SCOPE Site Selection Presentationsss.pptx

2025 SCOPE Site Selection Strategy: A Comprehensive Overview
Introduction
Effective site selection is critical for the success of clinical trials. The 2025 SCOPE Site Selection presentation outlines a data-driven approach to identifying, evaluating, and optimizing clinical trial sites to improve patient recruitment, retention, protocol adherence, and overall trial efficiency. This strategy integrates real-world data (RWD), artificial intelligence (AI)/machine learning (ML) models, and diversity considerations to enhance trial outcomes.

The 3 Ps of Site Selection
The presentation highlights three essential criteria for site selection:

1. Patients
Lower screen failure rates by choosing sites with access to high-quality patients.
Increase patient retention and reduce loss to follow-up (LTFU).
Ensure lower protocol deviations and fewer data queries.
2. Protocol
Sites should demonstrate strong protocol adherence.
Mandate rigorous training and use on-demand training modules for reinforcement.
Utilize AI-driven virtual training to maintain consistency across sites.
3. Performance
Optimize site collaboration and proactive issue resolution.
Invest in site preparation and training to improve compliance.
Implement real-time monitoring tools and AI alerts for protocol deviations.
Optimizing Site Selection with Data-Driven Strategies
The presentation emphasizes leveraging data science to improve site selection outcomes.

1. Selecting High-Performing Sites
Prioritize sites with a strong track record in patient recruitment.
Engage physician networks to drive patient referrals.
Utilize pre-screening tools, EHRs, and AI-based patient matching to enhance recruitment efficiency.
2. Community-Centric & Decentralized Approaches
Conduct culturally inclusive outreach and provide language-specific recruitment materials.
Partner with advocacy groups and local clinics to build trust and engagement.
Implement hybrid and decentralized trial models, including remote screening and telemedicine options.
3. Enhancing Patient Retention
Reduce visit and procedure burden by optimizing protocol design.
Improve patient engagement with mobile apps, SMS/email reminders, and real-time updates.
Offer incentives, reimbursements, and concierge services to address logistical barriers.
Diversity, Equity, and Inclusion (DEI) in Clinical Trials
Health Equity & Diversity Considerations
Conduct epidemiological research to understand disease prevalence, incidence, and disparities.
Benchmark historical recruitment patterns and assess FDA diversity guidelines.
Develop diversity plans that align with site selection and protocol strategies.
Data-Driven DEI Initiatives
Utilize Social Determinants of Health (SDoH) data to refine enrollment targets.
Implement AI/ML models to predict diverse enrollment and assess site suitability.
Engage underrepresented populations by tailoring recruitment strategies to their needs.

Site Performance & Predictive Modeling
The presentation introduces Clinical Operatio

1. Site Selection Overview Asma Kasuba Global Development, R&D Data Science and Digital Health Johnson and Johnson Innovative Medicine 2025

2. Choosing the Right S ites - 3 Ps Lower number of screen f ailures Higher Patient Retention/ Lower LTFU PATIENTS Good Quality Patients Lower protocol deviations Lower number of data queries PROTOCOL Higher Protocol Adherence PERFORMANCE Invest in Preparation and Training 2 Quantitative Strong site & sponsor collaboration Proactive issue resolution Qualitative

3. Patient Recruitment Strategies Optimize Site Selection – Choose high-performing sites with a strong track record of recruitment and access to the target patient population Leverage Physician Networks – Engage referring physicians early to identify and direct eligible patients to the trial Use Pre-Screening & Real-World Data – Utilize electronic health records (EHRs) and AI-based patient-matching tools to find eligible participants efficiently Culturally Inclusive & Language-Specific Outreach – Tailor materials to different demographics and address concerns specific to diverse patient populations Community-Based Recruitment – Partner with advocacy groups, local clinics, and support networks to build trust and improve patient engagement Hybrid & Decentralized Trial Models – Offer remote screening, telemedicine options, and home health services to reduce patient travel burden 3

4. Patient Retention Strategies Minimize Visit & Procedure Burden – Reduce unnecessary tests, long visits, and frequent hospital trips by optimizing the protocol design Enhance Patient Engagement & Communication – Provide mobile apps, SMS/email reminders, and regular updates on trial progress to keep patients informed and motivated Dedicated Patient Support & Concierge Services – Assign trial coordinators to assist with scheduling, transportation, and real-time queries Incentives & Reimbursement – Offer financial assistance (travel, childcare, lost wages) to alleviate economic barriers Address Common Dropout Reasons Proactively – Identify and mitigate issues like side effects, logistical difficulties, and unclear expectations through patient feedback loops 4

