Predictive risk assessment and risk stratification models based on postcode-level consumer classification are widely used for life insurance underwriting. However, these are socio-economic models not directly related to health information. Similar to precision medicine, precision life insurance should aim to tailor policy pricing/reserving to the individual health characteristics of each client.
In real life, people develop new health conditions, lifestyle habits, and they can start and stop a certain treatment regime at any time.
Additionally, clinical guidelines are regularly updated with new evidence, resulting in new eligibility criteria and treatment courses. This requires the ability to dynamically classify clients into time-varying subgroups with the predictable life expectancy based on the presence of evolving health-related conditions, treatments and outcomes.
In this talk, we demonstrate how the landmark survival modelling of electronic health records (EHR) can be used for dynamic prediction of individual and population life expectancy.
We discuss a case-study based on landmark analysis of the use of statins. For this case study, we consider a cohort of 110,243 participants who reached age 60 between 1990-2000 with no previous history of cardiovascular disease or statin prescription at baseline. Participants’ medical history was updated at ‘landmark’ time points occurring every six months.
Speaker: Elena Kulinskaya, University of East Anglia
This webinar will be streamed from 12.00-13.00. There will be time at the end of the session for Q&A.
Nearest Public Transport