Predicting and preventing all forms of cardiac arrest in hospitals with CodeRhythm™ AI software.

在英国，住院期间心脏骤停是导致死亡的主要原因，每年影响15000名患者。只有五分之一的受害者最终活着离开了医院。心脏骤停通常可以通过在心脏停止跳动之前及时改变患者的医疗护理来防止，但只有在医生提醒患者处于危险之中并有足够的时间采取行动时，才能做出这样的改变。当心脏骤停确实发生时，如果医生立即采取行动，患者存活的可能性要大得多。仅仅几分钟的延迟就会大大降低患者的存活几率。目前的心电监测技术只在心脏骤停发生后才能识别，对预防没有帮助，使快速反应变得困难。未来的预测技术可以在患者出现即将发生心脏骤停的高风险时向护士和医生发出警报，从而能够提供挽救生命的预防性护理，并在心脏骤停发生时做出更快的反应。Transform正在构建这项技术。使用最初在CERN大型强子对撞机开发的先进人工智能技术，并积累了世界上最大的医院连续心电监测数据数据库，Transform构建了一种专有的机器学习算法-CodeRhythm(Tm)--可以预测可电击节律导致的心脏骤停。根据初步的气候测试，CodeRhythm对电击IHCA的预测准确率高达97%，最长可提前8小时。在证明了准确预测可电击节律导致的心脏骤停的能力后，Transform Now在这个拟议的项目中旨在改进CodeRhythm，使其也能够预测非电击心脏骤停。在Innovate-UK的支持下，并与Barts NHS Trust合作，一个为期15个月的项目将提供一个预测所有类型心脏骤停的概念验证模型。如果成功，这种方法将为所有形式的心脏骤停提供第一个可靠的预测监测软件，这可能会改善对高危患者的护理，并将存活率提高一倍以上。

In-hospital cardiac arrest is a leading cause of death in the UK, impacting \>15k patients per year. Only one-in-five victims ultimately leave the hospital alive.Cardiac arrest can often be prevented by making timely changes to a patient's medical care before the heart stops beating, but such changes can only be made if physicians are alerted that a patient is in danger with enough time to act. When cardiac arrest does occur, patients are much more likely to survive when physicians act right away. Delays of mere minutes can dramatically lower a patient's odds of surviving.Current ECG monitoring technology only identifies cardiac arrest after it occurs, providing no help for prevention and making rapid response difficult. Future predictive technology that alerts nurses and physicians when a patient develops a high-risk of imminent cardiac arrest could enable life-saving preventive care and faster response when cardiac arrest nonetheless occurs. Transformative is building this technology.Using advanced AI techniques originally developed at the CERN Large Hadron Collider, and having amassed the world's largest database of hospital continuous ECG monitoring data, Transformative has built a proprietary machine learning algorithm---CodeRhythm(tm)---that can predict cardiac arrests caused by a shockable rhythm. Based on initial climnical testing, CodeRhythm has a 97% accuracy at predicting shockable IHCA up to 8 hours in advance. Having proven the ability to accurately predict cardiac arrest due to shockable rhythms, Transformative now aims, in this proposed project, to improve CodeRhythm so that it can also predict non-shockable cardiac arrest.With Innovate-UK support and in collaboration with Barts NHS Trust, a 15-month project, will deliver a proof-of-concept model that predicts all types of cardiac arrest. If successful, the approach will offer the first reliable predictive monitoring software for all forms of cardiac arrest, which could improve care for high-risk patients and more than double survival rates.