Roche Holding Ltd · SIX:RO
Roche — Investor Day (November 2023)
97 slides from Roche's investor day presentation, Nov 29, 2023
Slide 1: Roche Digitalization Day
Roche Digitalization Day Virtual IR Event 29 November 2023
Slide 2: Forward-looking statements
This presentation contains certain forward-looking statements. These forward-looking statements may be identified by words such as ‘believes’, ‘expects’, ‘anticipates’, ‘projects’, ‘intends’, ‘should’, ‘seeks’, ‘estimates’, ‘future’ or similar expressions or by discussion of, among other things, str
Slide 3: Introduction
Introduction Bruno Eschli Head of Investor Relations
Slide 4: Agenda
Agenda Introduction Bruno Eschli, Head of Investor Relations Informatics: Driving integrated data platforms and infrastructure Alan Hippe, Chief Financial and Information Officer Roche Information Solutions (RIS): Enabling laboratory efficiency & advanced clinical decision support Moritz Hartmann, G
Slide 5: Digitalization along the value chain
Digitalization along the value chain Use cases picked from Group functions, Pharmaceuticals and Diagnostics Group functions: Digital infrastructure 1 2 Treatment Clinical Regulatory & Manufacturing support for Early R&D Commercialization Diagnosis development reimbursement & distribution physicians
Slide 6: Digitalization along the value chain
Digitalization along the value chain What digitalization could deliver to Roche, patients & healthcare systems Group functions: Digital infrastructure Treatment Clinical Regulatory & Manufacturing support for Early R&D Commercialization Diagnosis development reimbursement & distribution physicians &
Slide 7: Digitalization and recent progress in GenAI: Where do we stand?
Digitalization and recent progress in GenAI: Where do we stand? Trying to generate a cover slide for today using OpenAI’s DALL-E “alphafold protein structure with a circuit board in the back” “a DNA helix and a light blue circuit board in the back” “show two separate things: a slightly blurred prote
Slide 8: Informatics: Driving integrated data platforms and infrastructure
Informatics: Driving integrated data platforms and infrastructure Alan Hippe Chief Financial and Information Officer
Slide 9: Technology trends driving the next wave of opportunities
Technology trends driving the next wave of opportunities Finding the right differentiating and disruptive opportunities to reshape our future Current digital waves Opportunities in healthcare • Artificial intelligence 11 12 13 14 • Data fluidity • Internet of things • AI-driven research 15 17 20 • A
Slide 10: Digital maturity in the healthcare industry is improving
Digital maturity in the healthcare industry is improving Strong investments in AI driven drug discovery Second highest increase of the Funding in AI driven drug discovery companies Digital Acceleration Index (DAI) by industry1 increasing substantially?2 1 BCG Digital Acceleration Index (DAI), global
Slide 11: Roche's global informatics network supporting the business
Roche’s global informatics network supporting the business Accelerating digital transformation and innovation through leading tech capabilities Pharmaceuticals + Diagnostics Switzerland (Basel/Kaiseraugst, Rotkreuz) Penzberg, Hillsboro Mannheim Barcelona Warsaw Mississauga Indianapolis Pharmaceutica
Slide 12: Transforming informatics: Building an interoperable digital core
Transforming informatics: Building an interoperable digital core Accelerating delivery of value through integration, technology and re-usable data Old structure: New structure: 4 separate units Interoperable digital core supporting business solutions Integrated business solutions Science & Clinical
Slide 13: Roche forming strong alliances with leading tech companies
Roche forming strong alliances with leading tech companies in the healthcare space Develop digital endpoints as regulatory approved First-of-its-kind collaboration to accelerate standards of outcome measurement, and design of drug discovery using generative AI next generation wearable technology Dig