5. Diversity, Equity, and Inclusion 5 Approach & Impact An Oncology indication trial have a target for African American patient enrollment of 20% in the United States Opportunity based on new data partnerships to construct a diversity index that is directly linked to recruitment impact, beyond patient populations diversity mix Approach Collecting the RWD with R&E, demographics (Census), SDoH (CDC) and enrollment data sources Leverage ML to evaluate potential of site in enrolling diverse patients for screening, based on historical enrollment patterns Created a site level diversity index for an indication across US Sites which represents the likelihood of sites to contribute to diverse enrollment High model precision in predicting diverse enrollment based on historical study-site data Enhancing the robustness of diverse sites recommendations by incorporating key factors contributing to diverse enrollment Embedded diversity considerations in a more targeted and systematic approach; complimenting Clin Ops and RWD f actors contributing to strong operational performance and diverse enrollment are not necessarily the same across sites Impact

6. Utilizing SDOH and Diversity Data to Meet Enrollment Targets 6 Inherited Demographic Traits Economic Stability & Education Neighbourhood & Physical Environment Consumer Habits Community Context & Social Support Housing & Transportation Behavioral aspects Age Gender Race Ethnicity Income level Accumulated wealth Wealth stability Credit stability Employment status Employment trajectory Highest education attainment Neighboorhood crime index Neighboorhood poverty index Food dessert status Median area income Percentage english speaking population Preferred grocery store type Restaurant habits Level of Discretionary spending Level of Alcohol consumption Marital status No of household members Presence of a caregiver Number of children in household Number of nearby relatives Homeownership status (own or rent) Dwelling type Rental affordability Address stability Number of registered vehicles Average area commute time Smartphone usage Credit card usage Level of excersise Interest in health product & ervices Interest in outdoor activities

7. Health Equity & Diversity Landscape 7 Review disease landscape and historical trials, from a health disparity and diversity perspective Conduct desk research, RCT / RWD data analyses, and/or RWE studies based upon study requirements Understand the disease, epidemiology, & historical trials of an indication, from a health equity and diversity perspective, including: Prevalence, incidence, mortality & risk factor assessment in RWD Benchmarking of historical recruitment in clinical trials Review of race & ethnicity FDA comments in post-marketing requirements and commitments Shape program & study strategy, across diversity plan, protocol & site selection strategy Developed across 3 programs Developed template deck for assessments to streamline deliverable generation Defined DEI landscape for the indications of interest to help clinical teams develop diversity plans (OR THE Grant awarded to develop engagement/site capacity strategy and to improve diverse recruitment of the disease ) Confirmed lower incidence and prevalence rates of an indication in US minority groups, and found high ascertainment bias for Hispanic and Asian groups (Submitted for publication to PLOS One) Overview

8. Protocol Adherence Site Training & Onboarding Mandate protocol training and use on-demand modules for reinforcement Provide AI-driven virtual training for consistency across sites Real-Time Monitoring & AI Alerts Implement EDC systems with real-time alerts for deviations Use AI to detect compliance risks early and automate performance tracking Simplified Protocol Design Avoid complex procedures to reduce deviations and patient dropout Engage investigators early for feedback before finalizing protocols Enhanced Communication & Feedback Establish direct communication channels for issue resolution Use automated dashboards to track deviations and share best practices Standardized Data Collection & Quality Checks Automate data validation checks and ensure centralized monitoring Set clear KPIs and performance benchmarks for site adherence 8

9. Clinical Operations – Protocol Analyzer 9 Approach, Impact and Overview Approach Impact Impact analysis has shown that the top 50% ranked sites through Clinical Operations Modeling enroll 14-150% more patients than the bottom 50% Modeling approach is applied to 30+ studies for 10+ indications Huge need to estimate performance of site in clinical trials Opportunity to utilize clinical operations data sources to estimate enrollment potential, likelihood of non-enrollment or default rate, and site start-up time or agility Define trial benchmark studies that are most similar to target study Master Drugdev , CTMS site and investigator data and leverage AI/ML to predict enrollment potential, non-enrollment (default) likelihood, and start-up time of a site Create a site level enroll, default (non-enrolling) and agility score across sites which respectively represent the enrollment potential, non-enrollment likelihood, and start-up time of a site Overview DP 1 DP 5 Time sensitive feature extraction Start-up time model Non-enrollment model DP 3 DP 2 Enro l lment model Enroll score Start up time Generative AI Feature extraction DP 4 Aggregator Default score

