PRISM changes what trial data is asked to do.It connects multi-source intelligence, governed conversational AI, and predictive readiness analytics.The result is an operating system that surfaces completion risk while there is still time to act.
Clinical trials have access to more patient data than ever. EMR/EHR systems, claims databases, genomics, digital biomarkers, real-world evidence platforms. Yet screen failure rates linger at 20-40%. Dropout rates haven't budged in two decades. The data exists. It's answering the wrong question.
The industry uses data to answer: "Is this patient eligible?" PRISM uses data to answer: "Is this patient ready? Where will friction emerge? What intervention will work? Will this patient complete?"
Eligibility is necessary but not sufficient. Readiness is the gap between them, and it requires data sources, analytical models, and intelligence infrastructure that no existing platform provides.
Better data does not improve outcomes. Better questions improve outcomes. PRISM asks the right questions.
PRISM's Data Layer integrates five domains to build multidimensional patient profiles across the 6-stage operating system. Each delivers signals no single source can provide.
What it tells PRISM: Protocol metadata is the demand side of the readiness equation. How many visits? How long? What procedures? How complex is the medication regimen? What amendments shift burden? PRISM maps patient capacity against protocol demand to find where fit breaks.
Why it matters for readiness: Readiness is relative. The same patient may be ready for a low-burden chronic disease study and unprepared for a high-burden oncology trial. Protocol metadata contextualizes every readiness call.
What it tells PRISM: Claims data reveals treatment history, visit patterns, medication adherence, comorbidity burden, and healthcare utilization. These predict whether existing care obligations will conflict with trial participation.
What it cannot tell on its own: Claims data shows what happened, not why. A patient with perfect adherence may still drop out from caregiver burnout or logistical barriers. Claims data is necessary but insufficient.
What it tells PRISM: EMR/EHR data provides clinical detail claims cannot: lab trajectories, provider notes on engagement, documented concerns, disease progression, and response to treatment changes. This anchors eligibility and feeds clinical readiness.
What it cannot tell on its own: EMR captures the clinical picture, not the behavioral or structural one. A patient with favorable labs may face transportation barriers, caregiver conflicts, or cognitive friction EMR records miss.
What it tells PRISM: SDOH data maps the structural environment around the patient. Transportation access determines if site visits are feasible. Economic stability determines if participation creates strain. Housing, food security, and safety determine if a patient can sustain a multi-month commitment. These are hard predictors, not soft metrics.
Why it matters for readiness: A patient can be clinically eligible and cognitively prepared but structurally unable to participate. SDOH data surfaces structural friction before dropout occurs, and each friction point maps to a specific support response, so barriers get resolved instead of just flagged.
What it tells PRISM: BDOH captures the behavioral dimension clinical and structural data miss. Health literacy, healthcare engagement patterns, decision-making under uncertainty, and motivation durability. This layer most directly predicts persistence when participation gets hard.
Why it matters for readiness: Behavioral readiness erodes fastest. A patient may intellectually understand the protocol but withdraw when fatigue, confusion, or competing priorities pile up. BDOH signals feed initial readiness assessment (Stage 2).
PRISM's Coordination Layer operates as agentic AI: an always-on agent that engages patients conversationally, validates readiness signals, resolves confusion, pre-screens, and coordinates site handoff. It doesn't follow scripts. It governs transitions.
"Agentic" is not decorative. It means the AI has authority to make governed decisions within defined parameters. It holds patients at a stage if readiness criteria aren't met. It escalates to humans when signals are ambiguous. It adapts based on patient response. It operates within the operating system's rules, not outside them.
Chatbots answer questions from a knowledge base. PRISM's AI evaluates whether questions reveal understanding gaps, misconceptions, or fear. The same question ("How long is the trial?") may mean curiosity in one patient and burden anxiety in another. PRISM distinguishes.
Chatbots follow scripts. PRISM's AI adapts based on friction profile, engagement history, and readiness signals. High health literacy gets different communication than patients needing comprehension scaffolding. Caregiver concerns trigger support-system pathways. The AI doesn't have one conversation. It has a governed, adaptive dialogue.
Chatbots hand off when they fail. PRISM's AI hands off when the patient is ready. Human escalation is a design feature for edge cases, not a fallback for AI limits. The AI governs the transition. Humans govern clinical decisions.
Most trial metrics tell you what happened after recovery is impossible. Monthly enrollment reports. Quarterly retention summaries. Post-hoc screen failure analysis. PRISM surfaces live execution signals enabling intervention while time exists. Three proprietary indices monitor the trial-health dimensions that determine whether a study completes on time, on budget, and with sufficient data quality.
What PXCI measures: Clarity, coherence, and stability of the patient experience during participation. PXCI tracks if patients understand what's happening, why, and what's next. It measures experience quality as a leading indicator of outcomes.
Why it matters: Degraded experience doesn't just create dissatisfaction. It creates confusion, site rework, protocol deviations, and data quality loss. Patients not understanding visit schedules, medication instructions, or obligations force sites to re-explain instead of execute. PXCI surfaces friction before it spreads.
What a declining PXCI signals: Rising site burden from confusion. Protocol deviation risk. Re-consent and re-education needs. Dropout risk from experience frustration, not clinical burden.
