Technology in Clinical Trials #13 – Advanced Prompting Guide for Medical & Regulatory Writers

In the rapidly evolving digital landscape, artificial intelligence (AI) has become an indispensable tool for medical and regulatory writers. As we've explored in a previous issue of the newsletter on basic prompting, the art of crafting effective prompts is crucial for leveraging AI in regulatory content creation. In this issue of the newsletter, we're taking a deep dive into advanced prompting techniques specifically tailored for regulatory submission documents.

More specifically, we'll explore sophisticated prompting strategies for key regulatory submission documents, including clinical study protocols (CSPs), clinical study reports (CSRs), and informed consent forms (ICFs). By mastering these techniques, medical and regulatory writers can significantly enhance efficiency and the quality of produced content, while navigating the unique challenges of each document type.

📈 The Evolution of AI in Medical and Regulatory Writing

The landscape of medical and regulatory writing is undergoing a rapid transformation, driven by advancements in AI and natural language processing (NLP). This evolution has opened new opportunities for more sophisticated prompting techniques in regulatory content creation (Foote HP et al, 2025).

AI-assisted writing tools are becoming increasingly adept at understanding context, maintaining consistency, and even suggesting improvements in structure and content. However, the key to unlocking their full potential lies in our (human) ability to guide these tools effectively through well-crafted prompts. As we delve deeper into advanced prompting techniques, remember that our role is not being replaced but rather augmented – our expertise in regulatory requirements, scientific understanding, and ethical considerations remains paramount.

📑 Understanding the Nuances of Regulatory Documents

Before we explore specific prompting techniques, it's crucial to understand the unique characteristics and requirements of different regulatory submission documents. Each document type serves a distinct purpose and audience, necessitating tailored approaches in our prompting strategies.

• Clinical study protocols (CSPs) outline the objectives, design, methodology, statistical considerations, and organization of a clinical study. They serve as the blueprint for the entire study and must adhere to Good Clinical Practice (GCP) guidelines, as detailed in the recently finalized ICH E6(R3).

• Clinical Study Reports (CSRs) provide comprehensive details about the design, execution, and results of clinical studies. They form the backbone of regulatory submissions and must present study data accurately and objectively.

• Informed Consent Forms (ICFs) are critical for ethical research conduct, requiring clear, understandable language to ensure study participants are fully informed about the study's nature, risks, and benefits.

To optimize our use of AI in creating content for the abovementioned documents, we need to craft prompts that address their specific requirements. For instance, consider this prompt for analyzing the latest GCP guidelines:

"Analyze the ICH E6(R3) guidelines and generate a concise summary of the key changes from the previous version, focusing on aspects that directly impact the writing of clinical study protocols. Highlight any new requirements for data integrity, risk-based quality management, and patient-centric approaches."

This prompt demonstrates how we can direct AI to focus on relevant aspects of regulatory guidelines, helping us quickly identify and incorporate critical updates into our writing process.

💉 Advanced Prompting for Clinical Study Protocols

When it comes to drafting CSPs, the challenge lies in ensuring comprehensiveness, consistency, and compliance with regulatory requirements. Advanced prompting techniques can significantly streamline this process. Let's explore some strategies:

1. Layered Prompting:

Layered prompting involves starting with a broad prompt and progressively refining it to address specific aspects of the protocol. For example:

1. Initial prompt: "Draft an outline for a Phase III clinical study protocol for a novel diabetes medication."

2. Refined prompt: "Based on the initial outline, expand the eligibility criteria section, ensuring inclusion of specific HbA1c ranges, age restrictions, and relevant comorbidities. Reference the SPIRIT-AI guidelines for AI-specific considerations if applicable."

3. Final layer: "Review the expanded eligibility criteria. Ensure alignment with the study objectives and endpoints. Suggest any additional criteria that might be necessary based on the mechanism of action of our novel diabetes medication."

