Endeavour Predict (EP-Predict)
Endeavour Predict CIC provides software and services that help professionals and patients make decisions about health.
It uses openly published algorithms, created by by university researchers, and validated on large scale health data stores such as primary care and acute care data. The algorithms are designed by experts in the field of clinical epidemiology and data science, and are all openly published in peer reviewed articles in main stream health journals.
EP-Predict consists of:
- A set of hosted internet based REST APIs that respond to requests from authenticated servers hosted by authorised organisations
- A set of SDKs or local components for use embedded within applications, provided in a variety of languages. These can be used for individual predictions or large population risk stratification.
- A number of web applications that provide the facility of a user to enter factors required for prediction.
- Advisory concept sets or value sets for use in health record query for inferring the state of a patient in order to populate variable values used in the formulas
- A post code to Townsend look up that operates at the level of output area (more granular than the Lower Super output area)
- A query Library that holds machine and human readable models of health data query.
- A common health data model that has been shown to work for both primary and secondary care data at scale.
The algorithms are made available (though the NHS in the UK) as a set of services via Web APIs, or local run time versions in several languages
The algorithms are regularly reviewed and the versions updated to ensure that new knowledge is brought rapidly to decision making.
For the tool itself and each algorithm, there are a set of resources covering,
- Information on this wiki, including documentation of the health record query rules for auto completion of variable values.
- Links to sign up information for use of the APIs in test mode
- Links to the publication and the open source licensed version of each algorithm
- Links to developer portals (Swagger pages ) for the APIs, including the algorithm APIs, the Endeavour IM Concept set (FHIR value sets ) API, and other supporting APIs
- Links to the Endeavour IM for of concept set definitions and rule definition viewing.
EP-Predict Calculator engine
The EP-Predict calculate engine is a technical component designed to be used by IT system suppliers that wish to integrate Prediction algorithms into their application.
The main article EP-Predict Calculator Engine describes the components in more detail.
Concept sets (FHIR Value sets)
The statistical models use sets of concepts to determine the independent variables. Often referred to as reference sets (Snomed-) or Value Sets (FHIR).
These are advisory . They must not be used as the definitive source of accurate sets as variations exist between supplier systems.
All concept sets used in Endeavour are available on the Endeavour ontology server via the Endeavour IM at
https://im.endeavourhealth.net/#/
with the Swagger for the FHIR Value set API at the free self sign up developer portal
Prediction scores
The following algorithms are supported.
QRISK®3 -
QFracture®
QCancer®