Slide 14: Roche recognized as industry leader through digital innovation and talent
Roche recognized as industry leader through its digital innovation and talent Ranked #1 pharma company due to leading levels of Genentech ranked #1 best place to work in IT by AI innovation1 Computerworld2 Top-scoring pharma companies rated on their leading levels of A testament to the exceptional t
Slide 15: Digitalization along the value chain
Digitalization along the value chain Group functions: Digital infrastructure 1 2 Treatment Clinical Regulatory & Manufacturing support for Early R&D Commercialization Diagnosis development reimbursement & distribution physicians & patients 11 12 13 14 18 19 20 7 8 9 10 3 4 5 6 15 16 17 1. GALILEO: G
Slide 16: GALILEO: Creating a global generative AI & LLM strategy for Roche
1 GALILEO: Creating a global generative AI & LLM strategy for Roche Establishing next generation AI & LLM capabilities Driving our generative AI strategy forward… … by following concrete actions Follow the value • Develop highest value use cases across divisions and functions Accelerate & scale best
Slide 17: GALILEO: Roche's global generative AI & LLM strategy
1 GALILEO: Roches’s global generative AI & LLM strategy Providing secure access for all employees to accelerate & scale best opportunities LLM powered apps Gen AI applications (for process & business innovations) GALILEO Model marketplace Operating System (for access to LLMs) (GalileoOS) Responsible
Slide 18: ASPIRE: Creating the digital process backbone for Roche
2 ASPIRE: Creating the digital process backbone for Roche Most comprehensive digital transformation program in the industry Value chain processes Idea to product (lifecycle) 1 Amongst SAP’s ~ 500 enterprise-wide Marketing plan to cash (commercial) largest Roche program program / Plan to deliver prod
Slide 19: ASPIRE: Integrated & standardized system
2 ASPIRE: Creating the digital process backbone for Roche Integrated & standardized system enabling analytics and data-driven decisions GEP SMART SAP Customer GEP Nexxe (myBUY) (iCollab) Web Portal (C/4HANA) OMP Supply Chain Planning Conga & Zilliant Integration Quote to Contract SAP Global Trade Se
Slide 20: Roche Information Solutions (RIS)
Roche Information Solutions (RIS): Enabling laboratory efficiency & advanced clinical decision support Moritz Hartmann Global Head of Roche Information Solutions, Roche Diagnostics
Slide 21: Digitalization along the value chain
Digitalization along the value chain Group functions: Digital infrastructure 1 2 Treatment Clinical Regulatory & Manufacturing support for Early R&D Commercialization Diagnosis development reimbursement & distribution physicians & patients 11 12 13 14 18 19 20 7 8 9 10 3 4 5 6 15 16 17 1. GALILEO: G
Slide 22: Roche Insights Business (RIS)
Roche Insights Business (RIS) Enabling laboratory efficiency & clinical decisions through digitalization Diagnostics Pharmaceuticals #1 in in vitro diagnostics A leading provider of medicines and pathology Committed to worldwide 29 billion diagnostic tests accelerating the power Over 14 million pati
Slide 23: Roche becoming a leading player in digital health
Roche becoming a leading player in digital health Informing treatment decisions with insights across healthcare settings Segment Ambition Uniqueness Provide differentiated patient Remote patient Extend clinical workflows and clinical decision monitoring solutions for management support with integrat
Slide 24: RIS value propositions
RIS value propositions Integrate data sources, improve efficiency and impact patient care Medical insights Enable confident care decision & impact patient outcomes Operational excellence Improve workflows & operational efficiency Digital infrastructure Integrate data sources & harmonize data streams
Slide 25: Roche's commercially available portfolio
