Chatbot Potential in Clinical Research

Pharmaceutical CommercePharmaceutical Commerce - November 2021
Volume 16
Issue 4

How AI-powered support automation can solve the data problem in clinical trial management

Throughout the life of a clinical study, research associates, site coordinators, principal investigators and patients contend with an overwhelming volume of data: documentation, visits, calls, video meetings and emails. Executing, collecting, recording and filing trial documents may seem like a mundane part of the study process. However, the information within those documents can mean the difference between a trial that moves to the next phase and one that stalls out.

From interviewing and prescreening hundreds of potential patients to onboarding site staff, maintaining compliance and tracking patient progress, clinical trial teams face a host of competing priorities in need of their valuable time and skills. Neglecting any facet of the life of a study has consequences, none more so than the ability to recruit and retain the right patients to give a trial the best chance to succeed.

AI-powered support automation platforms offer solutions for patients and trial administrators, providing an interactive, informative interface, integrated cloud storage, intelligent document processing and a centralized knowledge base that keeps critical information accurate, accessible, and secure. Potential patients can learn about trials and start enrollment processes, while trial team members can complete patient selection, site compliance and trial analysis tasks more efficiently.

Improve the patient experience, starting with recruitment

Deloitte projects that all clinical trial stakeholders will move toward processes that center the patient experience in an effort to recruit and retain committed patients to support clinical trial goals and get more trials to the finish line. An AI-powered support automation platform can improve clinical trial recruitment and retainment with features that cater to potential patients while facilitating faster, more efficient recruitment selection for trial goals.

The recruitment process often starts with a website designed to give straightforward information to prospective patients who fit the criteria needed for a given trial. While a website with study information helps patients learn about the study itself, it doesn’t help research organizations glean any information about their prospective recruits. For that, companies often rely on site visitors to contact them via an email address on the website. Then, trial team members spend time reaching out to each prospect, responding to questions and interviewing the prospects, first to see if they qualify for the trial and then to learn whether they’re interested in participating and likely to stay with the trial. Since not all the interested candidates will qualify, and some will decide against participating, team members waste time and resources on those lengthy interviews.

By implementing an AI-powered conversational chatbot, organizations open up the line of communication on their recruitment websites. Prospective patients can ask questions about the study and get instant responses about a trial. The chatbot answers by leveraging an organization’s knowledge base, which is constantly updated to ensure accuracy. That means the chatbot gives the most current answer every time. If it can’t find an answer, it seamlessly escalates the request to a human expert who can answer a site visitor’s inquiry. The human expert’s responses are then added to the chatbot’s knowledge base, improving its abilities as it works.

While the chatbot informs site visitors, it also determines whether prospective patients fit trial criteria, saving team members hours of time and effort and providing them with a smaller pool of better choices for a trial. A chatbot learns about those prospective patients through guided conversations created and customized by trial administrators, giving the chatbot clear paths to follow as site visitors respond to its questions. An AI-powered chatbot can start the recruitment process, collect patient information and conduct prescreening questionnaires, making it easier for study coordinators to step in and complete the process.

Equip team members with actionable information, quickly

While much trial data is processed through electronic data capture, built-in system redundancies often mean humans are still manually sorting and moving records, leaving room for human error in e-filing and categorizing. And because each trial uses multiple clinical sites required to provide the same kinds of paperwork at regular intervals, the chances of sending documents to the wrong people or filing them in the wrong place are high. Too often, clinical trial documentation takes a slow journey from clinical site to research associate, finally landing on the desk (sometimes literally, as paper documents continue to plague the industry) of a trial assistant already working through an avalanche of papers.

Clinical research associates need documents to be timely, organized and accessible. That ideal scenario facilitates accurate reporting, follow up and next steps. Frictionless retrieval of accurate trial documentation, enabled by support automation, is essential to a clinical study’s compliance, safety and success.

The same chatbot that engages potential patients—answering questions and directing them to relevant information—also works for trial team members, mining an organization’s knowledge base to respond to inquiries about trial processes, contact information and site communication. Rather than searching for a specific document or data point in multiple places or interrupting a co-worker, trial administrators can ask the chatbot. It will guide them to the answer, whether it’s the expiration date on a medical license, an overview of a study population or a site coordinator’s phone number.

As part of an AI-powered support automation platform, the chatbot can also initiate customized workflows that use robotic processing automation (RPA) to streamline work processes. Workflows initiate a series of actions, ensuring the right team members, documents and processes are engaged—all while integrating with other systems to send emails, upload records or create action items. By integrating workflows, team members don’t have to think through the trajectory of a document each time they encounter a new one. They can work with a chatbot to follow a path that includes what needs to happen with a record, who should see it and what follow-up processes it triggers.

Intelligent document processing can function in tandem with workflows, allowing a trial team to create and follow templates for different kinds of documents, then batch upload trial records for processing. Machine learning classifies the documents against the customized templates, mines the data and pulls a human expert into the process to deal with any missing information. From there, trial teams can export the data, push the record onto its next step in a workflow or refine the template to capture information more precisely. Teams can verify the templates and provide feedback, so the system improves with use. The more documents are uploaded, the better the AI gets at sorting. And once records have been added, they become part of the knowledge base team members can access through the chatbot.

While a good support automation platform equips trial teams with the information they need, it must protect the security and privacy of the clinical work. Its deep knowledge base should have robust access controls that mirror the permissions trial team members have across applications to ensure the right people have access to the right information. That means if a team member asks the chatbot for information they can’t access, the chatbot won’t give it to them. In the same way, when a team member requests sensitive data, the system should provide it but keep the data private by not caching it in its server.

About the Author

David Karandish is Founder and CEO of Capacity.

Related Videos
© 2024 MJH Life Sciences

All rights reserved.