Research Resource for Complex Physiologic Signals

点击翻译按钮获取中文摘要

PhysioNet, established in 1999 as the NIH-sponsored Research Resource for Complex Physiologic Signals, has attained a preeminent status among biomedical data and software resources. Its data archive was the first, and remains the world's largest, most comprehensive and widely used repository of time-varying physiologic signals. Its software collection supports exploration and quantitative analyses of its own and other databases by providing a wide range of well-documented, rigorously tested open-source programs that can be run on any platform. PhysioNet's team of researchers drive the creation and enrichment of: i) Data collections that provide comprehensive, multifaceted views of pathophysiology over long time intervals, such as the MIMIC (Medical Information Mart for Intensive Care) Databases of critical care patients; ii) Analytic methods for quantification of information encoded in physiologic signals relevant to risk stratification and health status assessment; iii) User interfaces, reference materials and services that add value and improve access to the resource’s data and software; and iv) unique annual Challenges focusing on high priority clinical problems, such as early prediction of sepsis, detection and quantification of sleep apnea syndromes from a single lead electrocardiogram (ECG), false alarm detection in the intensive care unit (ICU), continuous fetal ECG monitoring, and paroxysmal atrial fibrillation detection and prediction. PhysioNet is a proven enabler and accelerator of innovative research by investigators with a diverse range of interests, working on projects made possible by data that are otherwise inaccessible. The creation and development of PhysioNet were recognized with the 2016 highest honor of the Association for the Advancement of Medical Instrumentation (AAMI). PhysioNet's world-wide, growing community of researchers, clinicians, educators, trainees, and medical instrument and software developers retrieve about 380 GB of data per day and publish a yearly average of nearly 300 new scholarly articles. Over the next five years we aim to: 1) Enhance PhysioNet’s impact with new data and technology; 2) Develop new methods to quantify dynamical information in physiologic signals relevant for health status assessment, and for acute and chronic risk stratification, and 3) Harness the research community through our international Challenges that address key clinical problems and a new data annotation initiative.