European start-up Exactcure has developed a software solution to reduce the impact of inaccurately dosed medications and help patients avoid underdoses, overdoses and drug-drug interactions.
Their Digital Twin solution simulates the efficacy and interactions of medications in the body of a patient based on their personal characteristics such as age, gender, kidney status, genotype and any other individual parameters. With the Covid-19 pandemic, they made digital twins of molecules such as lopinavir-ritonavir, hydroxychloroquine and chloroquine. We spoke with Fabien Astic, Co-Founder, Business Development at Exactcure.
MedicalExpo e-magazine: The digital twin concept started in the industrial sector first. Now it’s being increasingly applied to the medical sector. What does this mean exactly?
Fabien Astic: In the industrial sector, we see digital twins of machines, equipment or an entire factory. They are made before starting high-scale and high-speed production. These virtual twins allow you to visualize products or evaluate production decisions.
For the medical sector, it is not exactly the same approach. We’re not going to see a virtual human model appear like in Star Wars anytime soon. As far as Exactcure is concerned, we are talking about virtual models of molecules, i.e. the active ingredients of a drug. Where it can become a patient’s digital twin is when we integrate all the personal parameters of this patient into our drug models in order to better adapt the dosage to the characteristics of each patient. So the drug models are weighted by the patient’s parameters.
ME e-mag: Can you explain to us how your system works and how you managed to apply it to the Covid-19 pandemic?
Fabien Astic: A public hospital in France came to us in March to find out if we could provide personalized models of several molecules that were being considered to fight Covid-19 in order to work on the dosage of these drugs. These molecules were lopinavir-ritonavir, hydroxychloroquine and chloroquine (Please note that on May 27, France banned the use of hydroxychloroquine for the treatment of Covid-19). We are currently working on other molecules to fight the pandemic—we are targeting about 15 of them.
We delivered a simulator to this hospital. How does it work? Through a browser, you arrive on a web page where you enter the drug you want to simulate, as well as the age of the patient you want to treat, their weight, the doses you want to give them and the times of dose intakes. You click on “simulate” and the system outputs a personalized curve, specific to this patient. You can add other parameters such as sex, renal status, liver status, we are also currently working on the genotype—but this is not for Covid-19.
The curve that comes out is a pharmacokinetic and pharmacodynamic curve of the molecule in question. It shows the evolution of the blood and plasma concentration of the molecule over time. It goes up when the patient takes a dose, then it goes down, then it goes up again when they take a second dose, and so on. It goes up and down more or less quickly depending on the parameters you entered. The goal is to visualize whether the patient is well within the therapeutic window, either up or down. This is really an application for hospitals, healthcare professionals, doctors and pharmacologists.
ME e-mag: You are also working on a more consumer-oriented application of your system?
Fabien Astic: Yes. For example, we have developed a paracetamol simulator available online (www.exactcure.com/simulation-paracetamol) where patients can enter their own parameters such as gender, age, weight and liver status. They will be able to see if they are taking too much paracetamol or not enough, or if they are well within the therapeutic window. We also provide a mobile app covering many other drugs. A clock will be displayed on the patient’s phone: if it turns red around the clock at a given time you reach an overdose or an underdose, if it’s green it’s good. It’s very visual and simple. It also tells the patient if another non-prescription drug interacts with their basic treatment. The patient will see that it’s red around the clock. If they want we can even send an automatic alert to their healthcare professional. Ideally they can do their simulation before taking their medication, and avoid a mistake if it turns red.
The objective is to inform people. Because the numbers are terrible. In France, iatrogenic drug complications (misuse of medication, inappropriate prescription, etc.) lead to five times more deaths than road accidents every year. And this phenomenon costs health insurance companies 10 billion euros a year. That’s the equivalent of more than a million hospital days a year.
ME e-mag: This is where personalized medicine is heading?
Fabien Astic: Yes. We are not replacing healthcare professionals, we are offering a complementary tool, a personalized simulation tool that they will use to better adapt the dosage to the needs of each patient. And thus converge more quickly on the right dose without sometimes losing weeks. Healthcare professionals remain responsible for their decisions. From the patients’ point of view, they now have a tool that allows them to visualize what is going on in their own body. Until now, this did not exist; it was like driving a car without a dashboard. In the long run, we would like to simulate models of almost all the drugs on the market.
ME e-mag: Why hasn’t this been done before?
Fabien Astic: The reason is partly technological because today everyone has a smartphone so the technology is available in everyone’s pocket, we can put personalized medicine in everyone’s hand. And processing power has become a commodity.
ME e-mag: What is the technology behind your system?
Fabien Astic: To make a priori simulations we use mathematical models; we will take the weight, age and/or other parameters depending on the drug, and we will simulate and see what it looks like. The primary goal is to have a model for each patient in order to see how he or she reacts to a particular drug and adapt the dose. But behind that there is a second aspect that incorporates artificial intelligence or machine learning. We are going to take into account the patient’s feedback to further personalize our models and to get Real World Evidence. When we have a large number of users, our clients will be able to identify patterns on a population.