徐州醫(yī)療大數(shù)據(jù)分析徐建業(yè)2015

1、Health Information Database Application巨量醫(yī)療健康生活數(shù)據(jù)分析與應(yīng)用Big Data for Biomedical Applications,1,2015-06-26徐州醫(yī)學(xué)院急診醫(yī)學(xué)兩岸學(xué)術(shù)交流綜合論壇,Chien-Yeh Hsu 徐建業(yè) PhD臺(tái)北護(hù)理健康大學(xué)信息管理系所 National Taipei University of Nursing and Health Science

2、s 臺(tái)北醫(yī)學(xué)大學(xué)醫(yī)學(xué)信息研究所Taipei Medical University臺(tái)灣醫(yī)學(xué)信息學(xué)會(huì)Taiwan Association for Medical Informatics TAMI,2,Roadmap for ICT Development in Taiwan,行政院,,,m-Taiwan（2005~）,NHIP & U-Taiwan （2008~）,Ubiquitous e-Service,Mobile Se

3、rvices,,,Web GovernmentServices,Web HealthcareServices,HealthInsuranceIC Card,HIS,HIN,NII,Tele Medicare,Construct Healthcare Informatics Infrastructure,PersonalHealthcareRecord,2002,2005,2008,,ElectronicMedicalRe

4、cord,2011,Governmente-Service& e-indus-trialization,International Trend,Application Trend,Development Trend of Taiwan Health Systems,4,The NHI VPN國(guó)家衛(wèi)生基礎(chǔ)設(shè)施,,,HIN,,,,,NHI VPN,,,IDC,,NHI local offices,,NHI head-quarte

5、r,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,CDC,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,17,000Clinics,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,

6、,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,DOH,,,SC,,,,,,,,,,,,,,,,,,,,,,,,Admin,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,

7、,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,BOH,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,HCA,600Hospitals,5,000Pharmacy,Other 100HIS,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,

8、,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,IDC – Internet Data CenterSC – Service CenterHIN – Health Information NetworkNHI VPN – Virtual Private Network,◎ 【1-1】個(gè)人健康照護(hù)信息整合云端服務(wù)[執(zhí)行單位] 醫(yī)學(xué)信息學(xué)會(huì),衛(wèi)生「健康云」規(guī)劃,◎ 【3-1】遠(yuǎn)距健康照護(hù)服務(wù)計(jì)劃[

9、執(zhí)行單位] 衛(wèi)生署照護(hù)處、工研院,◎ 【2-1】診所病歷云端備份服務(wù)[執(zhí)行單位] 衛(wèi)生署醫(yī)事處、工研院,數(shù)據(jù)源：資策會(huì)-創(chuàng)研所整理, 2010,1-1 個(gè)人健康照護(hù)信息整合云端服務(wù)2-1 診所病歷云端代管及備份服務(wù)2-2 署立醫(yī)院醫(yī)療照護(hù)云端服務(wù)3-1 遠(yuǎn)距健康照護(hù)服務(wù)計(jì)劃4-1 健康數(shù)據(jù)加值中心網(wǎng)絡(luò)化服務(wù),◎ 【2-2】署立醫(yī)院醫(yī)療照護(hù)云端服務(wù)計(jì)劃[執(zhí)行單位] 衛(wèi)生署醫(yī)管會(huì)、署立醫(yī)院、資策會(huì),【4-1】健康數(shù)據(jù)加值中心網(wǎng)絡(luò)

10、化服務(wù)[執(zhí)行單位] 衛(wèi)生署統(tǒng)計(jì)室、資策會(huì),,Medical services,Rehabilitation &Follow-up services,Healthcare services,from Dr. Hsu, Min-Huei, DOH, Taiwan,Big Data and Information,Innovation, Social NetworkingWellness, travel, sport, diet

11、ary,Value of service mode, Need more evidence, Insurance,7,HCA Card醫(yī)事人員卡及醫(yī)?？?RSA Card issues from HCAto health professionals,non-RSA Health insurance IC Card for all citizens,8,Establishing EMR in Taiwan,Vision: At any

12、 hospital, a patient can get his/her integrated medical records using the health insurance IC card under the agreement and authorization of the patient.Goal: By 2012, 80% hospitals(400, no clinics) DICOM and report, Tes