Roche’s commercially available portfolio Lab Primary care & hospital Home 3 Medical insights DP Algorithms algorithms* CDS apps: Guidelines*, Publication Search*, Digital biomarker Parkinson’s** Clinical Trial Match* 2 Operational excellence navify Monitoring navify Inventory navify Analytics navify
Slide 26: Roche's commercially available portfolio
3 Roche’s commercially available portfolio Use case on building a shared digital infrastructure & marketplace open to 3rd parties Lab Primary care & hospital Home 3 Medical insights DP Algorithms algorithms* CDS apps: Guidelines*, Publication Search*, Digital biomarker Parkinson’s** Clinical Trial M
Slide 27: navify Integrator including Marketplace & Portal
3 navify Integrator including Marketplace & Portal Harnessing the power of data and enabling access to our digital portfolio operational excellence and medical insights portfolio Easy access a broad range of digital Roche and non-Roche products through: navify navify Algo Clinical CDS 3rd party 3rd
Slide 28: Roche's commercially available portfolio
4 Roche’s commercially available portfolio Use case on providing advanced software solutions to increase lab efficiency Lab Primary care & hospital Home 3 Medical insights DP Algorithms algorithms* CDS apps: Guidelines*, Publication Search*, Digital biomarker Parkinson’s** Clinical Trial Match* 4 2
Slide 29: navify operational excellence applications
4 navify operational excellence applications Achieving operational excellence through integrated solutions navify Lab operations navify Sample tracking Sample results and data API offering to connect 3rd management party solutions to trace the pre- pre-analytic path of every sample navify Analytics
Slide 30: Roche's commercially available portfolio
5 Roche’s commercially available portfolio Use case on developing advanced algorithms to support clinical decision making Lab Primary care & hospital Home 3 Medical insights DP Algorithms algorithms* CDS apps: Guidelines*, Publication Search*, Digital biomarker Parkinson’s** Clinical Trial Match* 5
Slide 31: navify Algo Suite: Go-to for medical value algorithms
5 navify Algo Suite: Go-to for medical value algorithms Providing trusted diagnosis & decision support for clinicians A one-stop shop cloud-based platform Lab Algorithm library Algorithm is embedded into the clinical workflow Patients Algorithm output informs the clinical decision Partners Hospital
Slide 32: navify Algorithm Suite: Algorithm menu
5 navify Algorithm Suite: Algorithm menu Expanding clinical decision-making across multiple indications 2023 2024 ROMA* Briganti nomogram* GastroFlag** Oncology GAAD Roach* PREVAIL* LungFlag** ColonFlag* ALBI Score* MELD Na ADAPT-ProC3** LGI-Flag* Cardiology Heart Score* ADAPT Protocol* HAS-BLED* LA
Slide 33: GAAD algorithm
5 GAAD algorithm Aiding diagnosis of early stage hepatocellular carcinoma (HCC) Diagnostics GAAD algorithm Pharmaceuticals Sex cobas® e Age Analyser Treatment navify® Algorithm (e.g. Roche) Elecsys® Suite Elecsys® Report to aid follow-up AFP (+ navify AFP evaluation & Integrator) treatment Elecsys®
Slide 34: Roche's commercially available portfolio
6 Roche’s commercially available portfolio Use case on developing digital biomarkers to track disease progression 24/7 Lab Primary care & hospital Home 3 Medical insights DP Algorithms algorithms* CDS apps: Guidelines*, Publication Search*, 6 Digital biomarker Parkinson’s** Clinical Trial Match* 2 O
Slide 35: Remote Patient Management Solution
6 Remote Patient Management Solution* Multi-disease solution to enable better care for patients Remote Patient Management Solution Patient engagement Care team workflow Actionable care insights optimisation Device & Lab data Clinical data dBM data Track vital signs & digital Track relevant Bi-direct
Slide 36: Roche Insights Business (RIS): Our business model
