Expert: Boost Compliance By Monitoring Social Media

Companies that fail to monitor social media for reportable complaints might be missing a key piece of the compliance puzzle, device industry expert Mike King told Medtech Insight.

King, who is a senior director of products and strategy for health information company IQVIA, pointed to regulatory mandates that require companies to capture and report adverse events in a timely manner. While the specifics of those mandates vary from country to country, it exists in some form just about everywhere, he said, adding that they’ve been strengthened further by recent changes to European policy on postmarket surveillance under the Medical Device Regulation. (Also see "Vigilance And Postmarket Surveillance: Companies Must Avoid Getting Caught Out By May 2021 Deadline" - Medtech Insight, 15 Oct, 2020.)

“If your product could have harmed someone or has, that is something you need to jump on immediately, both for the benefit of the patient and also to make sure that the regulator has confidence that the company's processes are in place and suitable to protect patient safety,” King explained.

In addition to the compliance benefits, analyzing social media can also provide a “wealth” of information around product performance and product quality, King said. He recalled having worked for a company once that accidentally shipped some boxes with no product inside. Angry customers swiftly took to social media to report the issue.

“It was moderately amusing, but incredibly aggravating because it then took a lot of work to remediate,” King said. “The clinician wasn't happy, that dealer wasn't happy, the business unit wasn't happy. But due to the speed of social media that was caught very quickly and identified as a product quality issue. We were able to get into the production file, see what the issue was, put our arms around the batch, do a health hazard evaluation, and then ultimately make a decision around whether certain actions were needed.”

Step One: Define Your Scope

Typically, companies must report any adverse events that they are or should be aware of – which includes events documented on company-owned websites and company-managed pages on major social media sites. Additionally, firms may need to monitor online forums devoted to third parties such as equipment dealers who operate in the company’s name.

“Given the volume of information that's out there, cutting through that volume in an effective way, is often a challenge. And that's where tools like AI … can support with the gathering of that data.” – Mike King

 But as King noted, gathering and sorting through that information is not always easy.

“Given the volume of information that's out there, cutting through that volume in an effective way, is often a challenge,” he said. “And that's where tools like AI … can support with the gathering of that data.”

The first thing a company needs to do in developing a social media monitoring program is to lay out the scope of the information to gather, King said: Identifying what data sources you have, what type of information they can provide, and whether that data is company-owned or controlled by a third party. In addition to company assets, these might include patient groups and other public forums.

Data gathered from those sources can then be fed into software to identify patterns that might lead to product performance claims, safety issues, or other concerns, King explained.  The software needs to be programmed to take multiple factors into account, such as where products are used, potential failure modes and the purpose of the device.

Human Review Still Critical

But once those patterns are identified, human review is still needed to follow up on the captured information. The human can use other methods, such as reaching out to the customer, to validate which cases seem to be legitimate product quality issues, which may require further review as a potential adverse event, and what’s been rejected as not needing a follow-up.

“Where there's information that's lacking, particularly whether it's a potential safety issue, that [human] follow-up is often critical for a company to gather the right information to make the right assessment around that product,” he said, adding that information gathered by humans can also be used to further train the algorithm.

There are two main challenges in deploying AI to review social media posts, King said. First, the software needs to be checked against a company’s existing complaints and adverse events data to ensure it’s watching for the right issues. And second, it’s important that the technology is just one piece of a functional end-to-end complaint handling process that’s validated to meet global regulatory requirements.