Current medicine falls short at providing systematic data-driven guidance to individuals and care providers. While an individual's medical history is the foundation for every medical decision in clinical practice and is routinely recorded in most health systems, the predictive potential and utility for most human diseases is largely unknown. We explored the potential of the medical history to inform on the phenome-wide risk of onset for 1,883 disease endpoints across clinical specialties. Specifically, we developed a neural network to learn disease-specific risk states from routinely collected health records of 502,460 individuals from the British UK Biobank and validated this model in the US-American All of US cohort with 229,830 individuals. In addition, we illustrated the potential in 24 selected conditions, including type 2 diabetes, hypertension, coronary heart disease, heart failure, and diseases not formerly considered predictable from health records, such as rheumatoid arthritis and endocarditis. We show that the medical history stratifies the risk of onset for all investigated conditions across clinical specialties. For 10-year risk prediction, the medical history provided significant improvements over basic demographic predictors for 1,800 (95.6%) of the 1,883 investigated endpoints in the UK Biobank cohort. After transferring the unmodified risk models to the independent All of US cohort, we found improvements for 1,310 (83.5%) of 1,568 endpoints, demonstrating generalizability across healthcare systems and historically underrepresented groups. Finally, we found predictive information comparable with current guideline-recommended scores for the primary prevention of cardiovascular diseases and illustrated how the risk scores could facilitate rapid response to emerging pathogenic health threats. Our study demonstrates the great potential of leveraging the medical history to provide comprehensive phenome-wide risk estimation at minimal cost. We anticipate that this approach has the potential to disrupt medical practice and decision-making, from early disease diagnosis, slowing of disease progression to interventions against preventable diseases.
Models that can effectively represent structured Electronic Healthcare Records (EHR) are central to an increasing range of applications in healthcare. Due to the sequential nature of health data, Recurrent Neural Networks have emerged as the dominant component within state-of-the-art architectures. The signature transform represents an alternative modelling paradigm for sequential data. This transform provides a non-learnt approach to creating a fixed vector representation of temporal features and has shown strong performances across an increasing number of domains, including medical data. However, the signature method has not yet been applied to structured EHR data. To this end, we follow recent work that enables the signature to be used as a differentiable layer within a neural architecture enabling application in high dimensional domains where calculation would have previously been intractable. Using a heart failure prediction task as an exemplar, we provide an empirical evaluation of different variations of the signature method and compare against state-of-the-art baselines. This first application of neural-signature methods in real-world healthcare data shows a competitive performance when compared to strong baselines and thus warrants further investigation within the health domain.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.