# How is risk quantified and measured by the AI?

We measure absolute risk as the chance that a given disease might, in purely statistical terms, develop over the next 5 years. This is best interpreted as follows: a 0.3% absolute risk of developing a heart condition is otherwise phrased as a “3 in 1000” risk. This means that if 1000 people provided identical question responses the statistics suggest that three people might develop that particular condition over the next 5 years.

A number for absolute risk can be calculated but it is much harder to define what is “high” or “low”risk. As such, we also provide a comparison of the statistical risk profile to others of the same age and gender, presented as the disease risk bar. We produce this bar by simulating a large population of people of the same age and gender and producing risk profiles for them, and then sorting these profiles in order from lowest to highest absolute risk. We then place the profile a users creates on this bar to illustrate the risk profile position relative to this simulated population, based on the information provided.

The green section of the bar represents the risk profiles below the “median” risk profile position. Amber and red represent those above the median, with red being those significantly above (specifically, those more than a standard deviation of the absolute risk values above the median risk value). Being in the red section does not mean the information entered suggests a “high” risk, only that that information suggests more risk in statistical terms than other people of the same age and gender. Increased age in most cases results in the risk level increasing.

Being more toward the amber and red sections will be correlated with a higher risk  caused by specific risk influences. In many cases, there will be healthy lifestyle suggestions which could be made (such as reducing alcohol intake, quitting smoking, etc.) which are known statistically to reduce the risk level. In other cases, elements such as family history of a disease may be a strong influence on risk and are non-modifiable. Again, this may result in a higher risk than someone of equivalent age and gender, but does not necessarily result in a high risk overall.