The Reference Model describes chronic disease complications in a population.
• An ensemble model assembling multiple other models
• A league of disease models that compete and cooperate
• A validation model
• A medical knowledge accumulator
The Reference Model is now:
• COVID-19 model for US states and territories
• The Most Validated Cardiovascular (CVD) Diabetes Model known
• United States Patent 9,858,390
The Reference Model can now:
• Estimate COVID-19 parameters for US states and territories
• Determine CVD models that significantly behave better on several diabetic populations
• Deduce that CVD probability halves every 5 years due to medicine improving - according to information from the last 3 decades
• Calculate life tables for diabetics
• Interface with ClinicalTrials.Gov
• Include human interpretation in the model
• Create an interactive map of our cumulative computational knowledge gap
The interactive plot below shows our cumulative knowledge gap by showing the error in the vertical axis for US states and territories listed on the horizontal axis. Circles at the bottom have a better fit between observed COVID-19 results and model results. Results are for normalized population of size 10,000 individuals. Hover over the circles to see additional details about each state. The slider determines the model optimization iteration. User can explore the map by changing size and color attributes.
The Reference Model is a good way to cross reference information to find out pieces of information and assumptions that fit together, and allow competition against accumulated known data to guide our perception. High Performance Computing is a key to those capabilities and it provided using capabilities of the MIcro Simulation Tool (MIST) .
MIST also provides advance population generation techniques using Evolutionary computation. The Reference Model uses publicly available data such as clinical trial publications. This allows it to access more information since it allows accessing data that otherwise will be restricted from sharing. The Reference Model has an interface that allows it to read information from ClinicalTrials.Gov while maintaining tractability and reproducibility.
The Reference Model was created in 2012 and evolved since then. You can find key developments and publications by year in the news section .
Here are some videos describing the Model:
This video explains the model in a larger context as presented in AnacondaCon 2019:
This video summarizes a decade of work as presented in PyTexas 2017:
This describes the evolution of the model up to 2016 presented in PyTexas:
This describes the work presented in PyData in 2014: