Vast amounts of longitudinal data accumulating in electronic health information systems present an untapped opportunity to improve medical screening and diagnosis. The Intelligent Histories project works towards realizing the vision of Predictive Medicine by developing new ways of using commonly available electronic medical information to predict people's future medical risks, helping doctors choose preventive interventions and improve medical care.
Working from large de-identified medical research databases, we have developed advanced Bayesian models, known as Intelligent Histories, that are capable of predicting a patient's risk of receiving certain future diagnoses, up to years in advance, based solely on the patient's past diagnostic history.
With increasing amounts of clinical and genetic data becoming available, this work has the potential to bring closer the vision of Predictive Medicine, in which vast quantities of information are used to predict individuals' future medical risks in order to improve medical care and diagnosis.
Our initial work focuses on long-term predictors of domestic abuse:
Reis BY, Kohane IS, Mandl KD. Longitudinal histories as predictors of future diagnoses of domestic abuse: modelling study. BMJ 2009; 339:b3677
Intelligent Histories models are designed to serve as the basis for a clinical decision support system capable of detecting long-term indicators of health risks, alerting a patient's doctor when a significant risk is identified.
The central part of this prototype system is the risk-visualization environment, known as a Risk Gel, which provides clinicians with instant overviews of longitudinal medical histories and related risk profiles at the point of care. In conjunction with alerts for high-risk patients, this visualization enables clinicians to rapidly review and act on all available historical information by identifying important risk factors and long-term trends.
Click to view sample visualizations 1, 2.