AI-driven laryngeal cancer assessment

我们的愿景是使用机器学习来从患者的智能手机语音记录中检测喉癌风险。这样做的目的是自信地将良性声带问题转移到更合适的护理途径。如今，在初级保健中出现“声音沙哑”的患者被转介到头颈部癌症诊所。据估计，英格兰每年有28,000人转诊，其中绝大多数(93%)没有患癌症。如果我们能够提供人工智能驱动的智能手机测试来简化这一过程，对患者和医疗系统有多种好处。来自利物浦大学医院NHS基金会信托基金的头颈部癌症外科医生和耳鼻喉科顾问是该项目的幕后推手。他们认为，倾听患者沙哑的声音可以提供足够的信息来排除癌症。在这个项目中，我们将使用机器学习通过智能手机来听声音，并开发一个原型系统来将患者送到合适的医疗机构。最终，这样的系统可以在每个全科医生的初级保健中部署，从而为NHS带来显著的患者利益和成本节约。

Our vision is to use machine learning to detect laryngeal cancer risk from smartphone voice recordings of patients. The purpose is to confidently divert benign vocal problems into more appropriate care pathways.Today, patients presenting at primary care with a "hoarse voice" are referred to a head and neck cancer clinic. Of the estimated 28,000 annual referrals in England, the vast majority (93%) do not have cancer. If we are able to provide an AI-driven smartphone test to streamline this process, there are multiple benefits to patients and the healthcare system. Head and neck cancer surgeons and ENT consultants from the Liverpool University Hospitals NHS Foundation Trust are the driving force behind this project. They believe that listening to a patient's hoarse voice provides sufficient information to rule out cancer. In this project, we will use machine learning to listen to a voice, via a smartphone, and develop a prototype system to route patients to the right care. Eventually such a system could be deployed in primary care, at every GP, resulting in significant patient benefit and cost savings to the NHS.