医疗领域的人工智能:深度学习时代带给我们的惊喜

2016年Alpha Go的惊艳亮相在全球范围内引爆了人工智能的热潮,AI 的深度学习能力让人类感叹其强大,那么如果将 AI 应用到医疗行业,会碰撞出怎样的火花呢?解释 AI 的医疗应用之前先科普一下什么是人工智能!

Humanity has been amazed with the implementation of artificial intelligence (AI) throughout the world by Alpha Go in the spree of 2016. Still, what about the implementation of AI technology within the healthcare industries? Before moving into its application, it is important to learn what AI is all about.

人工智能(artificial intelligence), 顾名思义,是制造智能机器,特别是智能计算机程序的科学和工程。 其致力于解决在计算机科学领域中与人工智能相关联的认知性问题,这些问题包括自主学习、问题解决和模式识别等。

AI, as the name implies, is referred to science and engineering which can create intelligent machines especially within a computer system. It is devoted to solving cognitive issues which entail self-learning, problem-solving, pattern recognition and so on.

那么AI在医疗领域的运用是怎样的呢?

So how does AI perform in the field of healthcare?

1

AI医疗是什么

-What is AI healthcare?

根据鲸准数据中心的《行业字典:一张图看 AI 医疗》解释:

According to the Jingzhun data center:

AI医疗是以互联网为依托,通过基础设施的搭建及数据的收集,将人工智能技术及大数据服务应用于医疗行业中,提升医疗行业的诊断效率及服务质量,更好的解决医疗资源短缺、人口老龄化的问题。

It is revealed from the Internet that AI technology within healthcare is developed through the building of basic infrastructure and data collection. It aims to improve the efficiency of diagnosing and the quality of service. Furthermore, it deals with the shortage of medical resources and the issue on the aging population.

来源:鲸准数据sourced byJingzhun Data

AI 医疗主要在三个层面进行技术革新:

On the innovation aspect of AI technology within healthcare, it can be viewed in three layers:

基础层:通过软硬件的基础设施,收集用户、药物及病理数据,并使数据互通互联,为人工智能的应用提供支持与可能。

Foundation layer:Through the infrastructure of software and hardware, AI collects data about the patient, the medication and the diagnostic information. Moreover, it interconnects the data to support the AIapplication.

技术层:通过语音/语义识别、计算机视觉技术,对非结构化数据进行分析提炼。“学习”大量病理学数据文本,使其掌握问答、判断、预警、实施的能力。

Technology layer:Through the recognition of voice and lexeme as well as vision technology of computers, AI can extract and analyze non-structured data. It studies vast amount of diagnostic information from patient’s data allowing it to have the ability to answer, judge, warn and implement.

应用层:是指人工智能与不同细分领域的结合,以解决医疗行业中的某种业务需求,如智能诊断、药物研发、智能健康管理、智能语音等医疗场景。

Application layer:This is a combination of AI and other areas which aim to solve the possible demands from certain medical industry operations. For example, the demands can be from situations like intelligent diagnosis, drug development, intelligent health management, intelligent voiceand others.

AI技术的三大基石

深度学习算法+计算能力+大数据

The three major cornerstones of AItechnology: deep learning algorithm, computing capacity and big data.

2

AI 在医疗领域中所解决的问题

-The problems AI can solve in the field of healthcare

总的来说,就是将已有的医学知识输入到计算机程序中,利用人工智能的数据挖掘和深度学习能力,从海量的数据中提炼出证据,根据证据在一定规则下对于病情进行推理和判断,从而给出诊断结果和治疗方案的推荐。

Briefly, AI technology in medical treatment is to input existing medical knowledge into a computer system creating a mass of data, which can then be extracted through data mining and deep learning. Following certain rules, the diagnosis is made according to reasoning and estimation on the basis of clinical evidence, thereby proposing the best-fit treatment possible.

