IMPLEMENTING BIG DATA IN TODAY S HEALTH CARE PRAXIS: A CONUNDRUM TO PATIENTS, CAREGIVERS AND OTHER STAKEHOLDERS - WHAT IS THE VALUE AND WHO PAYS 29 OCTOBER 2015 DR. DIRK J. EVERS
BACKGROUND TreatmentMAP Treatment Decision Support in Oncology
TREATMENT MAP Treatment Decision Support in Oncology
SEQUENCING IS AFFORDABLE BUT IS IT CLINICALLY APPLICABLE? nanograms of DNA 3,000,000,000 genomic positions 20,000 genes each cancer is molecularly different hundreds of cancer types countless drug combinations and co-medications complex comorbidities, medical conditions and adverse events > 8,000 approved medications * > 100,000 clinical studies * > 23,000,000 scientific publications *
WITH ONE PHYSICIAN HAVING TO MAKE SENSE OF ALL OF IT
AND AN INCREASING DEMAND
TreatmentMAP BRINGS IT ALL TOGETHER AND GBs of DNA sequencing data Demographics, indication, comorbidities, co-medication TreatmentMAP Nucleus
LEVELS OF EVIDENCE Treatment Decision Support based on Clinical Evidence - Guidelines, Off-label, Trials - Actionable mutations in 30-70% of cases depending on cancer type What about the huge number of variants of unknown significance (VUS)? - In cancer drivers, in druggable pathways
TreatmentMAP ANALYTICAL PROCESS Data generation & Genome analysis Evidence mining Clinical Interpretation Reporting Test requisition NGS Panel/ WES Test Genomic aberrations Actionable Biomarkers Biomarker & Drug Prioritization Derisking using SafetyMAP TreatmentMAP Report Analysis of @3 million oncology publications >320 Cancer Drug Targets >600 Genes >160 Predictive Biomarker Genes Observed Text Data Analytics World of Biomedical Knowledge Predicted Integration of the Worlds Biomolecular Databases Clinical meta-analysis of biomarker data Clinical Validity Endorsed by KoL Validation Clinical + #Cases context + Effect + = Patient Disease Other RECIST OS AE s Evidence Level >350 Cancer Drugs >120 FDA approved Observed in >100 Cancer Indications Endorsed Biomedical Curation Bioinformatics Analysis Clinically Observed Patient Disease Other RECIST OS AE s >230 IND s >5000 clinical trials Clinical Translational Single nucleotide variants, Polymorphisms, Insertions/ Deletions, Fusion Proteins, Copy Number Variants Genotypes & Outcomes Databases Predicted Disease Drivers Translational /Predicted Patient Disease Other RECIST OS AE s 9
DELIVERS A COMPREHENSIVE REPORT
PROVIDING EVIDENCE BASED TREATMENT DECISION SUPPORT IN MULTIPLE DIMENSIONS Evidence based prioritization Drugs with increased Approved drugs Toxicity / Adverse Events Off-label drugs Drugs in development,trials Ineffective drugs
ENGINES, TOOLS AND APPLICATIONS Cohort Analytics trial matching, patient profile matching, translational research Treatment Decision Support in Oncology Safety & Efficacy optimization Enabling Data structuring & referencing data Cohort Analytics trial matching, quality metrics, patient matching, research Safety Analytics Cohort analytics, mechanism based safety research, signal detection & prediction De-Risking poly-pharmacy derisking, safety biomarker testing, disease risk assessment, genetic marker testing Computer Linguistics, NLP InsightMAP TreatmentMAP Ontology & Dictionary Services OutcomesMAP SafetyMAP RiskMAP Nucleus comprehensive biomedical annotation warehouse
AN EXAMPLE FDA S AERS DATA 9468 15723 1935 515 (ATC level 4) AERS >7,000,000 SafetyMAP Cases 7 931 282 1095 2015 Molecular Health GmbH. All rights reserved. Confidential.
HEALTHCARE REGULATION CHALLENGES IN INNOVATION PART I: FACTS
TWO DIFFERENT MARKETS TWO DIFFERENT REGULATIONS EU: Software as Medical Device US: Laboratory Developed Test out of a CAP/CLIA certified Lab
US: LDT END TO END SERVICE Certified Lab Molecular Data Sample logistics & prep Signal Generation (Lab) Patient Clinical Data Data Analysis Software as Medical Device Medical Report Molecular Pathologist Clinical decision Treatment Efficacy Patient Safety
EU: SOFTWARE AS MEDICAL DEVICE Molecular Data Patient Sample logistics & prep Clinical Data Signal Generation (Lab) Data Analysis & Results Presentation Software as Medical Device Medical Report Clinical decision Treatment Efficacy Patient Safety
MEDICAL DEVICE REGULATION (EU) Medical devices in the EU are regulated by the medical device directive 93/42/EEC (MDD) Provides definition and classification of medical devices and is the basis for a harmonized legislation for safe and effective medical devices Medical devices must fulfill the essential requirements (93/42/EEC Annex1) The MDD establishes common ground but develops much slower than innovations in the field The MDD therefore provides a precise but general definition of medical devices All provisions of the MDD can and must be applied to the particular medical device
Application of the MDD provisions to innovative devices are challenging!!! Regulation moves slowly but you can t escape it!