10. Site List Selection Site Housing Hub : Establish a centralized, consistent, common master data foundation, serving as the source of truth across Trials360.ai applications & external partners R epository for mastered site data across sources and foundation of Trials360.ai and site list development process Site Metrics calculation : Allows final site recommendation to include site prioritization based on structured weighting of performance, experience, and custom metrics Automates and standardizes manual data extraction, curation and weighting so recommendations generated in structured and traceable weighting system Cumulative Score calculation: Integrates  machine learning pipelines into a single, tiered recommendation for site performance: Real-World Data (RWD): Tailored patient insights, patient site interactions, referrals, line of therapy, publications, census, etc . Clinical Operations data: Historical site experience, startup times, & performance, I/E criteria , contracting data, etc. Diversity, Equity, and Inclusion Index (DEI): Includes race, ethnicity, sex, age, social determinants of health in combination 10

11. 11 Benchmarking/ Competitive Studies Assess site performance & experience drivers desired by study teams Integrated Qualitative Insights Site Prediction AI/ML Model Identify sites with high enrollment potential and low likelihood of default to drive modeling and algorithms Real World Data Assess patient volumes & enrollment potential at HCO / HCP level to identify sites fitting the cohort profile Diversity, Equity, and Inclusion Data Geographic patient & investigator diversity Key Elements of Site List Development Site Housing Hub Historical feasibility & engagement data on potential study interests incorporating feedback from stakeholders and PIs Sources Collaborate with Clin Ops and Scientists to identify and HCPs with specialized capabilities that will impact study success Internal Stakeholders Overview

12. 12 Real-World Data Trials face challenges in an active and competitive research field to recommend new sites Opportunity to utilize Real-World Data (RWD) sources to suggest an untapped pool of sites with trial naïve and experienced sites Approach & Impact Approach Define patient cohort for upcoming study Extract RWD data and compute referral networks Unique patient cohorts & patient-site interactions for each indication Link RWD, Drugdev and CTMS data at site level and leverage AI/ML to predict enrollment potential of a site, based on RWD data and historical enrollment patterns Create a site level enroll score across US Sites which represents the enrollment potential of a site By utilizing RWD, we can inform site selection using patient cohort specific factors such as patient volumes, referrals and line of therapy RWD Coverage in certain is very high which allows to understand the entire landscape New sites were identified as promising sites, thus expanding the pool of sites for our trials Impact

13. 13 Prediction of S ite Level Enrollment Opportunity to more accurately forecast the completion date of the trial and to counteract delays early on Prediction of Active Sites (PAS) Approach Extract site level Live enrollment data and trial level data ( CiteLine , Clinical Operations Predictions, Contract data) Leverage ML to predict the potential of a site to enroll more patients / the risk of defaulting, based on site level screening performance data, COP predictions and enrollment payments. Group sites into low/medium/high potential sites based on ML predictions. Identify recruitment slow down on a site level early on. Develop mitigation strategies by increasing engagement with sites at risk for low enrolment, closing and replacing sites without enrolment and allocating resources to sites with highest potential. Reduce delays in study completion by continuously assessing site enrolment potential and correct for enrolment deficits early on Impact

14. Ad-Hoc Analytics 14 Advanced Analytics for Enhancing Clinical Trial Efficiency & Effectiveness Line of Therapy Analysis Leverage RWD to determine lines of therapy used by target patient population Patient Referral Analysis Understand HCP referral patterns and leverage PI referral networks to identify target patient hotspots Site Overlap Analysis Understand site overlap between competitive studies to inform site tiering and outreach strategies Site Outreach Analysis Guide prioritization for site outreach based upon study considerations and site enrollment potential Non-Enrolling Site Analysis Assess and mitigate contributing factors to non-enrolling sites to maximize site productivity Predicting Active Sites (PAS) Analysis Predict screening and enrollment for active sites to optimize clinical supply using site profile and performance data Long Term Extension (LTE) Analysis Assess benefits ( eg , approval, efficacy, recruitment) of LTE to study/program Investment Impact Analysis Quantify risks ( eg , cost, timeline, enrollment) of delayed investment EXAMPLES

