解锁有效利用现实世界数据的关键

Table of contentsReal World Data: navigate a crucial, complex ecosystemto generate insightsRWD and its usesFrom Real World Data to Real World EvidenceFocus on main RWD sourcesProfile of the main Real World Data sourcesReal World Data possible usesRWD in a nutshellData type relevance by research questionReal World Data challengesAddressing some of the FAIR principlesData findabilityData accessibility and interoperabilityData interoperability and reusabilityInnovative data access and usesMulti country studiesEnriched studiesArtificial Intelligence and machine learningGlossary 12224791011111114171922232528 iqvia.com | 1Maximizing the potential of Real World Data for insightsand evidence generationTraditional Randomized Clinical Trials (RCTs) are widely recognized as a reliable framework for generating evidence on theeffectiveness and safety of health technologies for regulatory purposes. However, when alternative sources of evidence,such as Real World Data (RWD) from diverse settings are the only option, the rigid processes of RCTs pose a challenge.Consequently, health authorities and national health systems worldwide are increasingly accepting the need to movebeyond relying solely on RCT data.… ALONG WITH CHANGES IN HOW IT IS CREATEDIn the rapidly growing Big Data universe, unprecedentedamounts of person-level information contained in ElectronicHealth Record (EHR) systems contributes to the fastestcompound annual growth rate of 36% forecasted till2025 (growth of volume of health related data).1The dataoriginates from primary, specialist and hospital care, drugand disease registries, medical devices, digital apps, andother sources of health data which are generated every day.VOLUME IS GROWING IN ABSOLUTE TERMS …36%30%26%25%27%Healthcare2018–2025 data – compound annual growth rate (CAGR)Source:1Coughlin et al Internal Medicine Journal article “Looking to tomorrow’shealthcare today: a participatory health perspective”. IDC White Paper,Doc# US44413318, November 2018: The Digitization of the World – From Edge to Core”.ManufacturingFinancialserviceMedia andentertainmentGlobaldatasphereThey include structured data in the form of diagnoses,medication, laboratory test results, etc. with differentlevels of granularity. They also include unstructureddata in clinical narratives, for example all of whichlikely contain invaluable insights for most researchprojects along the development lifecycle, some of thembecoming essential for regulatory purposes. Thesedatasets are often siloed by country, language, region,hospital and even department. Real World data alsooriginate from a multitude of disease specific contextsand so vary greatly in complexity.As a consequence of this great diversity, it is broadlyacknowledged that state-of-the-art tools are neededto make such healthcare data Findable, Accessible,Interoperable and Reusable (as per the FAIR principlesinitially published in March 2016 by a group ofscientists (Mark D. Wilkinson) in Scientific Data);opening the door to maximizing the use of Real-WorldData (RWD) in the regulatory context and beyond.FindableF.A.I.R.1Institutions/Health SystemsReference dataSales/ConsumptionClaimsResearchexperience1Scientific Data: 15 March 2016 The FAIR Guiding Principles for scientific data management and stewardship – Mark D. WilkinsonCountriesWholesalersTrialregistriesSalesDataExtraction,Curation, Privacyand SecurityTechnologiesPhysiciansReference dataResearchexperiencePrescriptionsChecklistMulti sourcesLab/BiomarkersPatientsSocial mediaMortalityDevicesPharmacyClaimsEMRClinicaltrialsPRORegistriesChartReviewGenomicsImagingWearablesAccessibleInteroperableRe Usable 2 | IQVIA Real World Data Assets: Navigate a crucial, complex ecosystem to generate insightsReal-world data and analytics are generally robust and widely available today across many stakeholders, butgenerating meaningful evidence remains complex and elusive. This is due to contingent factors, such as theheterogeneous level of data sources quality and the increased usage of AI techniques that are not yet framed by acode of conduct.The image above clearly differentiates between Real World Data and Real World Evidence: RWD needs to undergotwo types of processes in order to potentially become evidence:• They initially need to be cleaned, curated, harmonized and de-identified•They can then be analysed via diverse approaches, ranging from descriptive statistical models to complexArtificial Intelligence algorithmsReal World Data(RWD)Patient level data routinelycollected for medical and/orhealth administration purposesversus data collected inconventional randomizedcontrolled clinical trials(RCTs)Evidence (RWE)claims and address evidenceProcessingProcess (curate /harmonize / de-identifyRWD then analyse it usingappropriate techniquesReal World Data and its uses— from Real World Data to Real World EvidenceFocus on main RWD sourcesOne of the main reasons behind the initial data processing needs outlined above is that the initial intentunder