The EU Joint Clinical Assessment, mandatory for oncology and ATMP products from January 2025, expands evidence requirements significantly compared to most national HTA processes. JCA PICOs may require evidence on comparators not studied in randomised controlled trials, rare subpopulations underrepresented in trials, outcomes that were not measured in the trial programme, and long-term real-world follow-up that the controlled trial setting could not generate. Under the JCA timeline, all of this evidence must be ready before submission, not commissioned years later as a condition of conditional reimbursement. [1] This timeline makes clinical trial evidence gaps AI identification not a post-hoc documentation exercise but a pre-Phase III strategic requirement.
At Pienomial, we see evidence gap identification as the most consequential application of KnolAI in the pre-submission phase of a product's lifecycle. A HEOR team that understands precisely which clinical evidence gaps their trial programme will leave, and which of those gaps are addressable before submission versus which will require RWE supplementation, is a team that can build an evidence strategy with a realistic chance of satisfying NICE, G-BA, HAS, and JCA assessors simultaneously. This post explains how to identify clinical trial evidence gaps systematically using AI, what types of gaps matter most for different HTA bodies, and how Knolens enables this analysis at the speed and depth that pre-Phase III evidence planning requires. [9]
The most common form of evidence gap identification in pharma is retrospective: the HTA submission has been filed, the assessor has issued a formal clarification request identifying the gap, and the HEOR team is now in a four-to-six-week response window trying to address it with whatever data is available. This is the wrong time to discover an evidence gap. By the time a NICE technical team issues a clarification letter about an indirect treatment comparison methodology problem or a G-BA hearing reveals that the ZVT comparator was not adequately justified, the trial that could have addressed the gap has been completed and the data it could have generated no longer exists.
The correct time to identify clinical trial evidence gaps is Phase II, before Phase III protocol lock, when every evidence architecture decision, comparator choice, endpoint selection, subgroup pre-specification, and RWE data generation plan, can still be changed. Evidence gaps identified at Phase II become protocol design improvements. Evidence gaps identified at Phase III initiation become RWE study commissions. Evidence gaps identified at submission become clarification requests.
The challenge that prevents most HEOR teams from conducting systematic evidence gap analysis at Phase II is the manual effort required. Understanding which evidence gaps will matter to NICE requires analysing NICE appraisal history for the indication. Understanding which gaps will matter to G-BA requires analysing G-BA benefit assessment decisions and IQWiG methodology comments for analogous products. Understanding which gaps will matter to JCA requires analysing EUnetHTA methodology guidance and the specific PICO scope communicated for analogous products. This is months of manual analysis, which is why most teams do not complete it before Phase III. KnolAI compresses this to weeks.
Not all evidence gaps are equal in the eyes of HTA bodies. Understanding which gap types create the most significant reimbursement risk enables HEOR teams to prioritise their evidence generation strategies and address the highest-risk gaps first. [1]
Gap Type 1, Comparator misalignment: The trial was designed against a comparator that differs from the comparator the HTA body defines as the appropriate standard of care. This is the most consequential and least recoverable evidence gap. G-BA defines the ZVT independently, and IQWiG does not consider analyses against the wrong comparator as evidence of added benefit regardless of their statistical quality. NICE's comparator definition may differ from the FDA's. JCA may require comparisons against multiple comparators simultaneously. A comparator gap identified at submission is not addressable without a new trial or a methodologically defensible indirect treatment comparison, which may not be feasible.[3]
Gap Type 2, Endpoint patient-relevance gaps: The trial measured endpoints that the HTA body does not accept as sufficient evidence of patient benefit. G-BA accepts only patient-relevant endpoints: mortality, morbidity, quality of life, and adverse events. PFS as a surrogate endpoint requires validation as a surrogate for OS in the specific indication and patient population. NICE accepts PFS with survival extrapolation but scrutinises the extrapolation methodology. JCA assessors require documentation of the correlation between surrogate endpoints and patient-relevant outcomes before accepting surrogate-based evidence as the primary clinical effectiveness demonstration. [1]
Gap Type 3, Missing subgroup evidence: The trial did not pre-specify subgroup analyses that HTA assessors will require to evaluate benefit in specific patient populations. G-BA routinely requests subgroup analyses by age, disease stage, prior therapy, and biomarker status. JCA may require subgroup evidence for patient populations relevant to specific member states that differ from the overall trial population. Subgroup analyses not pre-specified in the protocol are classified as exploratory and carry minimal evidential weight. This gap is addressable only if the protocol is modified before the Phase III statistical analysis plan is finalised.
Gap Type 4, Long-term outcome data gaps: The trial follow-up is insufficient for the HTA body to assess long-term clinical benefit. NICE frequently challenges survival extrapolation models when trial median follow-up is short relative to the expected disease course. JCA requires documentation of real-world evidence to supplement trial evidence where long-term outcomes cannot be assessed within the trial timeframe. This gap is partially addressable through prospective RWE study commissioning at Phase III initiation, which can generate two to three years of real-world follow-up by the time the JCA submission is due.
