Google DeepMind and United Kingdom medical institutions: using artificial intelligence to prevent disease-artificial intelligence-IT information
Google DeepMind AlphaGo beat human chess championships today, it began to enter the medical field. According to the guardian report, DeepMind NHS (United Kingdom national health service) again, along with Moorfields Eye Hospital development learning system identification of Visual diseases. Through a eye scan, the system can identify Visual disease early symptoms, achieve the purpose of Visual diseases prevention in advance.
This is DeepMind second effort with the NHS. Prior to this, DeepMind had cooperated with the Royal free hospital in North London, with Smartphones to monitor the patient's renal function. DeepMind co-founder Mustafa Suleyman said, this partnership is the first purely medical research company. Meanwhile, the company first applied machine learning medical projects.
(Picture from theverge)
The core part of this study is to share 1 million eye scans, DeepMind researcher will be used to train a machine learning system to better identify Visual early symptoms of the disease.
"This study is very important, especially in the diagnosis of diabetic retinopathy. If you have diabetes, then you will increase the chance of blind 25 times. If we are able to detect this situation and start treatment as soon as possible, then 98% of severe vision loss can be avoided. "Suleyman said.
Moorfields with DeepMind thanks to the hospital's consultation ophthalmologist Pearse Keane. DeepMind official website, Pearse Keane discussed the analysis of how to do eye scans, he then made contact with Suleyman, initiated this project.
DeepMind previous co-operation with the NHS has led to concerns about privacy. Relatively speaking, this dispute will be smaller, because sharing information is anonymous. "This means that from the scanned image cannot be identified in any of the patients. Meanwhile, these scans are history, that is to say, our results can be used to improve future health care, but does not affect the nursing care of patients accepted until now. When you use this kind of data research, anonymous data that researchers could not identify individual patients, then the explicit consent of the patient is no longer necessary. ”
(Photo from the Telegraph)
Moorfields Eye Disease Research Center director Peng Tee Khaw said that the cooperation is the key to accurate eye scans are fast increasing. "The sophistication of these scans is incredible, more fine than other parts of the body scans: we can see the very microscopic level, the question is, how do we deal with such a large amount of data. ”
"You want to track a patient's history, I need to use on a lifetime of experience, predict the patient's future, they have to rely on my experience. If we are able to use machine-assisted deep learning, we are able to work better, because I will have 10,000 times the life experience. ”
谷歌DeepMind和英国医疗机构合作：用人工智能预防疾病 - 人工智能 - IT资讯
Google DeepMind曾以AlphaGo战胜了人类围棋冠军，如今，它又开始进军医疗领域了。据卫报的报道，DeepMind与NHS（英国国家医疗服务体系）再次合作，同Moorfields眼科医院一起开发辨识视觉疾病的机器学习系统。通过一张眼部扫描图，该系统能够辨识出视觉疾病的早期症状，达到提前预防视觉疾病的目的。
这是DeepMind与NHS的第二次合作。在此之前，DeepMind曾与伦敦北部的皇家自由医院合作，用智能手机监控病人的肾功能。DeepMind联合创始人Mustafa Suleyman说，这次合作是公司首次进行纯粹的医学研究。同时，这也是公司首次把机器学习应用于医疗项目。
(图片来自 theverge)
这项研究的核心部分是分享100万张眼部扫描图，DeepMind的研究员将用来训练一个机器学习系统，更好地辨识出视觉疾病的早期症状。
“这次研究是非常重要的，特别是糖尿病视网膜病的诊断。如果你有糖尿病，那么，你变盲的机率会增加25倍。如果我们能够检测出这种状况，并且尽快展开治疗，那么，98%的严重视觉丧失都可能被避免。”Suleyman表示说。
Moorfields与DeepMind的合作要归功于该医院的咨询眼科专家Pearse Keane。在DeepMind的官方网站上，Pearse Keane探讨了如何做好眼部扫描图的分析，随后，他与Suleyman取得了联系，启动了这个合作项目。
DeepMind与NHS的上次合作曾导致人们对隐私的担忧。相对来说，此次合作的争议会小一些，因为分享的信息都是匿名的。“这意味着，从这些扫描图像中无法辨识出任何的病人。同时，这些扫描图属于历史了，就是说，我们的研究结果能用来改善未来的医疗，但不会影响到现在病人接受的护理。当研究中使用了这样的数据，即研究人员无法辨识个体病人的匿名数据，那么，病人的明确许可就不再是必要的了。”
(图片来自 telegraph)
Moorfields眼科疾病研究中心的主管Peng Tee Khaw说，此次合作得以进行的关键在于，精确的眼部扫描图正在快速地增多。“这些扫描图的精细程度是不可思议的，比其它身体部位的扫描图更为精细：我们能够看到非常微观的层面，问题在于，我们如何处理如此大量的数据。”
“要跟踪一个病人的历史，我需要用上毕生的经验，而预测病人的未来时，他们要依赖于我的经验。如果我们能够使用机器辅助的深度学习，我们就能更加出色地完成工作，因为我将有1万次人生经验。”