All human beings desire physical and mental well-being. However, traditional, empirically oriented approaches to achieving and maintaining personal health have reached their limits. Instead, an individualized, biophysics-based, Digital Human Model with predictive power could potentially revolutionize our thinking about personalized healthcare. This would especially affect the understanding, diagnosis, and treatment of diseases and injuries. It would also pave the way to identifying and curing diseases with a pervasive socioeconomic footprint – including cardiovascular diseases, disabilities caused by neuromusculoskeletal defects, cancer, or rare diseases affecting smaller populations. Furthermore, a Digital Human Model could be a key to developing individually tailored technologies that support an aging population, creating ergonomically optimized work and life spaces, and inaugurating a new generation of safety concepts
Achieving in silico predictions for complex multiscale organs by shifting simulation technology toward an integrative systems science has been a driving force for EXC 310. It produced a broad range of different simulation methods that make integrating physiological and structural properties, from cellular levels to multibody biomechanical systems, feasible. However, for a Digital Human Model we need highly complex data-integrated models covering length scales ranging from nanometers to meters and time scales spanning picoseconds to years.
To create a holistic understanding, for example, of a neuromuscular disease and how it affects movement, observed motion data alone does not suffice. The entire sensory feedback system controlling the interaction of molecular, cellular, and organ system processes must also be identified. This calls for new techniques capable of integrating multiscale models with data-rich (e.g., omics data, motion tracks, electrical activity) and data-poor (e.g., in vivo muscle fiber type distributions) phenomena. Simulations will significantly enrich data-poor scales and domains as a prerequisite for utilizing advanced machine learning techniques on complex data sets in HPC environments. In addition, there is a strong need to incorporate patient-specific data into simulation processes through novel homogenization techniques to clear the way for computer-assisted diagnosis and clinical decision making. The quality of simulations in a real-time (that is, clinical) environment is limited by time restrictions, low computational resources, and even missing data access due to data protection laws. Therefore, it is imperative to innovate model reduction techniques and self-adaptive numerical schemes suited for optimally exploiting the available hardware resources. By keeping the human in the loop through integrating cognitive models, utilizing mobile devices, and running simulations in real-time, pervasive simulation will allow us to reach for new horizons. These could include training on how to actively control and naturally operate (neuro-)prostheses or prevent lower back pain by subtly influencing the person’s environment. Moreover, this technology allows the interaction with bio-inspired robotics such as humanoid robots, personalized exoskeletons in the workplace, or biorobotic devices for guiding natural movements in rehabilitation settings. Research in the Cluster is aiming to provide the methodological basics for a Digital Human Model that enables delivery of the personalized healthcare of the future as contemplated here.