KnolAI identifies clinical trial evidence gaps AI by querying the Knolens knowledge graph for two interconnected evidence sets: what evidence the target HTA bodies have required and accepted in prior assessments for analogous products, and what evidence the current trial programme will actually generate. The gap is the structured difference between what is required and what will be available.
The KnolAI evidence gap analysis workflow operates in three stages for each target HTA body.
Stage 1, HTA precedent analysis: KnolAI queries all prior assessments by the target HTA body for products in the same indication and mechanism class. For each assessed product, it extracts: the comparator the body defined, the endpoints it accepted as patient-relevant, the subgroup analyses it requested, the evidence it found insufficient and why, and the RWE it required or accepted. This analysis produces a structured profile of what the HTA body expects, grounded in its actual assessment behaviour rather than its published methodology guidance, which often understates the body's expectations in practice.
Stage 2, Current trial evidence inventory: KnolAI analyses the current trial programme: the Phase III protocol, the planned endpoints, the comparator, the pre-specified subgroup analyses, the trial population characteristics, and the planned follow-up duration. This produces a structured inventory of what evidence the trial will generate.
Stage 3, Gap identification and prioritisation: KnolAI compares the HTA body's evidence requirements with the trial's evidence inventory and identifies the specific gaps: which comparator requirements the trial does not address, which endpoint patient-relevance standards the trial does not satisfy, which subgroup analyses are not pre-specified, and which long-term outcome requirements the trial follow-up cannot meet. Each gap is categorised by addressability: gaps that can be addressed by protocol modification, gaps that require RWE study commissioning, and gaps that are not addressable and must be documented and managed in the dossier strategy.
The comparator gap is the evidence gap that most frequently determines the commercial outcome of a submission, and it is the gap that is most difficult to address after Phase III is underway. At Pienomial, we recommend that every product with EU market ambitions conduct a KnolAI comparator gap analysis before Phase III protocol lock, without exception.
The KnolAI comparator gap analysis queries the Knolens knowledge graph for all G-BA ZVT decisions, all NICE comparator definitions, all HAS reference therapy selections, and all JCA PICO scope communications for products in the same indication and mechanism class. The output is a structured comparator map showing, for each target HTA body, which comparator has been required, which has been accepted, and how often the comparator required by each body differs from the comparator used in the trial. This map enables the clinical development team to choose a trial comparator that either directly addresses the most likely HTA body comparator requirement or that enables a methodologically defensible indirect treatment comparison against all required alternatives.
The value of AI for this analysis is in breadth and recency. Manual analysis of the comparator landscape for a complex oncology indication, covering multiple HTA bodies and multiple analogous products, requires weeks of analyst time and may still miss recent assessments that have changed the comparator precedent. KnolAI completes the same analysis in hours, drawing from a knowledge base that is continuously updated with new assessment decisions as they are published.
The subgroup evidence gap is particularly consequential for G-BA and JCA assessments. G-BA routinely requests subgroup analyses by age, disease stage, prior therapy, and biomarker status for oncology products. IQWiG does not consider 73% of subgroup categories presented by health technology developers under the current G-BA template because they were not appropriately pre-specified. [3] This means that a large proportion of subgroup evidence submitted to G-BA is classified as exploratory rather than confirmatory, carrying minimal evidential weight in the benefit assessment.
The solution to the subgroup gap is not to add more subgroup analyses at the submission stage. It is to pre-specify the right subgroup analyses in the Phase III statistical analysis plan, before the trial begins, based on what G-BA and JCA assessors have historically requested for analogous products. [1]
KnolAI identifies the specific subgroup analyses that G-BA, NICE, and JCA assessors have requested in prior assessments for the indication and mechanism class, producing a subgroup requirement map that becomes a direct input to the statistical analysis plan. Every subgroup that appears in the map as a historical assessor request is pre-specified in the statistical analysis plan as a confirmatory analysis before the trial enrols its first patient. The evidence cost of this analysis is negligible. The evidence value is substantial: confirmatory versus exploratory subgroup evidence is the difference between evidence that can support a broad benefit rating and evidence that supports nothing. [9]
Some clinical trial evidence gaps cannot be addressed by trial design modifications because they reflect structural limitations of controlled trial methodology: the trial population is narrower than the clinical practice population by design, the follow-up is shorter than the disease course by necessity, and the comparator is defined by trial feasibility constraints rather than current clinical practice in every market. [9]
For these gaps, the evidence strategy is prospective RWE study commissioning. JCA explicitly requires RWE to address PICOs that span multiple member states, provide external control arms where the trial comparator does not reflect the real-world standard of care in specific countries, and extend follow-up to longer-term outcomes in broader patient populations. The critical constraint is timing: JCA RWE must be ready before submission, not commissioned after the conditional reimbursement decision. [1]
KnolAI's RWE gap analysis identifies which evidence requirements of each target HTA body cannot be met by the trial data, assesses which available RWE data sources can address each gap with existing data, and identifies which gaps require prospective study commissioning with the estimated time to evidence generation from each commissioning point. This analysis produces a prioritised RWE commissioning plan that maps evidence generation timelines to submission requirements, enabling the HEOR team to commission studies early enough to have data ready when the submission needs it.[9]
The evidence gap profile for an indication is not static. New competitor approvals shift the standard-of-care comparator. New HTA assessments change the endpoint patient-relevance standards. New methodology guidance from EUnetHTA, NICE, or IQWiG revises the evidentiary requirements. A gap analysis conducted at Phase II initiation may be outdated by Month 12 of Phase III if the landscape has shifted.
