The confluence of machine learning (ML) data-driven approaches and increased computational power, alongside access to the wealth of electronic health records (EHRs) and other emergent types of data (e.g., omics, imaging, mHealth), are accelerating the development of biomedical predictive models. Such models range from traditional statistical approaches (e.g., regression) through to more advanced deep learning techniques (e.g., convolutional neural networks, CNNs), and span different tasks (e.g., biomarker/pathway discovery, diagnostic, prognostic, etc.).

We are implementing a prototype next-generation decision support system called SmartDx, which learns the optimal sequence of diagnostic tests tailored to a patient’s unique characteristics and circumstances. The goal of this project is to optimize clinical pathways based on diagnostic accuracy, timeliness, and cost for individual patients using routine data collected longitudinally in the electronic health record.

Annually, it is estimated that more than 795,000 Americans experience a stroke. The severity of neurological damage due to an acute stroke is mitigated by the early restoration of blood flow to the affected area; and more people are now surviving strokes through earlier intervention with thrombolytic agents and interventional clot retrieval devices. Unfortunately, the rapid development of new drugs and devices in this area has made it difficult to provide treatment guidance for a given patient, and metrics for comparing outcomes between treatment groups are lacking.

Intracranial aneurysms (ICAs) are an increasingly common finding, both from incidental discovery on imaging studies and on autopsy; it is estimated that anywhere from 1-6% of the American population will develop this problem. Unfortunately, while our ability to detect ICAs has grown, our fundamental understanding of this disease entity remains lacking and significant debate continues in regards to its treatment.