13、t reports, and medications, 60% hospitals can exchange EMRs。By 2014-(2016) complete EMR and exchange for all hospitals,9,Medical Information Exchange Center – MIEC 2000,10,,,,,,,,,Hospital Information System,Laboratory

14、Information Systems,TMT mini-server,個(gè)人化健康信息整合架構(gòu)TMT File Exchange Pathway,,,,,Hospital Information System,,,Paperless server,,TMT mini-server,,LaboratoryInformation Systems,,,,1,2,3,TMT viewer,Pre-Authorized,,4,Internet

15、 Health and Life Supporting Data Bank,,,EMR Exchange Center,National EEC Center,EMR providing Hospitals,EMR Reading Hospitals,Download, Querying, and Reading,12,Ministry of HealthImage Exchange Center,Health Insurance C

16、enter,Hospital A,Hospital B,Index Server,CPOE,CPOE,Download image,Download image,Radiology,Radiology,Image Report Database,Image Report Database,eSignature,Dual Card system and inform consent,Dual Card system and inform

17、consent,Request for image,Request for image,134 hospitals, 2010-2011 upload index 2,168,063 request: 6,592 download: 81,108,Let’s ask google about “big data”,,What Is Big Data?,High-volume (大量)High-variety (多種類)High-

18、velocity (快速)sources such as online personal activity, commercial transactions, and sensor networksRelating to health is a component of a growing field.e.g., e-health, m-health, digital health, health information tec

19、hnology, health 2.0, e-medicine, etc. Nilsen W, et al. J Health Commun, 2012. Laney D. META Group, 2001.Kumar S, et al. Computer, 2012,Bio-Medical and Health Informatics needs Analytics,醫(yī)院信息系統(tǒng)快速的發(fā)展,各類數(shù)具快速大量的累積,需

20、要分析數(shù)據(jù)來(lái)改善健康照護(hù),What is BIG DATA?,Wikipedia: a collection of data sets so large and complex that it becomes difficult to process using on-hand database management tools. The challenges include capture, curation, storage, se

21、arch, sharing, analysis, and visualization. The trend to larger data sets is due to the additional information derivable from analysis of a single large set of related data, as compared to separate smaller sets with the

22、same total amount of data, allowing correlations to be found to "spot business trends, determine quality of research, prevent diseases, link legal citations, combat crime, and determine real-time roadway traffic con

23、ditions”.,http://www.scientificamerican.com/article.cfm?id=what-is-the-memory-capacity,What about human brain? The Human brain’s memory storage capacity compares to something closer to around 2.5 petabytes (or a thousand

24、 Terabytes); which hold three million hours of TV shows. You would have to leave the TV running continuously for more than 300 years to use up all that storage.,How Big is BIG DATA?,,Database Size,Mb,Gb,Tb,Pb,Eb,,,,,,Hum

25、an brain capacity ~ 2. 5 Pb,NIH BTRIS DW 42K patients with 4 billion rows; ~ 1 Tb,Human Genome (compressed) < 1Gb,Facebook ~ 100 Pb,,NSA?Google?,NCBI GenbankDownload ~ 600 Gb,Amazon.com~ 42 Tb,Variety of BIG DA

26、TA Sizes,,Google Map~ 80 Tb,,WikiPedia> 10 million articles <50 Gb,Leveraging Big Data to Prevent Disease,Disease prevention requires two steps.Identify modifiable risk factors for disease e.g., diet, exercise,

27、 smoking, alcohol consumption, and environmental pollution.Interventions to improve disease risk factors To help that person achieve these goals.,International Cooperation ExampleGlobal AllianceJust announced: 7

28、 June 201370 organisations joining to promote sharing andstandardisation of genomic data,A Global Alliance for sharing genomic and clinical dataA White Paper circulated in early 2013 has the support of nearly 70 organ

29、isations in Asia, Australia, Africa, Europe, North America and South America who are committed to creating a common framework that supports data analysis and protects the autonomy and privacy of participating individuals

