How to Create a Risk Assessment with the Help of AI

To start with AI-Enhanced Risk Assessment generation, follow the steps below using the Protocol Analysis Tool.

1. Navigate to the Protocol Analysis Tool within the relevant study.

   For detailed instructions on how to access the tool, refer to the article: How to Enable and Access the Protocol Analysis Tool.

2. Click Generate Risk Assessment under Step 3 (see below).

3. To Generate a Risk Assessment, upload your Study Protocol in .docx or native (machine-readable) .pdf format by:
   1. Clicking Choose file
   2. Selecting the desired protocol file
   3. Clicking Open.

4. Click Generate Risk Assessment.

5. Once the request is processed (usually within a few minutes), the requestor will receive and email with a link to the generated Risk Assessment.

The generated Risk Assessment contains a list of Risk Statements in the "if-due to-then" format. Each Risk Statement has an associated Risk Question and an allocated Risk Category from the TransCelerate Categories List.

You can edit the Risks by:

1. Selecting a Risk Category from the drop-down list
2. Editing the Risk Question
3. Editing the Risk Statement
4. Deleting the Risk entirely

When all edits have been made, insert the Risk Assessment into RACT by clicking Insert in RACT. (The newly identified risks, risk questions and risk categories will be added to the existing ones in your RACT).

To download:

1. The uploaded Protocol file, click Download [Protocol name].
2. The recognized protocol text, click Download Recognized Protocol text.
3. The Log file(s), click Download Log file.

To start a new request, click New Request. The Risk Assessment generation process will start from the beginning.

Unedited results of any previous request can be accessed using the link in the email.

Important Notes:

• The artificial intelligence (AI) model used for Risk Assessment is based on Microsoft Azure OpenAI's GPT technology.
• It is designed to generate human-like responses to text prompts based on recognized patterns and correlations found in large datasets.
• It is not a subject matter expert (SME) and should not replace human professional judgment.
• All AI-generated results should be validated by qualified professionals and interpreted with caution.
• No study data is used to train the AI model. Your uploaded content remains confidential and secure.