Big-Data-Analytics-In-Medicine-大数据分析-公开课课件.pptx

1、1Big Data Analytics In MedicineBig Data Analytics and Drug Side Effects2BigBig DataData AnalyticsAnalytics study has found an associationassociationbetweenbetween the use of proton-pumpproton-pump inhibitors质子泵抑制剂and thelikelihoodlikelihood of incurring a heartheart attackattackShah,SH.Clopidogrel D

2、osing and CYP2C19.Medscape.July 1,2011.Medscape web site.Heart AttackElectronic Health Records(EHRs)of nearlynearly 3 3 millionmillionpeoplepeople and trillionstrillions ofof piecespieces of medical dataBigBig DataDataAlzheimers Disease EvolutionBig Data Analytics and Disease Predisposition体质BigBig

3、DataDataResearchers analyzedanalyzed moremore thanthan 7,700 brainimages from 1,171 people in various stages ofAlzheimersAlzheimers progressionprogression using a variety oftechniques including magnetic resonance imaging(MRI)and positron emission tomography(PET).BigBig DataData AnalyticsAnalytics st

4、udystudy hashas foundfound that changes inbloodblood flowflow are the earliest known warning sign ofAlzheimers.https:/ Data:Medical ImageLarge volumesvolumes ofof datadata produced by ImagingImaging TechniquesTechniquesVolume of medicalmedical imagesimages is growinggrowing exponentiallyexponentiall

5、yAnnotatedAnnotated datadata or structured methods to annotateannotatemedical images is challengingchallengingBigBig DataData AnalyticalAnalytical Methods allow for the interpretabilityinterpretabilityof depicted contentsScalableScalable MethodsMethods for collecting,compressing,sharing,and anonymiz

6、ing medical dataMedicalMedical ImageImage Processing from Big Data Point of View4Electronic Health Records(EHR)from Big Data Point of ViewBig Data:Clinical DataLarge volumesvolumes of Clinical Data that needs to bestoredstored,retrievedretrieved,and aggregatedaggregatedScalableScalable MethodsMethod

7、s for collecting,compressing,sharing,and anonymizing medical dataScalableScalable MethodsMethods for signal processingprocessing and fordeveloping Big Data based clinical decision supportsystems(CDSSCDSSs)5Genomics from Big Data Point of ViewBig Data:Genomics DataHuman GenomeGenome consists of 30,00

8、0 to 35,000 genesScalableScalable MethodsMethods for pathway analysisanalysis and for the discoverassociationsassociations between observed gene expression changes andpredictedpredicted functional effectsScalableScalable MethodsMethods for Reconstruction of MetabolicMetabolic andRegulatoryRegulatory

9、 Networks67Challenges of Big Data Management and Analytics8Challenges of Big Data Management and Analytics9Big Data PipelinePrecision Medicine14MedicalMedical modelmodel that proposes thecustomization of healthcarehealthcare,with medicaldecisionsdecisions,practicespractices,or productsproducts being

10、tailored to the individualindividual patientpatientPrecision Medicine15Stratified MedicineTargeted TherapyPersonalized MedicineP4P4 MedicineMedicine:Personalized,Predictive,Preventive,ParticipatoryMedicalMedical modelmodel that proposes thecustomization of healthcarehealthcare,with medicaldecisionsd

11、ecisions,practicespractices,or productsproducts beingtailored to the individualindividual patientpatientPilot 1:Lung CancerMotivation:Lung cancer among the most common and deadly diseases costly cancers Lung cancer is a heterogeneousdisease.Characteristics differ among patients tumor regions22iASiS

12、will enable:Discovery of correlations amongtumor spread,prognosis,response totreatment Unraveling molecular mechanismsthat predict response to differenttumor types(signatures)Big Data in the Lung Cancer Pilot Pharmacological knowledge extractedfrom publicly available datasets Biomedical ontologies a

13、nd taxonomies terminology standardization semantically describing the EHRs EHRs in Spanish PET/CT Images Genomic Data/Liquid BiopsySamplesPilot 2:Alzheimers DiseaseMotivation:Approximately,10%of people over65 suffer from Alzheimersimpedes accurate diagnosis andtreatmentsiASiS will enable:Discovery o

14、f patterns associated withprognosis,outcomes,and responseto treatments Heterogeneity of the symptoms Association of medical and lifestyleadvice to Alzheimers risk and stagesof severity EHRs in English MRI Brain Images Genomic Data Pharmacological knowledge extractedfrom publicly available datasets B

15、iomedical ontologies and taxonomies terminology standardization semantically describing the EHRsBig Data in the Alzheimers Disease PilotClinicalNotesNLPDataPreprocessingMedicalImagesDeep LearningPredictiveModelsClinical Big Data AnalyticsMedicalVocabulariesKnowledgeData MiningKnowledgeGraphGenomic B

16、ig Data AnalyticsIdentification ofRNAs regulatedby RBPComparisonwith availableinformationIntegration withtranscriptomicdataHospital-deriveddataKnowledgeGraphIdentification ofkey genes andinteractionsOpen Big Data AnalyticsHeterogeneous open dataSemantic indexing of the data via ontologies and thesauriKnowledge extraction from the data NLP and network analysis technologiesBig Data Management&Analytics34Big Data Management&Analytics35Big Data Management&Analytics36Big Data Management&Analytics37