Developer tools

The algorithms that power Project Big Life are statistical and machine learning algorithms developed by medical experts and researchers. Using our APIs and/or supporting references, your applications can use the same tools that our scientists use in their research. These algorithms provide your users with meaningful information about their health or the health of people in their care.

Two ways to make calculations

There are two ways to use our algorithms:

1. Our JavaScript Library

All the algorithm calculations done on the Project Big Life website are powered by our proprietary JavaScript library. At its core, the library is an evaluator for models written in Predictive Modelling Markup Language (PMML) but it also does much more. In addition to simple risk calculations, APIs provide additional outcomes, measures and features. For example, 5-year all-cause mortality risk is the outcome/target for the Mortality Population Risk Tool, but the API can also calculate:

• life expectancy;
• survival to specified age ("Will you live to see it?);
• health age (users age compared to a reference population); and,
• mortality risk compared to reference population.

Mortality risk can be estimated considering the effect of:

• health behaviours (smoking, alcohol, diet and exercise);
• health interventions; and,
• external risks not included in the original algorithm (e.g., the effect of air pollution on life expectancy).

Additional features include:

• support to assess calibration and recalibrate the algorithms in new populations; and,
• validation tools and data to ensure implementation is correctly calculating risk estimates.

For information regarding licencing our library for your use please contact us at projectbiglife@toh.ca or click here

2. Your scoring engine

All Project Big Life algorithms are available in Predictive Modelling Mark-up Language that provides a standard way to represent data mining models, enabling implementation in different statistical application and computing environments.

Our questionnaire is developed using Lime survey, an open-source survey system. Our Lime files are available, providing further information on how we implement Project Big Life algorithms.

Validation data is available, enabling you to check whether your implementation is generating correct calculations.

Additional tools and resources

Project Big Life algorithms are published in peer-reviewed medical literature. The research studies have additional information to help you to gauge whether the algorithms are helpful in different settings and for different applications.

Web appendices for these studies include additional resources such as:

• interactive algorithm visualization tools;
• calculated mean risk in different populations;
• ‍recalibration coefficients for different populations;
• bootstrap files of 500 algorithm coefficients to enable calculation of statistical error; and,
• ‘batch mode’ version of our API that allows you to upload data files in your web browsers and then have your data validated and scored (under development). The same as main Project Big Life API, all calculations are performed locally, allowing calculations to be performed in your secure environment..

See our science (link) section and our GitHub repository for more information.