Regulatory Intelligence (RI) Meets Artificial Intelligence (AI): A Revolution in the Waiting?

Artificial Intelligence (AI) is revolutionizing the world, particularly the Life Sciences domain. Be it large leaps that we have taken in the field of genomics research and drug discovery or large language models such as ChatGPT, which is making inroads into traditionally less-automated areas such as medical transcriptions – one area which remains sheltered from the exponential benefits of Artificial Intelligence (AI) is the domain of Regulatory Intelligence (RI).

It is surprising, given how critical and ubiquitous Regulatory Intelligence is across the Life Sciences industry- be it pharmaceuticals or medical devices or even consumer goods such as cosmetics and food supplements.

A recent third-party survey* (https://gens-associates.com/2023/05/01/regulatory-intelligence-2022-study-whitepaper/) has revealed that less than 30% of the organizations have put AI to good use, while only 17% have harnessed the power of AI to automate day-to-day RI tasks. Thus, organizations still have to rummage through tons of documents (most of which will become obsolete in no time), spend countless hours to derive that one critical insight and bear the brunt of ballooning technology and consultant costs as they go global.

There is always a risk that something will slip through the gap, leading to unforeseen compliance costs, reputation loss from product recalls, and unacceptably long timelines for market expansion.

Can AI plug this gap? In this blog, we shall explore some of the key shortcomings of the RI space and the key use cases where AI can be put to good use.

Role of AI in RI

RI activities can broadly be considered as a set of sequential steps – a journey from data management to decision-making. The four (04) key steps in the journey include:

1. Data Gathering
2. Data Curation  
3. Data Analysis
4. Data Consumption
   1. Data Search
   2. Data-driven Decision-making

Here is a detailed look into each stage:

• Data Gathering: One of the key complexities of RI information is that it has got dual dynamicity. There are many sources, and the sources update information regularly, rendering the corresponding older information obsolete. This presents a humongous challenge for busy RI professionals, who spend hours monitoring hundreds of websites or spend thousands of dollars on consultants. Even with the most diligent record-keeping, it is practically impossible to keep track of all the regulations and guidelines, which significantly increases the risk of non-compliance.
  A potential solution is to deploy intelligent web-scraping bots that will automate the data collection processes. While certain RI information providers offer Robotic Process Automation (RPA) services, the options are limited to only a few websites. Specific configurations to cater to customers’ requirements are either not possible or have long Turn Around Times (TATs).
• Data Curation: The second challenge that Regulatory personnel might face in their journey toward decision-making is related to data curation. The voluminous data obtained from diverse sources are often in local languages, are super lengthy, or can hide critical insights under a mound of complex text matter.
  This leads to escalating costs of consultations and translations, long timelines, and a critical risk of missing out on the key essence of the guidelines. While off the shelf translation services and auto-summarizers might make a weak attempt to fix these problems, the need of the hour is a robust summarizing and translating tool that has been trained specifically for Regulatory content.
  In addition, the organization of information is often too simplistic to capture the complex interrelationships between Regulatory concepts. This presents an uphill challenge – classifying documents into need-based buckets, which is a prerequisite for speedy access to curated information.
• Data Analysis: Even in a “perfect world” scenario, where the curated data is clean and immaculate, making sense of the data requires someone to engineer information and present it in a way that enables the Regulatory personnel to get a bird’s eye view of all the data and take critical decisions. The two (02) roadblocks in this step of data analysis are that the Regulatory personnel lack both time and great visualization software that is feature-loaded, easy to use, and economical at the same time. The Regulatory domain needs a workforce to diligently manage data and make complex information amenable to analysis and decision-making.
• Data Consumption: While the internet has freed the data from the locked doors of policymakers, a subsequent side effect of this data tsunami is an overwhelming sense of “analysis paralysis.” Cutting through this clutter requires intuitive search tools that can comprehend what the end user is looking for and a strong recommendation system that can complement the Regulatory personnel’s decision-making.

With ChatGPT on the horizon, it might seem enticing that these challenges will be resolved! But we know AI hallucinates and ChatGPT is a black-box tool that does not provide a direct reference to the source of information. In a domain with very little margin of error and a strong need for information traceability, such drawbacks would render the garden variety of ChatGPT to be of not much value.

The Freyr Solution

At Freyr, we understand the challenges of this journey. Therefore, at every step, we are not using AI for AI, but we are trying to solve customer problems.

Journey Map

Step 1 – Data Collection: Intelligent web-scraping bots.

Step 2 – Data Curation: Auto-translators and auto-summarizers trained on Regulatory ontology.

Step 3 – Data Analysis: Decision tree based on decades worth of precedence data and AWS Quicksight with rich, easy-to-use, cost-effective managed dashboards (Expert-curated dashboards).

Step 4: Data Consumption: Context-trained and context-aware ChatGPT.

Conclusion

While AI has revolutionized various domains, RI has remained relatively untouched by its benefits, with only a very small percentage of organizations harnessing AI for RI tasks.

By employing intelligent web-scraping bots, auto-translators, summarizers, advanced visualizations, and context-trained ChatGPT, Freyr aims at providing effective RI solutions. Dive into a world of cutting-edge technology and witness firsthand the remarkable capabilities of Freyr’s dedicated RI platform – Freyr Freya.Intelligence. Request a demo!

Don’t miss out on this incredible opportunity to explore the future of regulatory intelligence.