Roche Insights Business (RIS): Our business model 1 Assessment and validation of “willingness to pay” as 2 Monetisation of digital part of product products development process 3 Modularisation of products to allow upsell 4 Products sold as service (annual subscriptions, of future features on recurri
Slide 37: Roche Diagnostics Global Operations
Roche Diagnostics Global Operations: Improving efficiencies in manufacturing and distribution by AI/ML Kent Kost Global Head of Diagnostics Operations, Roche Diagnostics
Slide 38: Digitalization along the value chain
Digitalization along the value chain Group functions: Digital infrastructure 1 2 Treatment Clinical Regulatory & Manufacturing support for Early R&D Commercialization Diagnosis development reimbursement & distribution physicians & patients 11 12 13 14 18 19 20 7 8 9 10 3 4 5 6 15 16 17 1. GALILEO: G
Slide 39: Deploying digital solutions across operations value chain
Deploying digital solutions across operations value chain Unlocking value of digitalization at each step PLAN SOURCE MAKE DELIVER Streamline planning process Connect to suppliers for Improve manufacturing Optimize distribution using real time data to lower rapid demand adjustments efficiency, reduce
Slide 40: Machine learning tools reduce inventory & write-offs
PLAN SOURCE MAKE DELIVER 7 Machine learning tools reduce inventory & write-offs Advanced analytics and demand sensing EU Automated forecasting Forecast accuracy affiliate +7x +9pp +14% • Machine learning engine for time-series forecasts • Leverages classical statistical models and modern machine lea
Slide 41: Digital risk monitoring improves supply security
PLAN SOURCE MAKE DELIVER 8 Digital risk monitoring improves supply security Automated and real-time supplier risk management • Closer supplier risk management needed based on 1800+ suppliers monitored daily for learnings from the pandemic, supply chain disruptions and proactive risk control and miti
Slide 42: AI-enabled manufacturing efficiency & quality improvement
PLAN SOURCE MAKE DELIVER 9 AI-enabled manufacturing efficiency & quality improvement Manufacturing yield improvement for Point of Care sensors Before digitalization Point of Care sensor manufacturing yield (%) ● Low manufacturing yield upon product transfer resulting in high costs ● Product and proc
Slide 43: Harnessing transportation data insights
PLAN SOURCE MAKE DELIVER 10 Harnessing transportation data insights Enabling timely decisions to optimize the distribution mode mix Data driven transport management ● Transportation mode mix ● Air/sea ratio analytics +8% ● Short-term improvement increase in sea to air ratio opportunities -18% CO2 re
Slide 44: Digitalization driving Diagnostics operations excellence
Digitalization driving Diagnostics operations excellence The future of Roche Diagnostics operations ● Fully digitized operations with predictive analytics ● Streamlined processes through end-to-end connection ● Enhanced speed of execution and cost efficiency through digital optimization 44
Slide 45: gRED Lab-in-a-loop
gRED Lab-in-a-loop: Multiplicative levers to transform drug development Aviv Regev Head of Genentech Research and Early Development (gRED)
Slide 46: Digitalization along the value chain
Digitalization along the value chain Group functions: Digital infrastructure 1 2 Treatment Clinical Regulatory & Manufacturing support for Early R&D Commercialization Diagnosis development reimbursement & distribution physicians & patients 11 12 13 14 18 19 20 7 8 9 10 3 4 5 6 15 16 17 1. GALILEO: G
Slide 47: Drug discovery, preclinical assessment, clinical studies
Drug discovery _ Preclinicalassessment _Clinical studies Target discovery
Slide 48: 10! as 2x10! 2) 424x102)
10! as 2x10! 2) 424x102) 193 coc] ceuLtypes 7c] fet variant 2] gene [ee “.i) AND STATES +s". <} | COMBINATIONS |.:,"2,] COMBINATIONS [':,": = Soha BT pee wiles te at wae ee tea ee ee OY SEAR P OTR Soa bet aici See Poets , 10 be 9022 [es 190 ess ANTIBODY |<%-".3 SMALL | SEQUENCES |.:1) MOLECULES ot .