30、. http://www.ebi.ac.uk/about/news/press-releases/Global-Alliance,Integrated Biomedical Informatics for Clinical Research醫(yī)學(xué)研究之道，就是整合研究數(shù)據(jù),2003年9月30日，美國(guó)國(guó)家衛(wèi)生研究院（NIH）院長(zhǎng)塞烏尼（Elias Zerhouni）宣布：對(duì)美國(guó)政府資助的醫(yī)學(xué)研究進(jìn)行重整?！秆芯柯肪€圖」的計(jì)劃 NIH

31、 RoadmapNational Electronics Clinical Trials and Research (NECTAR) network 建立完整的路線圖以及徹底更新醫(yī)學(xué)數(shù)據(jù)的收集、儲(chǔ)存及共享 methods for collecting, storage, and sharing把龐大且分散的數(shù)據(jù)庫(kù)結(jié)合成一個(gè)巨大的數(shù)據(jù)庫(kù) integrating data bases發(fā)展軟件，使實(shí)驗(yàn)計(jì)劃的撰寫(xiě)能夠簡(jiǎn)化并標(biāo)準(zhǔn)化 dev

32、eloping software for helping experiment design and common data element 減少紙張的使用 reducing paper use,… needs a re-usable, extensible, sharable and interoperable informatics infrastructure to enable and streamline collabora

33、tion and data sharing for translational research…,Starts with “Quality Datasets”,BIG DATA in Biomedical Research…,Yang C. Fann, Ph.D. 2014 NIH/USA,Databanks related to health,Health-related Databanks,24,Use of cloud tech

34、nology to provide health information value-added services加值應(yīng)用,BUILDING A CLOUD-BASED CLINICAL DATA REPOSITORY (CDR),Bio-medical and Healthcare Data are BIG Data = EMR data + genomic data,New Technology TriggersThe Nexu

35、s of IT Forces:Social, Mobile, Cloud & Big Data/Information,Source: Gartner, 2013,Facebook can predict your breakups,3/23/2024,©2011 Healthcare Information and Management Systems Society,27,Eating Habits,Stop

36、 counting calories start eating better https://eatery.massivehealth.com/,國(guó)家級(jí)的健康資料中心,Providers,TJCHA,FDA,,,National Data Data Center,Local Health Bureau,Bureau of Health Promotion,NHI,DOH,CDC,NHRI,MOI,,,,,,,,,,,NHIRDB i

37、n Taiwan,NHIRDB (National Health Insurance Research Database) 12 years of de-identified claim database for 23 million peopleCohort DB (Five 1-million people groups for 13 years)Disease-specific DB (16 disease groups)

38、Random sample DB (outpatient 1/500, inpatient 1/20)generates >100 research papers a year,健康加值數(shù)據(jù)的價(jià)值,Secondary Use(加值應(yīng)用或二次運(yùn)用)去識(shí)別化之健康資料為世界趨勢(shì)美國(guó)早已在20年前開(kāi)放全國(guó)住院數(shù)據(jù)供研究者使用新的治療方式、疾病的診斷、藥物之副作用、疾病之關(guān)聯(lián)性…等若沒(méi)有完整開(kāi)放健康加值資料將嚴(yán)重?fù)p害廣大病人之權(quán)益

39、,From Prof. Jack Li,健康數(shù)據(jù)加值應(yīng)用思維健康與社會(huì)關(guān)聯(lián),社會(huì)結(jié)構(gòu)(Social structure),物質(zhì)環(huán)境(Material factors),勞動(dòng)環(huán)境(Work),心理環(huán)境(Psychological),社會(huì)環(huán)境(Social environment),健康行為(Health behaviors),生理病態(tài)的變化(Pathophysiological changes)器官損害 (Organ

40、impairment),健康 (Well-being)罹病 (Morbidity)死亡 (Mortality),腦(Brain)神經(jīng)內(nèi)分泌與免疫系統(tǒng)的反應(yīng)(Neuroendocrineand immune response),幼兒期環(huán)境(Early life),遺傳因素(Genes),文化因素(Culture),數(shù)據(jù)源(Source)：Social determinants of health,2006,33

41、,Cross-databank analysis,Databank A downloadingEncryption of individual data,Databank B downloadingEncryption of individual data,,,,,,Generating processed collective results,Abolish the downloaded database, only coll

