By Jacob Barhak, PhD (https://sites.google.com/site/jacobbarhak/)
The Reference Model for disease progression is used to compare existing models that forecast chronic disease progression. When given a population, the model attempts to predict the number of complications and morbidities such as stroke, MI, or number of deaths that will occur within a given number of years. Given additional local specific information, the model can predict costs and quality of life.
The Reference Model is not a single model; instead it is a league of models that compete amongst themselves to gain the best fit to known clinical trial results. In a sense, this is similar to asking the opinions of multiple specialists about a disease and figuring out who is the best specialist for which conditions. The Reference Model is therefore more than a forecast tool, it is a tool that allows us to gage our understanding of disease progression by comparing multiple models and populations.
The Reference Model for disease progression has rapidly advanced since its first launch in 2012.
The first important advancement is visually showing model and population fitness information in a color coded matrix form. This fitness matrix allows comparing the behavior of multiple models with multiple populations. This is actually the main idea behind the model.
Another important advancement was demonstrating that disease models become outdated and need correction/update for treatment improvement in time. This shows the necessity in a new generations of models that can cope with recent and future changes.
Another advance is demonstrating the uncertainty associated with unknown correlations of biomarkers. This is important since it shows how much model results can be believed considering hidden unknowns/assumptions in the population inputs. After all the inputs for the model are all publicly available data. More specifically, the populations are reconstructed from summary data of clinical trial results that are freely available in the literature. Using publicly available data increases the information base of the model.
The Reference Model is not a perfect solution, there is still a lot of knowledge missing and many assumptions are made. Yet it is a good way to cross reference information to find out pieces of information and assumptions that fit together, and allow competition against known accumulated data to guide our perception. This is preferable in many cases to expert opinion since using the model the opinion can cross referenced and tested against evidence.
High Performance Computing is used to cross reference all this information. The MIcro Simulation Tool (MIST) is free software used to implement the model and run it. MIST can run over the cloud! More information about MIST in my next post. For more information or questions, please contact me at https://sites.google.com/site/jacobbarhak/.