醫療領域的人工智慧：深度學習時代帶給我們的驚喜
2016年Alpha Go的驚艷亮相在全球範圍內引爆了人工智慧的熱潮，AI 的深度學習能力讓人類感嘆其強大，那麼如果將 AI 應用到醫療行業，會碰撞出怎樣的火花呢？解釋 AI 的醫療應用之前先科普一下什麼是人工智慧！
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.
人工智慧（artificial intelligence）， 顧名思義，是製造智能機器，特別是智能計算機程序的科學和工程。 其致力於解決在計算機科學領域中與人工智慧相關聯的認知性問題，這些問題包括自主學習、問題解決和模式識別等。
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.
那麼AI在醫療領域的運用是怎樣的呢？
So how does AI perform in the field of healthcare?
1
AI醫療是什麼
-What is AI healthcare?
根據鯨准數據中心的《行業字典：一張圖看 AI 醫療》解釋：
According to the Jingzhun data center:
AI醫療是以互聯網為依託，通過基礎設施的搭建及數據的收集，將人工智慧技術及大數據服務應用於醫療行業中，提升醫療行業的診斷效率及服務質量，更好的解決醫療資源短缺、人口老齡化的問題。
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.
來源：鯨准數據sourced byJingzhun Data
AI 醫療主要在三個層面進行技術革新：
On the innovation aspect of AI technology within healthcare, it can be viewed in three layers:
基礎層：通過軟硬體的基礎設施，收集用戶、藥物及病理數據，並使數據互通互聯，為人工智慧的應用提供支持與可能。
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.
技術層：通過語音/語義識別、計算機視覺技術，對非結構化數據進行分析提煉。「學習」大量病理學數據文本，使其掌握問答、判斷、預警、實施的能力。
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.
應用層：是指人工智慧與不同細分領域的結合，以解決醫療行業中的某種業務需求，如智能診斷、藥物研發、智能健康管理、智能語音等醫療場景。
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技術的三大基石：
深度學習演算法+計算能力+大數據
The three major cornerstones of AItechnology: deep learning algorithm, computing capacity and big data.
2
AI 在醫療領域中所解決的問題
-The problems AI can solve in the field of healthcare
總的來說，就是將已有的醫學知識輸入到計算機程序中，利用人工智慧的數據挖掘和深度學習能力，從海量的數據中提煉出證據，根據證據在一定規則下對於病情進行推理和判斷，從而給出診斷結果和治療方案的推薦。
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
醫療影像輔助診斷-減少誤診漏診率
To reduce the ratio of mis- or missed diagnosis by supporting thediagnostic process of radiographic images
利用人工智慧的感知能力，對臨床影像進行識別。比如對肺癌的早期篩查，以及糖尿病性視網膜病變的識別。用人眼觀察很容易漏診，人工智慧可以通過對數十萬的數據影像進行分析學習，定位標記病灶並判別良惡性，從而提高診斷準確率。 在癌症的早期篩查方面，人工智慧的影像學技術能夠為醫生的診斷提供比較好的補充，相關技術目前已經比較成熟。
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
診療結果預測-提早預估風險
Predicting diagnostic results – pre-estimating risk
人工智慧通過對病症的發展規律進行推測，可以幫助醫生確立最佳診斷方案，幫助患者病情的治療。比如上海兒童醫學中心人工智慧系統的應用，針對小兒先天性心臟病手術，系統能夠建立包括手術、麻醉、體外循環等在內的一套最佳的治療方案，還能夠預測病人術後的出血風險、出血量、在 ICU 的停留時間、以及術後綜合症的風險等。
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-醫生與患者的共贏
Health management - double wins of both patients and doctors
根據個人健康檔案數據分析，人工智慧可以幫助設計個性化的健康管理計劃。1.風險識別-通過大數據分析為用戶繪製患病風險隨時間變化的軌跡；2.通過「虛擬護士」提醒用戶執行健康管理計劃，如提醒用戶按時服藥，何時接種疫苗等；3.在線診斷-基於用戶過往病史和在線對話中對癥狀的描述，提出初步診斷結果及應對措施。
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.
在目前的人工智慧醫療研究中，電子病曆數據的互聯互通性不僅能夠幫助人工智慧發揮最大的作用，同時也能幫助醫護人員對於患者進行有效管理，改善患者的就醫體驗。
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公司所研發的臨床信息系統 - 「SystemOne」，是一套全面整合的電子健康檔案（EHR）。其採用中央數據託管技術，並能夠通過資料庫進行自主學習，進而聯動門診管理、患者管理、醫院管理（參見官網https://tpp-china.com/products）實現患者醫療數據共享，為患者提供精準的治療服務支持。
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致力於創造最佳醫療體驗，願與您攜手共創美好明天
SystemOne is dedicated to creating the best medical experience, and wish to work together for better tomorrow.
長按下方關注tpp
TAG:sherryxinyu |