42、ective results can be taken out from isolated area,,,Algorithm of Cross-databank Analysis With Physical Isolation,35,健康數(shù)據(jù)加值應(yīng)用,世代追蹤應(yīng)用,,From 衛(wèi)生署統(tǒng)計(jì)室,健康數(shù)據(jù)加值應(yīng)用,,健康與社會(huì)的關(guān)聯(lián)社會(huì)經(jīng)濟(jì)、勞動(dòng)條件、幼兒期、遺傳、文化等對(duì)健康的影響,衛(wèi)生政策的評(píng)估醫(yī)療、保健、防疫、全民健保政策實(shí)施

43、成效的衡量、評(píng)估與建議,數(shù)據(jù)整合應(yīng)用,36,From 衛(wèi)生署統(tǒng)計(jì)室,Examples of value-added application,Cohort Study for Hemodialysis,38,Source: Translated from Huang SM,Suicide vs. Psychotropic Medication,39,Medical visit in same year: 83.2%,No medical

44、 visit in same year: 16.8%,Psychiatric visit38.3%,Non-psychiatric visit44.9%,No1.3%,Yes17.6%,Never psychiatric medication: 1.3%,Suicide no. (2002-2005): 10,945,PsychotropicDrugs, yes:37.0%,Ever psychiatric medicati

45、on: 35.7%,No27.4%,Source: Translated from Huang SM,20 – 80 Rule?,Gini coefficientY2008 = 0.711Y2003 = 0.696,23% patients account for 80% expense,*X-axis shows cumulated medical expenditure, Y-axis shows cumulated pati

46、ent number (both by % of total)Source: Translated from Huang SM,41,集成健康加值應(yīng)用中心Application of Health Databanks,Phase 1,Phase 2,Distant User,臺(tái)北醫(yī)學(xué)大學(xué)健康暨臨床研究資料加值中心平面圖與現(xiàn)況圖Taipei Medical University,42,左側(cè)隔間使用者坐位8個(gè)，2名管理人員坐位(左邊隔間

47、左下角以及右邊隔間左下角),Health Insurance DB,Cause of Death DB,Cancer Register DB,Household (census) Register DB,Integrated Data Center for Bio-medical Informatics,Public Health Administration,Epidemiology,Health CareManageme

48、nt,,Others,? ? ?,,,Limited Data Set,Continuity Research,Tool Kits,User,主動(dòng)式具有分析評(píng)估能力的主題式資料架構(gòu) 2010Data Architecture by Subjects with Active Analysis and Assessment,,Regular Data Released,,,Provide Data,,Redundant /Feedbac

49、k,,,,未通過(guò)專法，本中心不直接釋出資料,De-identification,Data Released after IRB approval,,,,,,,,資料庫(kù),生統(tǒng)報(bào)表資料集:報(bào)表資料/彙整資料集,臨床研究資料集:線上即時(shí)分析報(bào)表,次級(jí)資料/資料超市,連結(jié)資料庫(kù),,ETL工具,資料查詢與維護(hù),行政院衛(wèi)生署統(tǒng)計(jì)室,,,,,糖尿病確診後罹患為肝癌之預(yù)測(cè)模型-預(yù)測(cè)表現(xiàn),糖尿病確診後罹患為肝癌之預(yù)測(cè)模型應(yīng)用系統(tǒng),CLOU

50、D COMPUTING FOR PERSONALIZED HEALTH CARE Achieving Meaningful Use of EMR/PHR,,Meaningful Use of Health/Medical Information -- 4P Medicine,PersonalizationParticipationPrediction PreventionMore PsHealthcare Promotion

51、Precision medicinePayment system,Dr. Leroy E. Hood,49,A Definition of Personalized Medicine,Personalized medicine is the use of information from a patient's Phenotype/genotype to: initiate a preventative measure ag

52、ainst the development of a disease or condition, or select the most appropriate therapy for a disease or condition that is particularly suited to that patient.,Definition paraphrased from www.wikipedia.orgOther sources:

53、 Jones, D. Nature Reviews Drug Discovery 2007; 6:770-771; Katsanis et al. Science 2008; 320(5872):53-54; Feero et al. JAMA 2008; 299(11):1351-1352,健康照護(hù)雲(yún)端運(yùn)算服務(wù)--A Personalized Wellness Ecosystem on Cloud,50,個(gè)人健康管理議題是全球健康醫(yī)療