Slide 49: Four multiplicative levers to transform R&D
Four multiplicative levers to transform R&D Human biology High resolution & massive scale Study disease processes directly Do deep and high resolution experiments at in patients or in human derived extreme scale for only a marginally added models cost Therapeutic modalities Artificial intelligence /
Slide 50: Lab in a loop
LAB IN A LOOP Integrating our levers to build the portfolio of the future EXPERIMENT ALGORITHM
Slide 51: Lab in a loop: Integrating our levers
Lab in a loop: Integrating our levers to build the portfolio of the future 51
Slide 52: Lab in a loop: Integrating our levers
Lab in a loop: Integrating our levers to build the portfolio of the future 52
Slide 53: Finding disease cells and targets is essential across R&D
11 Finding disease cells and targets is essential across R&D Primary and additional disease indications Growth conditions for screens Bi/multi specifics targets On-target toxicities 53
Slide 54: First, index the web: SCHub
11 First, “index the web”: SCHub has >200M cells from >2.2K studies Rood et al. Nature Medicine 2022 54
Slide 55: Next, you do reverse image search
11 Next, you do “reverse image search” 55
Slide 56: SCimilarity: Reverse cell search using deep learning model
11 SCimilarity: “Reverse cell search” using deep learning model Answer: Query cell is similar to macrophages in the catalog of labeled reference cells Heimberg et al. BioRxiv 2023 56
Slide 57: Portfolio impact: Target OSM/OSMR pathway in IPF and IBD
11 Portfolio impact: Target OSM/OSMR pathway implicated in both IPF and IBD through different cells Original goal: Antagonizing OSMR to prevent progressive lung In which tissues and diseases can we find OSMR-expressing fibrosis inflammatory fibroblasts? Deep metric learning (‘reverse cell search’) H
Slide 58: Portfolio impact: Target OSM/OSMR pathway in IPF and IBD
11 Portfolio impact: Target OSM/OSMR pathway implicated in both IPF and IBD through different cells Original goal: Antagonizing OSMR to prevent progressive lung Additional goal: Antagonizing OSMR to prevent IBD fibrosis Hypothesis: Targeting OSMR in IBD to block inflammatory Hypothesis: Blockade of
Slide 59: Lab in a loop: Integrating our levers
Lab in a loop: Integrating our levers to build the portfolio of the future 59
Slide 60: Massively parallel, high content pooled screens for function
12 Massively parallel, high content pooled screens for function Perturb-seq Perturb CITE-seq Optical pooled screens (OPS) In vitro screens Phenotype and in situ barcode readout In vivo perturb-seq & spatial Combinatorial perturbations In vivo screens Dixit et al., Cell 2016; MacFarland et al., Natur
Slide 61: Complex phenotyping at scale: Role of E3 ligases in dendritic cells
12 Complex phenotyping at scale: Role of E3 ligases in immune dendritic cells (DC) PERTURBATIONS • 3,390 guides targeting 1,130 genes • >1M cells profiled • 519,535 with single perturbations • 177,871 with 2+ perturbations Geiger-Schuller, Eraslan et al. BioRxiv 2023 61
Slide 62: A regulatory model of E3 modules controlling programs
12 A regulatory model of 6 E3 modules controlling 11 programs Geiger-Schuller, Eraslan et al. BioRxiv 2023 62
Slide 63: E3 ligase control every aspect of the dendritic cell life cycle
12 E3 ligase control every aspect of the dendritic cell (DC) life cycle Geiger-Schuller, Eraslan et al. BioRxiv 2023 63
Slide 64: The DC regulatory circuit is associated with genetic risk in inflammatory disease
12 The DC regulatory circuit is associated with genetic risk in inflammatory disease in humans Module 6 enriched for Multiple gene programs explain Programs enriched for inflammatory disease GWAS inflammatory common disease heritability rare risk variants in CD Geiger-Schuller, Eraslan et al. BioRxi
Slide 65: Double KOs allow prediction of genetic interactions between E3 ligases
12 Double KOs within and across modules allow prediction of genetic interactions between E3 ligases from sparse data Input: CombVAE: Output: Cells profiled with 2+ perturbations to predict interactions More intra- than inter-module interactions Geiger-Schuller, Eraslan et al. BioRxiv 2023 65
Slide 66: Portfolio impact: Lab in a loop in neuroscience and colorectal cancer