What IEQS measures: Depth and durability of patient engagement. Not if patients interact, but whether interactions reflect genuine understanding and intentional participation. IEQS distinguishes between engaged patients who understand and engaged patients who haven't hit the friction triggering withdrawal.
Why it matters: High engagement metrics mask fragile participation. A patient attending every visit but not understanding the medication regimen is "engaged" by traditional measures but at high protocol deviation and withdrawal risk. IEQS captures the quality dimension activity metrics miss.
What a declining IEQS signals: Screen failure risk pre-screening. Early withdrawal risk post-enrollment. Re-consent risk when amendments hit. Dropout driven by misaligned expectations, not genuine inability.
What CRI measures: Recruitment speed, efficiency, and predictability during execution. CRI tracks if enrollment trajectory is stable, accelerating, or deteriorating. It measures not just if enrollment happens but if it's durable.
Why it matters: Velocity without stability predicts rescue recruitment. A trial enrolling on pace but with falling quality (high screen failure, rising dropout) heads toward collapse. CRI surfaces stability enrollment counts can't reveal.
What a volatile CRI signals: Upstream readiness issues creating inconsistent enrollment quality. Site-level degradation. Rising timeline extension probability. Rescue recruitment necessary in 30-60 days.
PRISM was designed for regulated environments from the ground up. Patient data governance, audit trails, consent management, and access controls are structural, not optional.
PRISM integrates five data domains: claims and billing data, EMR/EHR clinical records, social determinants of health (SDOH), behavioral determinants of health (BDOH), and protocol metadata. Together these create multidimensional patient profiles that serve readiness assessment across all 6 stages.
Agentic AI means PRISM's conversational intelligence has governed authority to make decisions within defined parameters. It can hold patients at a stage, adapt its engagement approach, conduct structured pre-screening, and coordinate site handoff. It operates within the operating system's governance rules, with human escalation for clinical ambiguity.
No. Chatbots follow scripts and answer questions from a knowledge base. PRISM's AI evaluates readiness signals from conversational interactions, adapts its approach by patient profile, conducts structured pre-screening, and governs stage transitions. It is an intelligent agent with defined authority, not a Q&A interface.
PXCI (Patient Experience Composite Index) measures the clarity, coherence, and stability of the patient experience during active trial participation. A declining PXCI indicates growing confusion, rising site rework burden, and increasing risk of experience-driven dropout. It is one of three live execution indices in PRISM.
IEQS (Informed Engagement Quality Score) assesses whether patient engagement is informed, intentional, and durable. Low or declining IEQS predicts screen failure, early withdrawal, and re-consent risk. It distinguishes genuine understanding from surface-level participation that traditional engagement metrics cannot differentiate.
CRI (Composite Recruitment Index) tracks recruitment speed, efficiency, and predictability during active execution. Volatility in CRI signals upstream readiness issues, inconsistent enrollment quality, and operational fragility. A volatile CRI often predicts rescue recruitment needs 30-60 days before timelines slip.
PRISM was designed for regulated clinical environments. Features include role-based access controls, complete audit trails for all AI decisions, integrated patient consent management, data minimization to required readiness signals only, and configurable retention policies.
Enrollment dashboards report what happened. PRISM's indices diagnose what is happening and predict what will happen. They surface instability (experience degradation, engagement fragility, recruitment volatility) while recovery is still possible. They are diagnostic and corrective, not retrospective.
SPUR is PRISM's readiness diagnostic framework. It evaluates four dimensions traditional eligibility ignores: Social (caregiver support, household stability), Psychological (fear, confidence, trust in research), Usage (logistics, transportation, financial friction), and Resource (health literacy, cognitive load). SPUR produces readiness tiers, dropout risk indicators, and activation priority levels before advancement decisions are made.
Once SPUR identifies where readiness is fragile, COM-B provides the prescription logic. It evaluates three drivers of sustained behavior: Capability (does the patient truly understand?), Opportunity (does the environment support participation?), and Motivation (is willingness durable?). COM-B translates readiness gaps into specific interventions: redesigning education sequencing, activating caregiver involvement, adjusting logistical support, or modifying advancement timing.
Fogg governs execution at the moment of action. Clinical trial participation is a series of micro-decisions: booking a screening visit, completing consent, attending visit three, submitting an ePRO, staying enrolled after a protocol amendment. Fogg ensures three conditions are met simultaneously: motivation (willing right now), ability (task is simple enough right now), and prompt (triggered at the right moment). This prevents the most common automation failure: asking patients to act when they are not ready to succeed.
Together they answer three distinct operational questions. SPUR: Is this patient realistically ready for this trial? COM-B: What must change for participation to be sustainable? Fogg: What action will succeed right now, in this moment? Most clinical trial approaches apply behavioral concepts descriptively. PRISM applies them governance-first, meaning they directly influence who advances, when they advance, and what support is deployed.
Proprietary constructs defined by Jumo Health. Each one is a measurable operational concept, not a marketing term. View all definitions →
In two days, you get three execution diagnostics on your program.
How likely each eligible patient cluster is to activate, persist, and complete, before any intervention.
The barriers most likely to suppress activation, enrollment, and completion in each cluster, and how to mitigate them.
Which clusters to target, which barriers to address, which interventions to deploy, and what completion lift to expect.