54、關(guān)注的焦點(diǎn)Pervasive Personal Health Management ServiceContext aware health monitoring健康監(jiān)測(cè)Personal Health-aware devices個(gè)人裝置Intelligent alert management智慧管理Pervasive lifestyle incentive management生活方式Pervasive access to h

55、ealthcare information健康資訊Preventive Care & Chronic Disease Mgmt疾病管理Social Health Promotion社會(huì)健康,Source：Pervasive Healthcare as a Scientific Discipline, Methods Inf Med 2008.,,The importance of this projectBuil

56、d infrastructure so that citizens own their health record and receive basic health care services at the right time and right placeFee for illness ? Fee for healthBenefitReduce the waste in medical resourcesImp

57、rove healthcare qualityPromote the health for all citizens,Meaningful Use of EMR,51,Business Model Focus on healthcare industry,Interventions to improve disease risk factors,To help person achieve these goals.In the p

58、ast a brief word of advice from one’s physician at the annual checkup. e.g., avoid smoking, exercise, and eat healthy foods.Big data offeroutside of the clinic in a personalized manner.more sophisticated program wou

59、ld include algorithms that provide personalized feedback to assist with behavior modification at key moments of decision making.e.g., suggesting healthy recipes while the patient is shopping; encouraging exercise at the

60、 end of the workday, or giving a personalized warning about location based environmental triggers for asthma,Example 1: Big Data and Physical Activity,Smartphone apps that have the potential to passively and continuously

61、 track physical activity.More detail datahow physical activity is affected by the social and environmental context.Directly help real-time reminders to increase physical activity before the end of an unusually sedent

62、ary day to avoid missing one’s daily activity target.linking groups in order to increase motivation.Donaire-Gonzalez D, et al. J Med Internet Res, 2013,Example 2: Big Data and Asthma,Sensor snaps onto asthma metered-d

63、ose inhalers, that passively captures the time, location, and GPS coordinates of inhaler use by communicating with a smartphone.App allows users to provide further contextual information, such as symptoms, perceived tri

64、ggers, activity at time of use, and whether.Creating a data feedback loop to improve adherence behavior.Reducing asthma symptoms and improved control.city of Louisville, Kentucky, has adopted this technology to addres

65、s their elevated asthma burden.Van Sickle D, et al. Resp Drug Delivery Europe, 2013MacDonald C. The Environmental Magazine, 2012,PERSONALIZED MEDICINE,It is estimated in 2014, a personal Genome can be sequenced under $

66、1,000 USD3 billion DNA and 33K genes ? more than 100K proteins ? metabolic pathways ? all the functions of body,From Jack Li, TMU,Cloud computing will quickly change the use of medical information,The fact that Goog

67、le and Microsoft are heavily invested “in the cloud” extends to their new offerings for medical records services, such as Microsoft’s HealthVault and Google Health.Google 23andMe, 3.9 million USD and more, The integrati

68、on of biological information, the use of new technology to establish a standardized DNA database, work with pharmaceutical and biotech industry to develop new drugs and personal medicine, Alzheimer’s foundation, Direct-t

69、o-Consumer research: recruit 10,000 patients with Parkinson’s disease to enroll. Brin’s Search for a Parkinson’s Cure, Brin proposes a different approach, one driven by computational muscle and staggeringly large data se

70、ts. For example, a mutation to the GBA gene is 5 times more likely to have Parkinson’s 23and Me: Parkinson’s Genetics initiative1. Tool Construction: Survey 2. Recruitment: 10,000 subjects with Parkinson’s. 3. Data a

71、ggregation: Community members DNA analyzed and surveys. 4. Analysis: database query based on 3,200 subjects. The results are returned in 20 minutes. 5. Presentation: People with GBA are 5 times more likely to have Parkin

72、son’s. Total time elapsed: 8 monthsTraditional Model1. Hypothesis: 2. Studies: 3. Data aggregation: 5,500 subjects 4. Analysis: 5. Writing: 6. Submission: 7. Acceptance: NEJM 8. Publication: The paper notes that peopl