1

医疗影像辅助诊断-减少误诊漏诊率

To reduce the ratio of mis- or missed diagnosis by supporting thediagnostic process of radiographic images

利用人工智能的感知能力,对临床影像进行识别。比如对肺癌的早期筛查,以及糖尿病性视网膜病变的识别。用人眼观察很容易漏诊,人工智能可以通过对数十万的数据影像进行分析学习,定位标记病灶并判别良恶性,从而提高诊断准确率。 在癌症的早期筛查方面,人工智能的影像学技术能够为医生的诊断提供比较好的补充,相关技术目前已经比较成熟。

Clinical imaging can also be seen making use of the advances of artificial intelligence. For instances, AI can screen for early signs of lung cancer as well as detect diabetic retinopathy. It is possible to miss a diagnosis with the naked eye but with AI, the correct ratio of diagnosis can be improved. This is developed through the learning and analyzing of hundreds of thousands of images, followed by marking the nidus and differentiating between benign or malignant. As for the screening of early lung cancer, the imageology technology of AI has been able to complement diagnosis made by doctors and the currently evolving technologies.

2

诊疗结果预测-提早预估风险

Predicting diagnostic results – pre-estimating risk

人工智能通过对病症的发展规律进行推测,可以帮助医生确立最佳诊断方案,帮助患者病情的治疗。比如上海儿童医学中心人工智能系统的应用,针对小儿先天性心脏病手术,系统能够建立包括手术、麻醉、体外循环等在内的一套最佳的治疗方案,还能够预测病人术后的出血风险、出血量、在 ICU 的停留时间、以及术后综合症的风险等。

AI can speculate the pattern of disease progression and assist doctors to determine the best fit treatment proposal for patients. For instances, an AI system at Shanghai Children Medical Centre can build a set of schemes, targeting children with congenital heart diseases, which include surgery, anesthesia, extracorporeal circulation, etc. Besides this, it can predict the risk of post-operative bleeding, the potential amount of blood loss, the duration of admission at ICU, post-operative syndromes, etc.

3

健康管理-医生与患者的共赢

Health management - double wins of both patients and doctors

根据个人健康档案数据分析,人工智能可以帮助设计个性化的健康管理计划。1.风险识别-通过大数据分析为用户绘制患病风险随时间变化的轨迹;2.通过“虚拟护士”提醒用户执行健康管理计划,如提醒用户按时服药,何时接种疫苗等;3.在线诊断-基于用户过往病史和在线对话中对症状的描述,提出初步诊断结果及应对措施。

By analyzing patients’ health information, the AI can design health plans which are individualised toward a particular patient as follow: 1) risk identification: through big data analysis, AI can track the changes as the illness progresses over time; 2) virtual nurses: AI can remind users to implement their health management plans; 3) online diagnosis: proposing the initial diagnostic results and response measures based on the patients’ past medical histories and description of their symptoms via online conversation.

在目前的人工智能医疗研究中,电子病历数据的互联互通性不仅能够帮助人工智能发挥最大的作用,同时也能帮助医护人员对于患者进行有效管理,改善患者的就医体验。

With the current development of AI, the communication of electronic medical records can not only help AI to perform better, but also assist medical staffs to manage patients more efficiently. This ultimately improve overall patient-care and treatment experience.

TPP公司所研发的临床信息系统 - “SystemOne”,是一套全面整合的电子健康档案(EHR)。其采用中央数据托管技术,并能够通过数据库进行自主学习,进而联动门诊管理、患者管理、医院管理(参见官网https://tpp-china.com/products)实现患者医疗数据共享,为患者提供精准的治疗服务支持。

TPP has developed a clinical information system, "SystemOne", which is a comprehensive integrated electronic health record (HER). It adopts central data management technology and through self-learning from the given databases, it can coordinate the outpatient management, patient management and hospital management (refer to https://tpp-china.com/products). The administration of medical informationcan thereby offer possible supports and precise treatments toward patients.

SystmOne致力于创造最佳医疗体验,愿与您携手共创美好明天

SystemOne is dedicated to creating the best medical experience, and wish to work together for better tomorrow.

长按下方关注tpp

  • 发表于:
  • 原文链接https://kuaibao.qq.com/s/20180530G08JBF00?refer=cp_1026
  • 腾讯「云+社区」是腾讯内容开放平台帐号(企鹅号)传播渠道之一,根据《腾讯内容开放平台服务协议》转载发布内容。
  • 如有侵权,请联系 yunjia_community@tencent.com 删除。

扫码关注云+社区

领取腾讯云代金券