How Merck Used Bots to Assist With Regulatory Filings in 23 Geographies
With a legacy going back 350 years, global pharma giant Merck opted to re-engineer its well-established processes four years ago. The organization broke conventions to embrace robotic process automation, RPA, and bots to process voluminous documentation for regulatory filings.
Internally, Merck faced resistance to adopting newer technology, but the India and Asia-Pacific teams at Merck implemented the use of bots for a project for Korea – and succeeded. Time-consuming documentation work that earlier took six to eight months to complete can now be done in a month. For Merck’s team in North America, it once took 30% of daily time to convert 6,000 purchase requisitions into purchase orders every month. With bots developed by the India and APAC teams, this is now done with 97% accuracy and a 60% reduction in turnaround time. They have also bumped up their delivery time for external vendors from 83% to 93%.
“We were not sure if using bots for order entry would work. We’re a scientific team and not an order-entry or customer-facing team. When we suggested using bots to automate tasks, we faced a lot of resistance and were asked to show a use case,” said Dr. Radhika Mahadev, head – RPA, Merck Life Science.
Merck Life Science (Merck) produces tools, services and expertise for researchers in the pharma and healthcare industries. Its products are used by researchers for scientific experiments and innovations. The Merck Group has expertise in life science, healthcare and electronics. Its numerous business activities include medical R&D, clinical research and diagnostics.
Automation in Regulatory Paper Submissions
A core task for the Merck team is submitting papers to regulators for product approvals. Before a product (medical device, drug, or service) is launched in any country, it needs to pass through stringent regulator processes and review by subject matter experts. This is a tedious process that requires excessive documentation.
There are two types of paper submissions. The first is healthcare submissions, which go to regulatory bodies such as the U.S. FDA and EMA. The other is license submissions that are presented to the country’s regulatory boards for trade compliance.
“There is a ton of paperwork involved for both these submissions. Because it is compliance related, there are numerous checkpoints that involve subject matter experts. So, we were asked if the paperwork could be reduced or done in another way,” Mahadev said.
As Mahadev and her team set out to address this requirement, they evaluated the use of automation and RPA. They wanted to change the status quo “and do things in a different way.”
Merck receives orders from global customers through various communication channels and in different formats – email, fax, PDF and postal mail. Purchase requisitions received from customers need to be converted to purchase orders, which follow specific formats as prescribed by the country regulator. This was performed by an order entry team at Merck.
The Merck team chose the Korea leniency regulatory program for their first bot. The program checks country regulatory compliance for toxic and non-toxic products. It is mandatory to file paperwork for products with the country council before launching products in Korea, and failure to do so would result in heavy fines or cancelled licenses.
Implementing RPA: Challenges
“[Since this was Korea] we had to first build the language capability in Bengaluru – and we had to do that in six to nine months,” Mahadev said.
It was a nail-biting experience for everyone on the team, as there was no precedent. Bots had never been used for regulatory work before, and the project was met with skepticism.
“No one was willing to take risk as this was related to regulatory compliance. The developers felt that if something went wrong, the submissions would not be accepted by the government of that country, and we would lose out. I knew that the subject matter experts from the regulatory side would have even more resistance to accepting such a critical job,” Mahadev said.
The order entry team was also worried about being replaced by bots. There were doubts about how a bot could accurately convert purchase requisitions from customers to purchase orders.
“Bots were welcome because we couldn’t expect highly qualified people from the scientific community to do mundane tasks like order entry,” Mahadev said. “The scientific community was more concerned about accuracy and efficiency.”
Using AI-powered bots from UiPath and Automation Anywhere, Merck cleared the entire backlog of paperwork for some 30,000 products within a month. And with that, there was more acceptance.
Prince Kohli, CTO, Automation Anywhere, said “the definition of automation” has changed and it is no longer “just about macros”.
“Seven years ago, we were the first to introduce AI into automation via document understanding and extraction of information from documents,” Kholi said. “Most enterprises work with documents but you need something that goes beyond OCR to understand the text in those documents.”
After the success of the Korea leniency regulatory program, Merck proceeded with the next program, the Japan Industrial Safety and Health Law, ISHL, inventory.
“Like the Korea program, a set of 10 documents per product needs to be submitted to the country regulator. But with a large portfolio of our products, the paperwork is voluminous. This was the second use case for bots,” Mahadev said.
Expanding RPA Footprint: Order Screening Level
After a successful attempt in two countries, Merck adopted the technology at the order screening level. This is a crucial step in the process where a decision is taken to launch a product in a country.
The checkpoints and validations are done by human subject matter experts in various countries. This ensures that the regulator’s criteria are met in accordance with the country’s regulation policies.
“In regulatory compliance, 98% is cleared by the bot, but 2% is always pushed to human subject matter experts. This is a manual safeguard mechanism that has been put in place in case the bot is confused or cannot fix something,” Mahadev said.
Today, Merck is using bots to assist with regulatory filings in 23 geographies globally. It is also using bots in areas like supply chain operations and customer care.
Merck is now considering bots for AI projects and RPA integrated with analytics. It is also looking to use disruptive technologies to address the challenge of unstructured data.
“Almost 60% of our activities are unstructured data. We have to go through it, understand it, and then translate it. So, there are immense activities within our organization. I’m sure we will need more automation, supported by AI and ML,” she said.
For years, humans were put to the task of extracting relevant information from lengthy documents. But now this laborious task can be done by bots and RPA, and Merck has proved that it can be done.