Close Search

Predictive Enrollment Planning for Clinical Protocols

Overview

As organizations aim to enhance Clinical Protocols planning and execution, providing a streamlined approach to protocol management becomes essential. When Clinical Research Associates (CRAs) plan new protocols for a specific program, leveraging insights from an enrollment prediction model can greatly improve accuracy and efficiency. By integrating an interactive Oracle Digital Assistant, CRAs can engage in guided conversations to refine protocol details and make data-driven decisions. At the end of the interaction, the assistant automatically updates the protocol records, ensuring seamless integration and reducing manual effort. This innovative approach not only highlights the effective use of AI in clinical trial planning but also improves productivity and ensures accurate, up-to-date protocol documentation.

Introduction

In Siebel CTMS, creating and managing protocols can be time-consuming and prone to inefficiencies. When CRAs need to plan new protocols, manual efforts often lead to increased costs and delays. Introducing an automated system for enrollment prediction and AI-driven protocol planning can greatly enhance accuracy and efficiency. The AI insights provides the Enrollment prediction insights when provided with details of the Protocol record values.

For example, when a Clinical Research Associate (CRA) plans a new protocol for an existing program, the system uses an advanced ML Prediction model to provide data-driven insights to predict Subjects turnouts given the protocol record values like # of sites, Subjects, Disease, etc. These predictions are delivered via an interactive Oracle Digital Assistant, allowing CRAs to explore enrollment scenarios, optimize site selection, and refine protocol parameters. At the end of the interaction, the system automatically updates protocol records, ensuring a seamless workflow with minimal manual intervention.

The Enrollment Prediction Challenge in Clinical Trials

Clinical trial protocols often face significant hurdles due to high dropout rates and unpredictable recruitment delays. These inefficiencies lead to extended timelines and increased operational costs. A major challenge lies in the inaccuracy of enrollment predictions, as well as site variability, which directly impact planning, resource allocation, and overall trial efficiency.

To address these concerns, there is a need for:

  • Improved accuracy in enrollment predictions to minimize unforeseen delays.
  • Enhanced resource allocation through data-driven insights.
  • A streamlined and user-friendly mechanism for gathering protocol-related information.
  • AI-driven enrollment predictions enhance planning accuracy and optimize resource use.
  • AI-driven solutions that optimize planning and mitigate common protocol challenges.

A potential solution involves leveraging the Oracle Digital Assistant (ODA) to simplify protocol information collection while utilizing predictive models to provide accurate enrollment insights.

AI-Powered Solution: Streamlining Protocol Planning

The proposed solution integrates AI-driven enrollment predictions within the clinical trial management process, streamlining protocol planning and execution. The workflow is as follows:

  • The Clinical Research Associate (CRA) selects a relevant Protocol record within the Siebel Clinical Trial Management System (CTMS) user interface.
  • Upon selection, a notification is triggered from the embedded Oracle Digital Assistant (ODA) widget, prompting further action.
  • The CRA interacts with the ODA chatbot, providing necessary details regarding new enrollment requirements and expectations.
  • The ODA interfaces with the AI-driven enrollment prediction model, analyzing input data and generating decision insights.
  • Based on the insights provided, the CRA reviews and confirms the new enrollment details. The Protocol record is then updated automatically with the revised information, ensuring real-time accuracy and efficiency in protocol planning.

Implementation

    The implementation of this flow involves

  • Developing the ODA Skill Flow: Designing a structured conversation flow within Oracle Digital Assistant (ODA) to facilitate user interactions.
  • Building a Custom ODA Component: Implementing a custom component to integrate with Clinical Protocol REST APIs, process responses, and invoke the Enrollment Prediction Model.
  • Embedding the ODA Widget: Seamlessly integrating the ODA widget within the user interface for an intuitive and efficient experience.
  • Customizing Siebel CTMS Open UI: Enhancing Siebel CTMS with tailored Open UI modifications to support the new workflow.

Architecture Diagram

Architecture Diagram
  1. CRA opens the ODA Widget window after selecting a Protocol record (ODA Sends a Notification).
  2. ODA connects to the ODA Skill via the Channel ID and Skill ID configured.
  3. ODA Custom Component calls the Clinical Protocols REST API to get the details for current protocol program
  4. ODA Custom Component calls Enrollment Prediction Model to get the Prediction results
  5. After confirmation from the CRA the ODA Updates the Protocol record with new values.

