Educational content on VJHemOnc is intended for healthcare professionals only. By visiting this website and accessing this information you confirm that you are a healthcare professional.

Share this video  

COMy 2022 | The role of artificial intelligence for precision medicine in multiple myeloma

In this video, Alberto Romagnoni, PhD, Owkin, Paris, France, discusses the role of artificial intelligence (AI) and machine learning in multiple myeloma and other hematological malignancies. Dr Romagnoni first highlights the importance of precision medicine in hematological oncology, and further explains how AI can help when working with rich and complex datasets. Dr Romagnoni then discusses a novel approach in AI and machine learning and the possibility of applying this approach to other modalities. This interview took place at the 8th World Congress on Controversies in Multiple Myeloma (COMy) 2022, held in Paris, France.

Transcript (edited for clarity)

Precision medicine is really important right now, because all our research in oncology is increasingly focusing, on this attitude where you try to identify the precise tumor, genetic subtype and immune environment. And then, all this heterogeneity that you have in cancer in general, this is extremely relevant in multiple myeloma. Because you have this heterogeneity at all levels, at the level of interpatient, intrapatient also, it’s a dynamic heterogeneity during, throughout all the different stages of the disease...

Precision medicine is really important right now, because all our research in oncology is increasingly focusing, on this attitude where you try to identify the precise tumor, genetic subtype and immune environment. And then, all this heterogeneity that you have in cancer in general, this is extremely relevant in multiple myeloma. Because you have this heterogeneity at all levels, at the level of interpatient, intrapatient also, it’s a dynamic heterogeneity during, throughout all the different stages of the disease. So, all this complexity, it’s complicated to deal with it. And artificial intelligence can help that, because it’s the real idea of machine learning. So, use very rich data, to explore very complex problems. The problem, in most of the cases is that, even if the modalities are very rich, so there is the information in the dataset, but the datasets are very sparse, especially, for very complex problems.

Now, the true issue is to combine datasets large enough to solve the problem. And typically, this is possible when the data are not sensitive or when there are no privacy or regulation reasons, but it’s more an issue in health science. In Owkin, we provide a different kind of solution, where we leverage federated learning technology, where the data do not travel, they stay in the centers. So providing security, and it’s the model that travels around the different datasets. So different models learn on different datasets, in different centers and then, the training is orchestrated and by a central model, which then can combine the information while keeping the privacy and the data where they are.

Actually, what happens for multiple myeloma happens for all other indications and especially, the possibility to explore different modalities. Also, histopathology, it’s a very rich feature so, a very rich modality of data, where machine learning and artificial intelligence can do a lot because, years and years of computer vision has helped us, adapted us to explore this kind of data. And then, several multimodal models can explore how to interact with all these modalities. Definitely, that’s a very rich field of research.

Read more...

Disclosures

Computational Biology Team Lead at Owkin