KnolAI monitors the evidence landscape continuously and alerts the HEOR team when a new event changes the gap profile: a competitor approval that shifts the standard-of-care comparator, a new NICE appraisal that changes the endpoint patient-relevance precedent for the indication, or a new JCA PICO scope communication for an analogous product that reveals the assessors' comparative evidence expectations. These alerts are not scheduled monthly reports. They are event-triggered notifications delivered within hours of the relevant event appearing in monitored source databases.
The HEOR team responds to gap profile changes before they become irreversible constraints. A comparator shift identified at Month 12 of Phase III can still be addressed by adding a comparator arm or by commissioning an RWE comparative study. The same shift identified at Month 30 cannot.
The output of a KnolAI evidence gap analysis is a structured evidence gap report that serves as the evidence strategy brief for the clinical development and HEOR teams. A complete evidence gap report for a product approaching Phase III contains six sections.
Section 1, Comparator gap map: The comparators required by each target HTA body versus the trial's planned comparator, with the source precedent for each body's requirement and an assessment of ITC feasibility for each comparator gap.
Section 2, Endpoint patient-relevance assessment: The endpoint patient-relevance standards applied by each body for the indication, the trial's planned primary and secondary endpoints, and the documentation required to satisfy each body's standards for each endpoint.
Section 3, Subgroup requirement map: The specific subgroup analyses requested by each body in prior assessments for the indication, mapped against the trial's current statistical analysis plan, with a prioritised list of pre-specification additions recommended before Phase III SAP finalisation.
Section 4, Long-term outcome gap assessment: The follow-up duration required by each body versus the trial's planned follow-up, with an assessment of survival extrapolation acceptability and the RWE evidence that would need to supplement trial data at submission.
Section 5, RWE commissioning plan: The prospective studies required to address the gaps that trial data cannot cover, with data source assessments, feasibility timelines, and protocol specifications for each required study.
Section 6, Gap monitoring plan: The source types and event triggers that KnolAI will monitor continuously to alert the team when new events change the gap profile, with defined response protocols for each alert type.
A comprehensive clinical trial evidence gaps AI analysis covering three target HTA bodies for a single indication typically takes a manual HEOR team eight to twelve weeks, accounting for literature searches across multiple databases, systematic review of prior assessment documents, comparative analysis against the trial protocol, and documentation of findings. KnolAI completes the same analysis in one to two weeks, drawing from the Knolens knowledge graph's pre-loaded precedent data for NICE, G-BA, HAS, and JCA.
Sprint 1, Week 1, HTA precedent analysis complete: KnolAI queries the Knolens knowledge graph for all prior assessments by each target HTA body for the indication and mechanism class. The comparator map, endpoint patient-relevance history, subgroup requirement history, and long-term data requirements are extracted and structured. The precedent profile for each HTA body is complete by end of week one.
Sprint 2, Week 2, Gap analysis and evidence gap report delivered: KnolAI compares the HTA precedent profiles against the current trial protocol and statistical analysis plan. The structured evidence gap report is generated covering all six sections. Each gap is categorised by type, severity, and addressability. The subgroup pre-specification recommendations are delivered to the clinical statistics team. The RWE commissioning plan is delivered to the HEOR evidence generation lead. The gap monitoring configuration is activated.
From this point, the evidence gap profile is maintained continuously by KnolAI. New assessment decisions, new JCA PICO scopes, and new methodology guidance updates are monitored and integrated into the gap profile automatically. When a new event changes the gap priority, the HEOR team receives a structured alert with the specific implication for their evidence strategy and the recommended response action. The evidence gap analysis is a one-time exercise that transitions immediately into a continuous intelligence function.
Evidence gap identification is not a documentation exercise. It is the most important strategic intelligence function that a HEOR team can perform before Phase III protocol lock. Every gap identified and addressed at Phase II is an evidence requirement satisfied at submission without a clarification request. Every gap missed at Phase II becomes a submission vulnerability, a conditional reimbursement requirement, or an HTA outcome that falls short of the commercial objective the trial was designed to achieve.
At Pienomial, we built KnolAI's clinical trial evidence gaps AI capability because we believe that the single most valuable application of AI in pharma market access is not generating dossier text faster. It is ensuring that the dossier is built on an evidence architecture that actually satisfies the assessors who will review it. KnolAI makes systematic evidence gap identification achievable before every Phase III, not just for the programmes with enough resources to commission a six-month manual analysis.
CTA: Run a KnolAI evidence gap analysis for your next Phase III programme. Book a demo with the Pienomial team.