This layered approach allows for a systematic development of the protocol, ensuring that each section is thoroughly addressed and aligned with the overall study design.

2. Comparative Prompting:

The comparative prompting technique leverages the AI's ability to analyze and compare information. For instance:

"Compare the key elements of our protocol's statistical analysis plan with those of three recently published diabetes studies in the New England Journal of Medicine. Highlight any innovative approaches in our plan and identify areas where we might need to provide additional justification for our chosen methods."

This prompt encourages a critical evaluation of the CSP in the context of current best practices, potentially identifying areas for improvement or modification.

3. Error-Proofing Prompts:

Consistency is crucial in protocol development. Use error-proofing prompts that ask the AI to check for internal consistency:

"Review the drafted protocol sections for internal consistency, particularly between the study objectives, endpoints, and statistical analysis plan. Flag any potential discrepancies, such as secondary endpoints not aligned with stated objectives or statistical analyses not adequately powered to address key research questions."

Such prompts serve as an additional quality control measure, helping to identify potential issues before the protocol undergoes formal review.

4. Regulatory Compliance Prompts:

Ensure your CSP aligns with the latest regulatory guidance:

"Analyze our draft protocol against the ICH E6(R3) guidelines. Identify any areas where our protocol may not fully address the updated GCP requirements, particularly regarding risk-based approaches to clinical trial design and conduct. Suggest modifications to ensure full compliance."

This type of prompt helps ensure that the protocol remains up-to-date with evolving regulatory expectations.

By employing these advanced prompting techniques, we can enhance the efficiency and quality of protocol development. Remember, while AI can significantly assist in drafting and reviewing, the final CSP should always undergo thorough human expert review to ensure scientific validity, ethical considerations, and regulatory compliance.

📝 Mastering Prompts for Clinical Study Reports

Clinical Study Reports are critical regulatory documents that present comprehensive details of clinical study design, execution, and results. Crafting effective prompts for CSR development can significantly enhance the quality and efficiency of the writing process. Let's explore some advanced prompting techniques:

1. Data Summary Prompts:

Clinical Study Reports require the synthesis of complex data into a paragraph format. Consider this example:

"Synthesize the efficacy data from Tables 1-5 into a 2-paragraph summary, highlighting the primary endpoint results and key secondary findings. Ensure alignment with the statistical analysis plan outlined in Section 9 of the protocol. Identify any trends or subgroup analyses that warrant special attention."

This prompt guides the AI to focus on the most critical data points while maintaining consistency with the pre-specified analysis plan.

2. Safety Summary Prompts:

Safety data in CSRs requires careful analysis and presentation. Try this approach:

"Create a comprehensive safety summary that includes an analysis of adverse events, laboratory abnormalities, and vital sign changes. Identify any notable trends or unexpected findings, and suggest potential areas for further investigation. Pay particular attention to any serious adverse events and their potential relationship to the study drug."

This prompt encourages a thorough review of safety data, highlighting areas that may require additional explanation or analysis.

3. Discussion Section Prompts:

The discussion section of a CSR requires critical thinking and contextual understanding. Consider this prompt:

"Draft the discussion section of the CSR, interpreting the study results in the context of current treatment guidelines for [disease area]. Address how these findings might influence clinical practice, and identify any limitations of the study design. Compare our results with those of similar recent studies, and suggest potential areas for future research."

This prompt guides the AI to produce a well-rounded discussion that places the study results in a broader clinical and scientific context.

Remember, while these prompts can generate excellent starting points, the nuanced interpretation of clinical trial data and its implications still requires human expertise. Always review and refine AI-generated content to ensure accuracy, compliance with regulatory requirements, and appropriate scientific interpretation.