12 Portfolio impact: Lab in a loop for target and compound prediction in neuroscience and colorectal cancer 66 In collaboration with
Slide 67: Lab in a loop: Integrating our levers
Lab in a loop: Integrating our levers to build the portfolio of the future 67
Slide 68: Improving molecule PTS: Lab in a loop in drug discovery
Improving molecule PTS: Lab in a loop in drug discovery 13 Small molecules 14 Antibodies 15 Tumor neoantigen property prediction optimization and generation prediction New ML/AI algorithms for small molecule, antibodies, vaccine design, etc. 68 PTS=probability of technical success
Slide 69: Lab in a loop for small molecule property prediction
13 Lab in a loop for small molecule property prediction 69
Slide 70: Successful prediction of small molecule potency
13 Successful prediction of small molecule potency 2017 ~1.2M compounds used in Increased hit rate in prospective GNEProp addresses key phenotypic HTS for E. Coli screens (membrane permeable tolC challenges: mutant and wild-type) E. coli wild type 0.03% Correctly predict activity of 2021 Trained o
Slide 71: Lab in a loop for antibody generation and optimization
14 Lab in a loop for antibody generation and optimization 71
Slide 72: Steady improvement in Prescient Design antibody optimization algorithms
14 Steady improvement in Prescient Design antibody optimization algorithms Lab-in-a-loop: Progress over rounds Round 9: Affinity results across targets / lead molecules 72
Slide 73: EquiFold diffusion: Fast antibody structure prediction
14 EquiFold diffusion: Fast antibody structure prediction EquiFold: Our protein structure prediction and design model for antibodies Movie: The heavy and light chain particles sorting themselves out and the CDRH3 assembling — General protein/complex model in development (with NVIDIA collaboration) 1
Slide 74: Lab in a loop: Integrating our levers
Lab in a loop: Integrating our levers to build the portfolio of the future 74
Slide 75: Autogene cevumeran personalized neoantigen vaccine
15 Autogene cevumeran (personalized neoantigen vaccine) against patient-specific tumor antigens in solid tumors AI/ML transformer-based model to design cancer vaccine Prime the immune system to generate a response to a specific tumor 75 RNA vaccine in collaboration with
Slide 76: Performance gains are observed across most of the MHC alleles
15 Performance gains are observed across most of the MHC alleles 132 different MHC alleles included in the training F1 scores across alleles Average precision Thrift et al., BioRxiv 2022: HLApollo: A superior transformer model for pan-allelic peptide-MHC-I presentation prediction, with diverse negat
Slide 77: Prediction of MHC II presentation and immunogenicity of therapeutic antibodies
15 Prediction of MHC II presentation and address immunogenicity of therapeutic antibodies (ADAs) Same model — multiple applications — Better joint performance 77
Slide 78: Lab in a loop: Integrating our levers
Lab in a loop: Integrating our levers to build the portfolio of the future 78
Slide 79: Algorithms as tools to enable our creative scientists
Algorithms as tools to enable our creative scientists 79
Slide 80: Three major LLM pillars in Genentech/Roche
16 Three major LLM pillars in Genentech/Roche USE TRAIN DEVELOP GALILEO: Architect, pre-train and fine Solve new problems by tune new LLMs on internal incorporating an LLM into Deploy LLMs (public and data another algorithmic setup private) into our work e.g., combine with Fine tune existing LLM for
Slide 81: USE an LLM to improve our day to day work
16 1. USE an LLM to improve our day to day work Example applications Suggesting analysis Writing / methods for a scientific optimizing code question Annotating Writing step-by-step lab cell types protocols Writing optimal prompts for Summarizing / another AI (DALL-E2) to searching literature make go
Slide 82: TRAIN: Build our own LLMs, leveraging extensive internal data
16 2. TRAIN: Build our own LLMs, leveraging extensive internal data Public LLMs (e.g. ChatGPT) Roche Large Language Model Trained on public corpus Trained on public and Roche corpora of of human language human language and lab data (notebooks, models, experiments); text-only and multi-modal Prompted
Slide 83: TRAIN: Roche LLM
