The EP-Predict Calculator engine is a set of components than can be used by an application to provide predictions according to a set of personal characteristics passed into the engine in order to obtain one or more predictions, usually a score or a likelihood of a condition.
The engine comes in 3 forms
- An internet facing REST API.
- Embedded software components in various languages such as C#.Net DLLs or Java Jars.
- A Batch processor executable application that can process millions of patients each patient having their characteristics containing the input model in the form of of a CSV file.
EP-Predict as a REST API
Information describing the API and a development version is available at https://dev.endeavourpredict.org/
This is a self registration portal where developers can register and subscribe to the EP-Predict API. Examples of the usage and the input model format can be seen and the API may be used to aid in development of a working application.
EP-Predict as a C# DLL or Java Jar
These are available on request from Endeavour Predict email : enquiries@endeavourhealth.org.
It is recommended that before requesting the integrated DLL or Jar that the technical team become familiar with the REST API approach as the inputs to the components are the same in all cases.
Using the EP-Predict DLL or Jar consists of calling the components main method passing in the input model in the same form as the API (i.e. an object in C# or Java)
EP-Predict batch executable
The batch processor will process a folder of CSV files and compare the calculated outputs with the associated JSON Output files. Example files are provided in the Example data folder of the release pack.
The CSV files must have a header row with columns representing the input parameters as shown in the example
Open a command prompt and call the EXE passing in the path to your input files, for example:
EP_QRisk3.Batch.exe .\TestPack
The batch processor will try and process any file in the folder with the “input.csv” extension.
Any file that does not have this extension will be skipped.
EP-Predict calculator input model
The main article EP-Predict calculator input model describes the model's content and in addition describes the health record query rules for obtaining default values.
All algorithms use the same model but different algorithms use different variables from the model. For example QRISK3 uses 20 or so.