MEDICAL DEVICE REGULATION AND INNOVATION (EU) THE CHALLENGE Application of the MDD provisions to innovative devices can be challenging E.g.: Ever growing body of scientifc evidence that is e.g. integrated in TME knowledge base has direct impact on quality of product as defined in the intended use, if data is not updated regulary. E.g.: Even though being in scope of the MDD stand-alone software medical devices that are provided as SaaS need tailored definitions for vigilance activities Understanding of innovative devices by authorities can be challenging. For TME as one of the first stand-alone software medical devices we received the question: To which medical device your software is belonging to?
REGULATORY LANDSCAPE MDD 93/42 requires MPG adds to MPV MPSPV requires requires requires requires SW Lifecycle Quality management Risk management Usability certifies certifies certifies certifies requires requires requires IEC 62304 ISO 13485 ISO 14971 IEC 62366 requires Source: C. Johner, CPMS Seminare
TRANSLATION INTO QM SYSTEM
THE RESULT
HEALTHCARE REGULATION CHALLENGES IN INNOVATION PART II: FUTURE
MEDICINE BECOMES AN INFORMATION SCIENCE Heinrich Nixdorf Museum: Moore s Law
DATA GROWTH BEYOND MOORE S LAW... Molecular profiling Personal health data Electronic Medical Record Biomedical Publications Clinical Studies Biomedical Data e.g. sequencing costs e.g. EMR adoption e.g. biomedical citations
WILL POWER HEALTHCARE MEGATRENDS Healthy Consumer Diseased Patient Empowerment of individual Commoditization, democratization and automation of diagnosis the economic product of healthcare Enabling the personalized health life-cycle with more precise and cost-effective prevention, diagnosis and care
STILL TODAY BIOMEDICAL PROGRESS IS SLOWING DOWN
AND THE NON-RESPONDER PROBLEM PREVAILS Anti Depressants Asthma Diabetes Arthritis Alzheimer s Oncology
BECAUSE WHILE EACH PATIENT IS DIFFERENT...
THEY ARE MOSTLY TREATED THE SAME WAY
RESULTING IN AN INADEQUATE ECONOMIC BURDEN (US only) *Projections - Source: Sean Keehan and others (2008): Health Spending Projections Through 2017: The Baby Boom Generation is Coming to Medicare
IT S TIME FOR A PARADIGM SHIFT TURNING DATA INTO BETTER HEALTHCARE Ø the practice of medicine (i.e. observational/empiric healthcare) to Ø the science of medicine (i.e. data-driven evidence-based healthcare)
DOUBLE BLIND STUDY CONCEPT DOESN T FIT DATA COMPLEXITY
ADDRESSING THE CHALLENGE: e.g. ASCO CancerLinQ
DO WE NEED AN ALTERNATIVE TO CLINICAL TRIALS? Clinical studies today are based on small cohorts. With Big Data technologies we have the option extend the evidence base to virtually all patients. Statistical analysis of e.g. efficacy, toxicity in relation to specific genetic alterations on a much larger dataset than in classical clinical studies (addressing the n vs N problem). Challenge: The regulation for clinical studies today is very strict an tailored to a trial approach that requires for good reasons e.g. an ethics vote. Based on data obtained from patient care, self learning systems can continuously improve the quality of treatment decision support.
DO WE NEED AN ALTERNATIVE TO CLINICAL TRIALS? Big Data technologies allow to extend the evidence base to virtually all patients. Statistical analysis of e.g. efficacy, toxicity in relation to genetic alterations will be possible on a much larger dataset (addressing the n vs N problem). Challenge: The regulation today is very strict and tailored to a classical trial approach. Based on data obtained from patient care, self learning systems can continuously improve the quality of treatment decision support.
UNLOCKING BIG DATA IN PATIENT GENOMICS 38
HOW CAN WE TAP INTO THE 97%? 39
THANK YOU! 2015 Molecular Health GmbH. All rights reserved. Confidential. 40
AN EXAMPLE FDA S AERS DATA 9468 15723 1935 515 (ATC level 4) AERS >7,000,000 SafetyMAP Cases 7 931 282 1095 2015 Molecular Health GmbH. All rights reserved. Confidential.
UNLOCKED IN SafetyMAP 2015 Molecular Health GmbH. All rights reserved. Confidential.