😷 Crafting Effective Prompts for Informed Consent Forms

Informed Consent Forms are crucial for ethical research conduct, requiring clear, understandable language to ensure participants are fully informed. Developing prompts for ICFs presents unique challenges due to the need for simplicity and cultural sensitivity. Here are some advanced prompting techniques:

1. Simplification Prompts:

Use prompts that encourage the AI to explain complex concepts in layman's terms:

"Explain the concept of 'randomization' in layman's terms, suitable for a patient with an 8th-grade reading level. Include an analogy to aid understanding, and ensure the explanation covers why randomization is important for the study."

This type of prompt helps in creating content that is accessible to a broad audience, which is crucial for ICFs.

2. Cultural Sensitivity Prompts:

ICFs often need to be appropriate for diverse populations. Consider this prompt:

"Revise the explanation of potential risks and benefits to be culturally appropriate for a diverse urban population in the United States. Ensure the language is inclusive and avoids cultural assumptions. Suggest alternatives for any idioms or phrases that might not translate well across cultures."

This prompt encourages the AI to consider cultural nuances in language and explanation.

3. Readability Optimization Prompts:

Maintaining appropriate reading levels is crucial for ICFs. Try this:

"Analyze the drafted ICF for readability. Identify any sections with a Flesch-Kincaid grade level above 8, and suggest simpler alternatives that maintain the essential information. Ensure that medical terms are clearly defined and that sentence structure remains simple and direct."

This prompt helps in creating an ICF that is accessible to participants with varying levels of health literacy.

As noted by Martinez JF et al (2022), improving the informed consent process is crucial for ethical research conduct. These prompting techniques can aid in creating ICFs that are more understandable and participant-friendly, but remember that final documents should always be reviewed by ethics experts and tested with potential participants to ensure clarity and comprehension.

🌟 Ethical Considerations and Best Practices

As we leverage AI in medical and regulatory writing, it's crucial to address ethical considerations and establish best practices. The use of AI in healthcare and clinical research raises important questions about transparency, bias, and the role of human oversight.

1. Transparency Prompts:

It's important to be clear about the use of AI in content preparation. Consider this prompt:

"Generate a statement for the methods section that clearly describes the use of AI assistance in the preparation of this document, including the specific tasks performed by AI and the extent of human review and validation. Ensure the statement is transparent without undermining the credibility of the document."

This kind of prompt helps in maintaining transparency about the role of AI in document creation.

2. Bias Check Prompts:

AI systems can inadvertently perpetuate biases present in their training data. Use prompts like this to check for potential bias:

"Review the drafted clinical study report for potential sources of bias in the presentation of results. Identify any areas where the language may unduly emphasize positive findings or downplay negative ones. Suggest revisions to ensure a balanced presentation of all study outcomes."

This prompt encourages a critical review of the content to ensure fair and balanced reporting.

3. Human Oversight Prompts:

While AI can greatly assist in document creation, human oversight remains crucial. Consider this prompt:

"Identify sections of the document that require careful human expert review, such as interpretation of statistical analyses, discussion of clinical implications, and benefit-risk assessments. Suggest specific questions or considerations for reviewers to address in these sections."

This type of prompt helps in highlighting areas where human expertise is particularly important.

Gerke S et al (2020) emphasize the need for a systemic approach to regulating AI in healthcare, including considerations of transparency and accountability. As we integrate AI into medical and regulatory writing processes, we must remain vigilant about these ethical considerations.

Best practices for using AI in medical and regulatory writing include:

1. Always having qualified professionals review and validate AI-generated content.

2. Maintaining clear documentation of how AI was used in document preparation.

3. Regularly updating prompts to reflect the latest regulatory guidance and ethical standards.

4. Conducting periodic audits of AI-assisted processes to ensure compliance and identify areas for improvement.

By adhering to these ethical considerations and best practices, we can harness the power of AI to enhance our medical and regulatory writing practices while maintaining the highest standards of scientific integrity and regulatory compliance.

❗ Disclosure statement ❗

Lastly, I would like to point out that I am NOT associated with and am NOT sponsored by any of the organizations or personas I refer to in this edition of the newsletter.