Siebel customisations

Leveraging Siebel Open UI, a Physical Renderer File is used to Embed the ODA Widget in Siebel CTMS UI and make it available in all the Views for the CRA to interact. Following functionalities are achieved

  1. ODA Widget Embedding.
  2. Collecting User information (Name) and Current Selected Protocol Row ID to be passed to ODA.
  3. Some fields are given fading Highlighting effect when Protocol record is updated with new values.
  4. Entry of the ODA Web SDK file and PR file in the Manifest administration and Manifest Files.
  5. Styling changes in ODA Widget UI (Buttons, Color Branding, Date Picker).
Configuring PR File and ODA Web SDK

  1. Place the below ODA Web SDK and PR file inside SAI Container in location: sai-ENT:/siebel/mde/applicationcontainer/siebelwebroot/scripts/siebel
    • oda.js
    • web-sdk
    • OR Download latest from https://www.oracle.com/downloads/cloud/amce-downloads.html
  2. Add the file entries in Manifest Files and Manifest Administration of type: Application, Usage Type: Common, Name: PLATFORM INDEPENDENT.

ODA Configurations

ODA Skills and Custom component responsible for interacting with the User as well as the Prediction Model

ODA Skill

Import the below ODA Skill in an ODA instance and intake the necessary changes required which includes and not limited to

  1. Changing the ODA Skill Flow as per requirement.
  2. Updating the Custom component with new Clinical Protocol REST Endpoint, Prediction Model REST Endpoint.

Clinical Protocol REST API

Please enable Inbound REST Access to Clinical Protocol


https://siebelhost:port/siebel/v1.0/data/Clinical%20Protocol/Clinical%20Protocol/88-2WGW3

Step-by-Step Demonstration of AI-Powered Enrollment Prediction

After you’ve configured the customisations, follow below steps to see the functionality.

  1. Launch your Siebel application and log in with CRA credentials.
  2. From the main menu, find and click on the "Protocol List" option which is under the section "Administration - Clinical"
  3. Search for the protocol record
  4. Open the Protocol Record detail page
  5. Now click on the Embedded ODA Widget
  6. Interact with ODA and provide all necessary information.
    7. Usecase diagram
  7. You (CRA) will receive AI Insights and ODA Updates the Protocol Record with new values.

Benefits

  • Enhanced Protocol Accuracy:By using AI-driven enrollment predictions, the system ensures more accurate planning and resource allocation, reducing the risk of protocol delays and cost overruns.
  • Improved Decision-Making:AI insights provided by the ODA assist CRAs in making better-informed decisions, ultimately optimizing enrollment strategies and increasing the likelihood of protocol success.
  • Automated Protocol Updates:The ODA automates the protocol update process, saving CRAs valuable time and effort while ensuring that protocol records are always up to date with minimal manual intervention.
  • Streamlined Data Collection:The interactive ODA widget simplifies the process of gathering protocol information, making it quicker and more efficient for CRAs to input the necessary data through a user-friendly interface.
  • Automated Protocol Updates:By using AI-driven enrollment predictions, the system ensures more accurate planning and resource allocation, reducing the risk of protocol delays and cost overruns.

Future Possibilities and Innovations

  • Provide Next Best Actions to the CRA by providing the CRA with Protocol record values that increases the Enrollment prediction probability.
  • Bring your own Model.

Notes

  • If the customers are not able to provide the feature values required for Model predictions then the enrollment prediction accuracy decreases.
  • Current Model feature values restriction (Eg. Restricted indications allowed, The CTMS MVG Phase field contains values that are incompatible with the allowed Phase values in the Model.)

Conclusion

By integrating Oracle Digital Assistant (ODA) and AI-powered insights into the protocol planning process, organizations can significantly enhance the efficiency and accuracy of clinical trial planning. The seamless interaction between the CRA and the ODA widget streamlines data collection, improves decision-making with AI-driven predictions, and automates protocol updates. This innovative approach not only saves time and reduces costs but also increases the likelihood of successful trials, ensuring more effective and timely protocol execution.

Get Your Copy of the Whitepaper Here!

Excited to explore this innovation? Get all the details in this recipe!

For more insights or design partnerships, reach out to us at siebel_coe_grp@oracle.com

References and Additional Resources
  1. A Simple Guide to Connecting ODA Custom Component to On-prem APIs
  2. Oracle Digital Assistant Native Client SDK for Web
  3. Oracle Digital Assistant (ODA) and Oracle Mobile Cloud (OMC) Downloads