This article is based on the interview led by Hazel Tang for AIMed. See original here.
AIMed: For the benefit of readers who do not know you, could you please tell us more about yourself?
Marc Raynaud: I have a background in pure mathematics and I am doing a PhD in biostatistics with the Paris Transplant Group (PTG) now. PTG is a think-tank which gathers a lot of different people, including nephrologists, cardiologists, epidemiologists, immunologists, statisticians and mathematicians.
Our main goal is to go beyond the boundaries between specialties and bring upon new and original perspectives to medical research. In my case, I have daily contacts with doctors and it has given me important insights that directly benefitted my work that is focused on kidney transplantation or more precisely, on risk predictions in kidney transplant.
AIMed: What inspired you to formulate the abstract idea?
Raynaud: As some of us might know, kidney transplant is a treatment of choice for people with renal failure. In past decades, there’s no improvement in the long-term graft survival in kidney transplantation. Most graft survival prediction systems were developed to improve risk stratification and patient monitoring but most of them come with limitations. Primarily, they have not been validated in different transplant centers; countries; continents, with distinct allocation systems and patient management. Importantly, none was validated in randomized controlled trials.
Furthermore, these systems do not integrate the full spectrum of parameters that predict graft survival and none of them are dynamic, meaning they do not take into considerations the parameters that are repeatedly assessed throughout patients’ follow-up. Because of all these reasons, these graft survival prediction systems have difficulties to move to a clinically implementable strategy.
As such, my team at the PTG and I aimed to address all these challenges in a rather ambitious study. We have developed a robust, reliable and extensively validated graft prediction system driven by AI and computation algorithms. As of now, we have already collaborated with 20 transplant centers from eight countries. From there, we also collated a unique, multi-dimensional dataset in six randomized controlled trials, which gave us an unprecedented and insightful phenotyping of every patient.
AIMed: What does winning the Abstract Competition mean to you and your proposed idea?
Raynaud: Being the winner means a lot to me and my team. This project is a long-standing effort started 10 years ago. We are honored to receive a recognition from the conference.
Given the excellent prediction performances of the model we have created, the next steps would be to implement this dynamic prediction system in clinical practice. We believe this will benefit both patients and clinicians and be of help to design the next generation of clinical trials involving kidney transplantation.
AIMed: How did you know about AIMed and what prompted you to take part in the Abstract Competition?
Raynaud: My team and I knew about AIMed through social media; it’s a prominent event which gathers a high number of talented researchers from around the world. Considering that we have a strong interest in AI, we thought this conference as an amazing opportunity to present our project.
All along, we believe our study is a proof of concept showing that AI and medicine can merge, not only via research but also from a human perspective and together they can yield a concrete and useful prediction model. We also believe that the model we have generated for kidney transplant is generalizable to other medical specialties with a need to improve risk stratification and patient monitoring.
AIMed: Now that you have won the competition, how do you intend to bring your idea further?
Raynaud: Of course, my team and I are working on refining the model but in the meantime, we also wish to have more transplant centers to come on board. As mentioned, our ultimate goal is to translate the model into actual clinical practice as we strongly believe this system will benefit clinicians and patients in their decision-making process.
AIMed: What is the greatest lesson that you have learnt from all these?
Raynaud: I think it will be to always keep informed of the advances in AI because some of the medical specialties like radiology, are moving in a speedy manner. So, it can be challenging to keep track of everything that is going on right now. Hence, the main lesson will be to maintain one’s passion, be proactive and keep a head’s up of the progress as much as possible.
Paris Transplant Group
Our global aim is to accelerate the translation of immunological and gene expression discoveries into the clinical field by filling the gap between basic science and applied biomedical researches.