15. Key Takeaways 15 Patient Recruitment & Retention Protocol Adherence Site List Selection Optimize Sites & Networks – Select high-performing sites and engage referring physicians Data-Driven & Inclusive Outreach – Use AI, EHRs, and culturally tailored strategies Decentralized & Flexible Models – Leverage telemedicine, remote screening, and community partnerships Minimize Burden & Enhance Engagement – Reduce visits, provide mobile updates, and offer support services Proactive Retention Strategies – Address dropout risks with financial aid, patient coordinators, and feedback loops Standardized Training – Mandate protocol training with AI-driven modules AI & Real-Time Monitoring – Use EDC systems and AI alerts for early risk detection Simplified Protocols – Minimize complexity and involve investigators early Effective Communication – Enable direct channels and automated dashboards Data Quality & Compliance – Automate validation, centralize monitoring, and set clear KPIs Centralized Site Data Hub – Unified master data source for Trials360ai & external partners Automated Site Metrics – Standardized weighting for performance, experience & custom metrics AI-Driven Site Scoring – Machine learning integrates RWD, clinical data & DEI factors Transparent Site Review – Centralized tool for reviewing, tracking, and refining site selection Integrated Applications – Supports site review, PI profiles, country insights & post-selection surveys

16. Q&A

17. Thank you!

18. Site Segmentation: M odel 18 Key Base Core Representing top-ranked global sites that satisfy the major selection criteria High volume ​, Scientifically high influence ​ Representing global ranked sites that satisfy the major selection criteria Shoulder a reasonable book of work , Potential to be Key ​ with support to raise performance ​ Representing about 60% of the global ranked sites Definitions Segmentation Criteria Site Performance and Quality Cycle time ( Major* ) Patient enrollment and Retention ( Major* ) Quality ( Major* ) Site Influence and Partnership Science Influence / Lead PI/National PI/KOL ( Major* ) Partnership (Past and current studies) Site Infrastructure Trial infrastructure Digital infrastructure Remote monitor capabilities ( Major* ) Net promoter score Network Affiliations Highly engaged site coordinators / Current responsiveness of site/PI Motivation ( Major* ) Clinical treatment focus and Patient Access Diversity Referral center ( Moderate** ) Internal Patient Database( Moderate** ) Biomarker capabilities Representing global ranked sites Sites for performance optimization Potential to move to Core *Major: That metric in the specified bucket is a maximum importance **Moderate: Metric is of moderate importance in that bucket

19. AI/ML Models & Approach Deep Dive

20. 20 Site List Development Algorithms Beyond Regular Benchmarking Metrics iAWARE Input Output Impact Clinical Operations Modeling Real-World Modeling Predicted site potential based upon previous research experience Leverage nuanced metrics and indication-specific criteria to drive performance-based site recommendations Insights on what drives site performance based upon real-world considerations (patient availability & scientific needs) Site enrollment predictions for both research experienced & research naïve sites Benchmark studies for each indication & target study Unique patient cohorts & patient-site interactions for each indication iAWARE Predicted site enrollment potential in areas that are likely to enroll patients from underrepresented populations Site considerations for trial diversity & patient representation Diversity Modeling Diversity data for patient cohorts and socio-demographic descriptors

21. Utilizing SDOH and Diversity Data to Meet Enrollment Targets 21

22. Utilizing SDOH and Diversity Data to Meet Enrollment Targets Inherited Demographic Traits Age Gender Race Ethnicity Economic Stability & Education Income level Accumilated wealth Income stability Credit s tability Employment status Employment trajectory Highest education Neighbourhood & Physical Environment Neighboorhood poverty & crime index Food dessert status Median area income % English speaking population Consumer Habits Preferred grocery store Restaurant habits Discretionary spending Alcohol consumption level Community Context & Social Support Marital status No of household members Presence of a caregiver Housing & Transportation Homeownership status (own or rent) Dwelling type Rental affordability Duration at current address Average commute time & no of vehicles in the area Behavioural Aspects Smartphone usage Credit card usage Level of excersise Interest in health product & ervices Interest in outdoor activities

