Patient-Level Assessment of Treatment Outcomes: Predictive models that assess the probability of a patient experiencing any outcome following initiation of any intervention, given his or her personal medical history.
Patients seek medical care to diagnose and treat illness. Current medical practice relies on limited aggregate information for prognosis and prediction of a patient’s health. When predictive models are used in healthcare they draw on data from hundreds to thousands of patients and consider small numbers of patient characteristics, often five or fewer. This contrasts sharply with the reality of modern medicine wherein patients generate a rich digital trail, which is well beyond the power of any medical practitioner to fully assimilate. The recent emergence of massive patient-level databases of electronic health records and administrative claims opens up extraordinary opportunities for massive-scale, patient-specific predictive modeling. Such models can inform truly personalized medical care leading hopefully to sharply improved patient outcomes.
We are developing predictive models using longitudinal data on over 100 million patients observed for multiple years and comprising over 5 billion clinical observations. Similarly large populations afford rich data to build highly predictive large-scale models and also provide immediate opportunity to serve large communities of patients who are in most need of improved quality of care. Effective exploitation of these data demands novel methodology and an interdisciplinary approach. We believe that OHDSI’s combination of backgrounds, our unfettered access to a unique data resource, and our substantial collaborative track record, our team is well positioned to advance the field.
Our research will focus on developing models and algorithms to derive clinically relevant predictive models from irregularly-spaced longitudinal electronic healthcare data, developing algorithms to use this information in large-scale multivariate modeling, and evaluating performance based on accuracy in predicting outcomes at the patient level. Predictive modeling in databases containing data for upwards of 100 million patients presents non-trivial engineering challenges. However, our team has considerable experience with these data and we have developed and honed a research computing environment tailored to our needs. The overarching goal of our work is to establish a standardized process for developing accurate and well-calibrated patient-centered predictive models that can be utilized for multiple outcomes of interest to patients and applied to observational healthcare data from any patient subpopulation of interest.
All patient-level predictive model tools will be developed as open-source solutions built against the OMOP common data model. Person-Level Assessment of Treatment Outcomes (PLATO) will be an integrated framework to allow all users to use the library of predictive models developed to produce individualized risk for all medical interventions and all health outcomes of interest, based on personal demographics, medical history, and health behaviors.
