by Rahul Jayaram, '21
If you have ever played a racing video game, that experience probably formed your first understanding of what driving a vehicle actually feels like. In fact, you probably developed a better intuition for the basic techniques of driving, such as accelerating, braking, and steering, from playing a game than you would have from observing another person drive. Despite not driving an actual car, the experience obtained from a simulation was meaningful. This scenario parallels the concept of neural modeling, or the use of mathematical representations of different brain aspects to better understand how our own brains work. Modeling uses equations to describe the changing relationship between different factors, in this case those that underlie human brain functionality, and can provide a functional understanding of the dynamic processes in our body.
Currently, dissected brain tissue and neuroimaging are our primary scope of vision into understanding the mechanisms behind neurodegenerative diseases. However, these are very static representations. There are in fact many other factors that must be considered if we want to truly understand how an active brain works, which cannot be easily provided by standard imaging.  For instance, misfolded proteins are responsible for many diseases that are universally known to be deleterious, yet there is wide debate about their exact kinetic mechanisms and structural properties with regards to a disease. One application of modeling would be to simulate these reactions and test how different molecules react within a certain context. Equations can be derived based on molecular structure and kinetic rate laws (how the molecule interacts with the neuron and how quickly it does so). The goal would be to run simulated reactions between different molecules using a model defined by the properties of molecular reactions.  This could lead to strong predictions into the results of different molecular reactions, providing insight into theoretical reactions that take place in neurodegenerative diseases. Modeling would also be able to map the macro effect of these proteins on the brain overall and provide predictions as to how a disease spreads throughout the brain over time.
The usefulness of modeling can be seen in the case of Alzheimer's disease, where a model was able to compare the effectiveness that of three potential approaches to lower amyloid protein accumulation. This protein is responsible for the disease, yet its exact molecular interactions with the brain remain elusive. The model demonstrated that of three drug based approaches that affect the amyloid protein structure, only one approach would substantially lower the amyloid protein levels, while the other’s would have been inefficient or even increase amyloid protein aggregation.  Since this was all computational, discovering that certain treatments are harmful is not ethically questionable. Yet if this same question was tested in a clinical trial, where subjects would have directly faced the effects of the treatments, this would not be the case.
In the future, instead of solely relying on sacrificed tissue or limited conjecture as a means of research, we will use these techniques as a complement to the knowledge brought by modeling. Modeling can serve as a primary tool to best predict what may happen and eliminate red flags before confirming a treatment through clinically based “real-life” experiments. Overall, modeling is playing a larger role in not just neuroscience, but also many other fields. Economics, engineering, astrophysics, psychology, and other disciplines are beginning to adapt the idea of using models to optimize effort, time and resources and to better understand underlying dynamic processes.
 Carbonell F, Iturria-Medina Y, Evans AC, Mathematical Modeling of Protein Misfolding Mechanisms in Neurological Diseases: A Historical Overview. Frontiers in Neurology. 2018; 9
 Masel J, Jansen VAA. Designing drugs to stop the formation of prion aggregates and other amyloids. Biophys Chem. 2000; 88(1-3).