Digital Twins for Predictive, Preventive Personalized, and Participatory Treatment of Immune-Mediated Diseases

Abstract

Digital twins are computational models of complex systems, which aim to understand and optimize those systems more effectively than would be possible in real life. Ideally, digital twins can be translated to individual patients, to characterize and computationally treat their diseases with thousands of drugs, to select the drug or drugs that cure the patients. The background problem is that many patients do not respond adequately to drug treatment. This problem reflects a wide gap between the complexity of diseases and clinical practice. Each disease may involve altered interactions between thousands of genes that vary between different cell types in different organs. To our knowledge, these altered interactions have not been characterized on a genome-, cellulome-, and organ-wide scale in any disease. Thus, clinical translation of the digital twin ideal for predictive, preventive, personalized and participatory treatment involves formidable challenges, which are close to the limits of, or beyond today’s technologies. Here, I discuss recent developments and challenges in relation to that ideal focusing on immune-mediated inflammatory diseases, as well as examples from other diseases.

An interactive atlas of genomic, proteomic, and metabolomic biomarkers promotes the potential of proteins to predict complex diseases

“…The clinical relevance lies in that prioritisation of biomarkers is complicated by each disease involving thousands of genes and gene products that can vary between the same patient before and after diagnosis, as well as between patients with the same diagnosis. Recent organome-, cellulome and genomewide studies show that the same complex diseases can involve variable cellular and molecular changes across multiple organs, and not only in the organ that shows symptoms or signs of disease 17 , 18 . Moreover, those changes can vary greatly between patients with the same diagnosis 19 .…”

AutoCore: A network-based definition of the core module of human autoimmunity and autoinflammation

“…Our work contributes to ongoing efforts to move toward a more unbiased, de novo definition of molecular mechanisms that is purely based on connectivity patterns ( 73 ). Another very promising route toward such a data-driven definition of molecular disease mechanisms are digital twin models ( 74 , 75 ), showing that multicellular network modules based on gene expression data can help identify biomarkers and putative drug targets for a number of inflammatory diseases. These findings are similar to our observation that the AutoCore subnetwork is enriched with markers of polygenic autoimmune and autoinflammatory disorders and can be used to identify endotype-specific treatments.…”

The Digital Twin: A Potential Solution for the Personalized Diagnosis and Treatment of Musculoskeletal System Diseases

“…DT is data-centric and various complex substructures are related to each other through technologies such as machine learning and AI to finally form a DT system [ 37 ]. At present, the DT system has made certain progress in the prevention and treatment of cardiovascular, musculoskeletal, and immune system diseases [ 38 , 39 ]. It has also made some breakthroughs in imaging diagnosis and radiotherapy [ 40 , 41 ].…”