23. 23

24. 24 A common master data foundation, serving as the source of truth across Trials360ai applications & external partners R epository for mastered site data across sources and foundation of Trials360ai and DDSL development process Site Feasibility & Insights Hub Site Feasibility & Insights Hub

25. 25 Key Features: Allows final site recommendation to include site prioritization based on structured weighting of performance, experience, and custom metrics Custom qualitative or quantitative features can help address data shortages or include unique features not databased Custom features are gathered through discussions with stakeholders to define parameters and weights Trials in indications with limited historical data can get structured recommendations Automates and standardizes manual data extraction, curation and weighting so recommendations generated in structured and traceable weighting system Recommendations with metrics weighting can be monitored to determine impact Site Selection Optimizer (SSO)

26. 26 Applications Included: Site Review and Decision Tools Site/PI Profile Country Profile Post Site Selection Survey Key Features: Enhancements will provide users with the ability to view sites that have been identified a tier directly from the system improved transparency by displaying the features that contributed to the Tier score Centralized place to review the initial site list, provide feedback and store historical decisions for future for site selection Site Selection Review Platform

27. Data driven capabilities to advance trial diversity and health equity from study planning through execution 27 Assess screening failures of historical trials across underrepresented subgroups & optimize protocol to improve inclusivity of study Protocol Optimization Enrollment Planning Support the Diversity Plan for upcoming studies with Census, RWD & RCT based benchmarks for enrollment goals Program & Study Strategy Desk research, data analysis (RCT & RWD) and/or real-world evidence study to assess Health Equity & Diversity considerations in priority indications Country & Site Prioritization Identify and evaluate recommendations of sites based upon data-driven & AI/ML site selection modeling Portfolio Assessment Study Planning & Execution Diversity Plan Analytics Tool Health Equity & Diversity Landscape Clinical Studio Site Selection Strategy Protocol Finalization Site Selection Site Activation FPI Assess diversity plan goals ​ vs proposed site selection scenarios, study milestones and budgets Diversity Index Screening Failure Analysis Existing Capability New Innovation Diversity Plan Draft Site Engagement & Patient Recruitment Tailor engagement strategy with sites based on SDoH data SDOH Informed Site Strategy Patient Heatmap

28. Key takeaways Site Selection Tools: Site Feasibility & Analytics Hub - AI ML Models in Site Selection: 28

29. SDoH highlight non-medical factors that influence health outcomes They are the conditions in which people are born, grow, work, live and age, and the wider set of forces and systems shaping the conditions of daily life SDoH Informed Site Engagement Strategy Social Determinants of Health generate Insights… 29 Multi-lingual recruitment & patient communication Engage community organizations/extra-clinical channels Emphasize cost coverage in material Offer off-hour appointments & transportation reimbursement To tailor Site Strategy Social Behavioral Environmental Genetics Medical

30. Real-World D a ta Model 30 Features and Validation Features Validation Patient volumes – cohort specific Patient visits – cohort specific Referral networks – cohort specific at HCO/HCP level Line of Therapy patient distributions at HCO/HCP level Census Potential and Penetration (EMEA & APAC) Open Payments’ features analyzing experience Publications # Specialists Research experience Treatments, procedures, pharmacy prescriptions volumes at site level – cohort specific Prescriptions data Top Factors determining the enrollment potential of a US site in a key IMM trial Validation is performed retrospectively at time of modeling on latest trial in test data cut separated from training data Impact analysis is done versus actual trial data once sufficient data is available in the trial or when it is completed to assess the impact and accuracy of the model recommendations Pro’s & Cons Info of sites for each trial Highly customizable Granular patient journey timeline 

31. Diversity, Equity, and Inclusion Model 31 Approach, Impact, Breakdown Situation Approach Impact High model precision in predicting diverse enrollment based on historical study-site data Enhancing the robustness of diverse sites recommendations by incorporating key factors contributing to diverse enrollment Embedded diversity considerations in a more targeted and systematic approach; complimenting Clin Ops Model and RWD Model Factors contributing to strong operational performance and diverse enrollment are not necessarily the same across sites Opportunity based on new data partnerships to construct a diversity index that is directly linked to recruitment impact, beyond patient populations diversity mix Collecting the RWD with R&E, demographics (Census), SDoH (CDC) and enrollment data sources Leverage ML to evaluate potential of site in enrolling diverse patients for screening, based on historical enrollment patterns Created a site level diversity index for Multiple Myeloma across US Sites which represents the likelihood of sites to contribute to diverse enrollment