16 TRAIN: Roche LLM • Roche LLM’s trained and tested 0.5B, 7B (potential production model) and 30B (scalability investigation) scale • Internal team; learning from open models, but training from scratch, to emphasize biomedical corpora and use own data • Internally trained models and chat in second-
Slide 84: DEVELOP: Build LLMs to assist our scientists
16 3. DEVELOP: Build LLMs (and beyond) to assist our scientists ChatGPT Roche Large Language Model Trained on public corpus Trained on public and Roche corpora of of human language human language and lab data (notebooks, models, experiments); text-only and multi-modal Prompted with human language Pr
Slide 85: Our AI strategy for R&D
Our AI strategy for R&D Leading Lab in a Loop, Scale & resolution partnerships and integrated acquisitions Partnership Accelerated computing Acquisition Full stack, across all aspects of R&D; up to Maximize benefit of large size: proprietary Top capabilities in house, to ensure full “self drive” leg
Slide 86: Lab in a loop: Integrating our levers
Lab in a loop: Integrating our levers to build the portfolio of the future 86
Slide 87: pRED: Transforming drug development through data & analytics
pRED: Transforming drug development through data & analytics Scott Oloff Global Head of Data & Analytics, pRED
Slide 88: pRED Digital strategy: Driving R&D productivity and innovation
pRED Digital strategy: Driving R&D productivity and innovation Data & analytics driven initiatives from pre-portfolio to early clinical development Data & analytics Systematic R&D engine Smart bets to change how is leading pRED to take productivity increase via we do R&D through data & the best data
Slide 89: Digitalization along the value chain
Digitalization along the value chain Group functions: Digital infrastructure 1 2 Treatment Clinical Regulatory & Manufacturing support for Early R&D Commercialization Diagnosis development reimbursement & distribution physicians & patients 11 12 13 14 18 19 20 7 8 9 10 3 4 5 6 15 16 17 1. GALILEO: G
Slide 90: Using AI to increase efficiency of small molecule R&D
17 Using AI to increase efficiency of small molecule R&D Leveraging Design Hub and Chemical.AI as part of “Lab in a loop” Integration of AI based design tools in small molecule research Key benefits Holistic data overview MLOps Collaborative, ML prioritization Increased research efficiency More effi
Slide 91: Leveraging RWD to optimize patient eligibility criteria
18 Leveraging RWD to optimize patient eligibility criteria Significantly increased recruitment rate into Ph I oncology trial Background Solution ● Ph I trial in oncology experienced slow enrollment due ● Flatiron Health RWD was used to build virtual study cohorts with different eligibility to string
Slide 92: Utilizing digital biomarkers to support clinical development
19 Utilizing digital biomarkers to support clinical development Clinical trial use for biomarkers established in Parkinson’s and Huntington's disease Accelerating and improving clinical development Digital biomarker developed for PD and HD Active tests Passive monitoring Status quo: Costly, large an
Slide 93: Digital biomarkers measure disease progression and Tx effects
19 Digital biomarkers measure disease progression and Tx effects Efficiency gains as higher sensitivity enables smaller and faster clinical trials Ph II (PASADENA) results of prasinezumab in PD1 Ph III (GENERATION HD1) of tominersen in HD2 Speeded tapping test HD digital motor progression score (HDD
Slide 94: Applying a deep learning algorithm in ophthalmology
20 Applying a deep learning algorithm in ophthalmology HRF as a potential biomarker for disease severity and progression in DME Hyperreflective foci (HRF), a potential predictor Performing post-hoc analysis of Ph III (YOSEMITE/RHINE) for of poor vision outcomes in DME Vabysmo in DME using a deep-lea
Slide 95: Placeholder: video to be played full screen
Placeholder: During event a 30’’ video will be played full screen 95
Slide 96: Vabysmo treatment leads to fast and durable HRF reduction
20 Vabysmo treatment leads to fast and durable HRF reduction DL algorithm approach suitable for post-hoc analysis across therapeutic areas 3D SD-OCT analysis of representative Comparison of Vabysmo vs. aflibercept treatment effect Vabysmo patient1 on HRF reduction2 Baseline BCVA: 57 letters CST: 553
Slide 97: Doing now what patients need next
Doing now what patients need next