Guest Editors
Yixiang Deng
Ydeng9@mgh.harvard.edu;
I am currently a Postdoctoral Fellow, advised by Prof. Daniel Lingwood at Ragon Institute of MGH, MIT and Harvard, and co-advised by Prof. Douglas A Lauffenburger at MIT. I am also a member of the systems serology lab lead by Dr. Ryan Mcnamara. I was advised by Prof. Galit Alter before October 2022 (now the Vice President of Immunology Research at Moderna).
Prior to joining the Ragon Institute, I received my Ph.D. degree in School of Engineering at Brown University, advised by Prof. George Em Karniadakis and co-advised by Prof. Christos Mantzoros at BIDMC, in February 2022.
My current research interest is in developing physics-informed data-driven digital twins to probe the mechanisms of diseases and effectiveness of therapeutics by leveraging data acquired from bench-side to bed-side and existing biological/medical knowledge, for example, diabetes in endocrinology and infectious diseases in immunology.
Summary
Over the past decade, the rise of data-driven models has acted as a catalyst for groundbreaking discoveries spanning a myriad of research domains, from fundamental sciences to various branches of engineering. These models, built upon vast volumes of data, have redefined how we approach problems, fostering innovation and expanding the boundaries of our current knowledge. Furthermore, when these models are intertwined with existing domain knowledge, such as principles of physics, they evolve into potent tools that are far superior, blending the empirical strength of large-scale data with the rigour and predictability of scientific laws. This synergy, often referred to as physics-informed machine learning, has gained a lot of attention especially in areas like computational mechanics and fluid dynamics.
However, the potential of physics-informed machine learning remains untapped in several crucial areas, particularly in the realm of biophysics and biochemistry, which play pivotal roles in our understanding of various diseases that plague humanity. For instance, while infectious diseases, cancer, and neurodegenerative disorders like Alzheimer's have seen considerable research, the intricate biophysical and biochemical mechanisms that mediate these conditions are still not fully grasped. Similarly, haematological disorders, such as sickle cell anaemia, and cardiovascular diseases present other areas where the nuanced interplay between the physical properties of blood cells and their biological pathways can benefit from a more in-depth application of physics-informed models.
In essence, while the progress made so far with the integration of physics and machine learning is commendable, there remains a vast expanse of biomedical challenges that await exploration. Delving deeper into these areas with the combined might of data-driven models and foundational scientific principles could pave the way for unprecedented advancements in our understanding and treatment of various diseases. In this special issue, we are looking for articles focusing on developing computational models that combine biophysics, biochemistry or biomedical imaging with machine learning to understand the fundamental mechanisms in those biological processes. Priority will be given to those researches addressing critical diseases impacting global health, for example, Respiratory Diseases like COVID, HIV, Cardiovascular Diseases, Diabetes, Neurodegenerative Diseases, Hemorrhagic Fevers like ebola, and Vector-Borne Diseases like Zika.
Keywords
Deep learning; physics-informed machine learning; biomedical imaging; biomechanics; computational fluid dynamics; molecular dynamics; multiscale-modeling
Login
Register