32. Diversity, Equity, and Inclusion Model 32 General data science approach for diversity site selection modeling Modeling Features Predict the enrollment rates using historical data on sites recruitment performance Use site characteristics such as:​ Area demography – Census​ Social indicators – SDOH​/AHRQ Patients RWE sources Exploit similar sites' historical data​ Similar Sites Site Characteristics Site History Metric: Accuracy, precision, AUC Strategy Model per therapeutic area but evaluated for a given indication and target group Goal: Predict if a site will enroll patients in a specified group of people ( age, gender, race & ethnicity ) Problem Approach: High vs low enrollment where: High – higher than the median Low – lower than the median Test Set: latest studies at all/a subset of all sites

33. Diversity, Equity, and Inclusion Model 33 Machine learning is used to combine the different data sources into a diversity score that is predictive of the diverse Enrollment potential Define cohorts for upcoming study and extract patient populations Align the sites & PIs across the different data sources (site matching) Analyze the relationship between the enrolled subject diversity and the racial and ethnic diversity Build a machine learning model that predicts likelihood of diverse enrollment Incorporate diversity findings in site selection outputs Extract diversity data at geography from Census & CDC Extract diversity data at cohort level from Komodo/ HealthVerity Extract  diversity enrollment data  of past Janssen clinical trials Pre-trial support site identification During trial assess appropriate actions Post-trial validate impact on recruitment

34. Diversity Plan Analytics Tool  34 Initialize new Diversity Plan Define indication and study level RWD cohorts and contrast to overall population 1 2 Investigate current trial recruitment status by comparing across study phase, sponsor & protocol 3 Align on DEICT goals with study team 4 Census and RWD Deep Dive Visualize race, gender, and ethnicity breakdowns at a territory or cohort level RWD Protocol Report RCT Deep Dive

35. Clinical Operations Model 35 Features and Overview Features Overview Historical enrollment performance Historical default performance Historical start-up times Experience in given indication Contracting data Publications Research experience Time-bound features to avoid leakage Features contain actuals and relative performance of the site with respect to peers in same trial Feature creation includes generative AI to extract valuable insights from unstructured data DP 1 DP 5 Time sensitive feature extraction Start-up time model Non-enrolment model DP 3 DP 2 Enrolment model Enroll score Default score Start up time Generative AI Feature extraction DP 4 Aggregator

36. Real-World D a ta Model Diversity, Equity, and Inclusion 36 Approach, Impact and Breakdown Situation Approach Impact By utilizing RWD , we can inform site selection using patient cohort specific factors such as patient volumes, referrals and line of therapy RWD Coverage in certain is very high which allows to understand the entire landscape Sponsor naïve sites can be identified as promising sites, thus expanding the pool of sites for our trials Trials face challenges in a competitive field, but RWD can help identify untapped trial-naïve and experienced sites Define  patient cohort for upcoming study Extract RWD data and compute referral networks Unique patient cohorts & patient-site interactions for each indication Link RWD, Drugdev and CTMS data at site level and leverage AI/ML to predict enrollment potential of a site, based on RWD data and historical enrollment patterns Create a site level enroll score across US Sites which represents the enrollment potential of a site Approach Impact High model precision in predicting diverse enrollment based on historical study-site data Enhancing the robustness of diverse sites recommendations by incorporating key factors contributing to diverse enrollment Embedded diversity considerations in a more targeted and systematic approach; complimenting Clin Ops Model and RWD Model Factors contributing to strong operational performance and diverse enrollment are not necessarily the same across sites Collecting the RWD with R&E, demographics (Census), SDoH (CDC) and enrollment data sources Leverage ML to evaluate potential of site in enrolling diverse patients for screening, based on historical enrollment patterns Created a site level diversity index for Multiple Myeloma across US Sites which represents the likelihood of sites to contribute to diverse enrollment Opportunity based on new data partnerships to construct a diversity index that is directly linked to recruitment impact, beyond patient populations diversity mix Situation Approach, Impact and Breakdown