. Things right now in healthcare are not perfect; there are specialties such as rheumatology where you have to wait for three months to get an appointment or where doctors just do not know the right diagnosis or which is the right drug to prescribe. We desperately need help. Doctors often believe they are good enough, but they also check out for diagnoses in google, as patients do. Around 5% of google clicks are healthcare-related. Diagnostic tools such as ADAHealth are useful. I think that digital solutions will help getting more information and assistance throughout the patient pathway, from the diagnosis to the treatment and follow-up.
How do you think we could accelerate the integration of innovative technologies in the healthcare system?
The first thing is the adaptation of the legislation as recently done in Germany where the use of these new drugs has been allowed as well as their reimbursement by health insurances. Then, regulations should be adapted as to ensure the quality of these solutions: EMA and FDA are getting into it. And then the players of the field should be identified: Will Google, Apple, Pharma or a new sector be doing this? Increasing funding of startups and research projects (SNF, Innosuisse) could support the clinical testing of these solutions.
What about your research activity? What are you currently working on?
Our main focus is deep learning: we leverage a large national dataset of roughly 10000 patients to predict diseases activity and patient clusters. We also work on disease recognition so that the patient could measure disease activity using the mobile phone and we are also developing a self-management application for gout.
Notably, we are actively working with members of the Digital Health Hub
for most of our projects. The DH2 is a really important interface to boost the field; it brings together the right people and through the Digital Pulse
it provides a neutral perspective on promising projects. And Lausanne is the place to be. I was recently at Stanford to present our work and I expected them to be far advanced compared with us here. But it is not. We are in an excellent position and the DH is an excellent opportunity. Everyone around, including the Canton, the CHUV, the universities are all going in the same direction and this will make the difference in the field.
Interview conducted by Beatrice Volpe, Vivactis Suisse (February 2020)