Part 2 - Accelerating the Medical, Legal and regulatory (MLR) review leveraging AI

Medical, legal, and regulatory review (MLR review) of promotional as well as non-promotional materials is one of the core responsibilities of medical and commercial teams within any life sciences organisation to ensure that all marketed, regulated medical product communication materials are authentic, fair, balanced, and compliant, as per the standards laid by the governing local regulatory bodies. This ensures complete transparency and safeguards against misleading or false promotion.

Part One covered initial screening and categorisation of assets using AI. In this second part, MLR review task automation use cases are explored, with faster reviews for compliant content dissemination.

Identification and validation of approved product claims (regulatory review task automation)

With optimised prompts for a specific therapy area in a certain geography, GenAI can identify and compare all product related claims in both the approved product label (used as reference document), as well as in an adapted collateral created for the same context of use. Instances where a full match is observed can be added to an autogenerated, dynamic, unique claims library (with unique claims identification number) removing any possibility of duplication.

Any mismatches found on comparison can be flagged and highlighted for a manual review with the option to either override, comment, edit and accept, or by a straightforward rejection with a non-approval comment (Off Label/Misleading/Inappropriate). If the claim is accepted, it can be automatically added to the dynamic unique claims’ library. Here again, based on the context of use and content similarity score (80-99.9%), any duplicate claims can be avoided.

All non-approved claims can be accumulated in a separate folder, along with the approved dynamic claims library so that the GenAI model continuously learns and readily recognises acceptable and unacceptable claims as per the context of use and improves the accuracy of automated claims checks in every evolution of model.

Claims substantiation (medical review task automation)

Regulations on promotional materials demand that all approved claims should always be accompanied by valid scientific references. Also, each reference should be appropriately cited, and a link to the full text should be included in the promotional copy. The reference should reflect the right context of use for each of corresponding claim statement in the promotional material.

For GenAI to perform claims substantiation accurately, it would need to perform following checks:

• Linking approved claim to appropriate reference/s with link to full text/s.
• Correctness of claim-reference linking: Claim statement should contextually match (90-100% accuracy) with the text in the referenced article.
• If the claim is not linked with the correct reference (non-contextual/outdated), GenAI should identify and flag these errors for subjective manual review. AI prompting can be used for a quick search through approved databases/reference libraries to list credible references (with a contextual match score of 90-100% accuracy for that claim). The reviewer can select and add this reference as a link or attachment to that claim with an audit trail.
• If a claim is made without any reference, then such claims should be flagged as ‘Unsubstantiated claims’ and there should be an option for either ‘To discard/Reject claim’, or ‘Add a valid reference’. The reviewer can trigger an automated reference search as mentioned above and can add a valid reference to that claim.
• Detection of prohibited Claims: GenAI may be able to detect and highlight use of any prohibited claims, such as those that include guarantees of product effectiveness, suggestions of no side effects, and implications of health risks without the product’s usage. Such claims can be directly discarded/rejected.

With the above use cases, the GenAI model continuously learns and recognises prohibited claims, unsubstantiated claims, and correct and incorrect references as per the approved context of use.

Fair balance check (regulatory review task automation)

Some of the areas where GenAI can offer relief by eliminating subjective bias for checking fair balance of benefit versus risk information representation are as below:

• Identification of risk omission in promotional materials: GenAI can screen through the safety section of an approved product label and identify, categorizes, and summarise all the warnings, precautions, contraindications, very severe, severe, and moderate risks associated with the product and check if such essential risk information is missed in the promotional material. It can compare the draft material with the approved label for presence of a link to important safety information (ISI). If missed, such materials should be flagged as ‘Lacking Fair balance - missing risk information.'
• Assessment of risk and benefit information in promotional material for equal weightage and prominence: GenAI can identify and compare the percentages of benefit/efficacy information representation in the promotional material with the representation of risk information and also check for placement (in proximity or far away) of both types of information in the document, perform font size comparisons, analyse, and suggest whether equal weightage and prominence is given for both the types of information for assessing fair balance representation of risk-benefit information.

GenAI can thus perform a preliminary check as to whether the risk benefit information in the promotional material is ‘fairly balanced’.

Superlatives check (legal review task automation)

GenAI may be able to identify and highlight use of any superlatives or prohibited terms in the promotional materials to check for ‘overpromotion’.

Localisation checks (legal review task automation)

AI can be used to screen the promotional material for the following checkpoints by comparing it with the product’s legal guidance or brand guidance (rule-based check):

• Use of appropriate disclaimers
• Use of appropriate footnotes
• Copyright statements
• Disclosures (company and/or HCP disclosures)
• Conflict of interest
• Logo branding & Trademark (based on brand book guidance)
• Highlight use of any defamatory statements for comparator brands or products

GenAI tools can streamline the MLR process by tracking the reuse of previously approved materials and claims supported with appropriate references, automatically reviewing materials for non-compliant and non-conforming phrases, drawing on previous learning experience of approved terminologies to instantly offer compliant rephrasing options, thus accelerating compliance checks and speeding up the approval cycles, allowing the reviewers to simultaneously focus on clinical accuracy and scientific merit.

GenAI has strong potential for accelerating the current MLR review process by introducing fragmented automations throughout the cycle and can offer speed to market, agility, and great relief for all reviewers involved (medical, legal, and regulatory, including marketing teams).