Software
The project has so far produced the following software components:
- The Indirect Estimation and Simulation Tool (IEST) using Python: General disease modeling software with a graphic user interface that provides simulation capabilities.
- Indirect Estimation Prototype for Matlab: a prototype that implements the Lemonade method for model parameter estimation. (Note: we are planning to convert this into Python so that is can be integrated into IEST.)
These software components are available for download. Below is a full description of the software components as well as links for their download.
The Indirect Estimation and Simulation Tool (IEST)
IEST is a general software application designed to model the progression of chronic diseases. The software allows modeling multiple disease processes in parallel based on a Markov model. The user defines states, transition probabilities, model variations, parameters, population data and many other details to specify the model and sample and then simulates the disease process for a number of years.
Currently only simulation is supported. Future versions intend to incorporate the estimation capabilities provided by the Lemonade Method.
Here are highlights of current capabilities of the software:
- Series of states: The model is specified as series of states. Each series is a sequence (usually defined by severity); each state may be a stage in a chronic disease or an instantaneous event such as Myocardial Infarction (MI).
- Multiple nested parallel series:Several disease processes (series) can be specified in parallel, such as complications or comorbidities. Such disease processes can also be nested within each other. This allows modeling multiple diseases and mortality from other causes in a conceptually easy manner. It is best to first describe the model as a diagram as shown in the diabetes example for the Chronic Disease Model.
- Transition probabilities as functions: The transition from state to state is specified by a probability; the probability can be a constant or described by a mathematical function of individual characteristics, of disease states, or of other parameters such as intervention parameters. These functions can use general mathematical functions such as "Exp", or "Log", or conditional functions involving "If" statements. or multi-dimensional tables.
- Monte-Carlo Simulation: The system can perform Monte-Carlo simulation of a model using a predefined population at baseline. The simulation can include pre-processing and post-processing stages that allow changing individual characteristics, intervention parameters, and cost parameters. The user can specify costs and health utilities for each stage of disease or event and the number of years to be modeled.
- Output: The user can control the amount of output provided by the simulation. The output can include the number of subjects who enter or pass through a stage of disease, the number who die and cause of death, the total cost and average health utility.
- Limited Version control: The system supports the storage of multiple versions of a model and of different population sets. This allows running the same simulation over different combinations of models and baseline population sets. Files saved by the system are backed up with a time stamp to allow rollback to a previously saved version.
- Import and Export: Simulation results and population set data can be exported using Comma Separated Values (CSV) format. Population data can be imported using the same format. This allows manipulation of system data in applications such as a spreadsheet or a database.
- Graphic User Interface (GUI) and Report Generator:The system has a dedicated GUI composed of forms. It allows users to define and modify states, transitions, models, population data and other entities. It contains wizards to help define cost and quality of life parameters. Finally it allows generating reports for system entities and specialized reports aid in analyzing simulation results. Such reports can help answer questions such as what is the mortality from a specific cause in the first year? in year 2? in years 1-5? What is the average yearly cost? Etc.
A set of examples is provided with the software to demonstrate its capabilities as well as a developer guide that explains the system from a technical point of view. Further information about the software can be found in its help system that is provided online for reference. To access the help system, please click here.
The software works in Python environment. For specifics regarding setup, read the README.txt file supplied in the zip archive or click on this link.
The software prototype is available for download using the link below under the GNU General Public License (GNU GPL).
A recent version of the Michigan Model for Diabetes was modeled using the IEST software. It can be downloaded from the model section of this web site.
Indirect Estimation Prototype for Matlab
This software implements the Lemonade Method. It enables estimating Markov model coefficients given a set of studies with sufficient data observations and associated population data.
The software works in Matlab environment. For specifics, read the README.txt file supplied in the zip archive. A set of examples is provided with the software to demonstrate its capabilities.
The software prototype is available for download using the link below under the GNU General Public License (GNU GPL).
Download the Indirect Estimation Prototype for Matlab (Version 0.38a - 29-Apr-2009)
Older versions of this software have been previously released and are no longer maintained.
