top of page
  • decodeMR Team

Adverse Event Reporting in Healthcare Market Research: Four Ways to Leverage Artificial Intelligence

In healthcare market research, professionals are bound by legal obligations to diligently report adverse events (AEs) related to any drug they encounter while conducting research studies. The reporting of AEs is imperative as it contributes to the real-world evidence for a marketed drug and plays a key role in assessing a drug's safety profile.

Healthcare market research agencies invest substantial efforts in ensuring that their training programs incorporate up-to-date modules on AE identification and reporting procedures. Their goal is to equip project team members to promptly recognize AEs and report them efficiently. However, despite rigorous training and established processes, these agencies still grapple with various challenges in achieving AE-reporting compliance.

This article seeks to shed light on our experiences and insights to date on how Artificial Intelligence (AI) and Natural Language Processing (NLP) can transform the approach to AE reporting in healthcare research.

Figure 1: Illustration on common complaints about AE reporting. This has been developed based on our more than 10 years of experience and interviews with our fieldwork partners

AI Technologies are Impacting Every Walk of Life

In recent years, AI has made a profound impact on various facets of our lives, redefining how we work, communicate, access information, and make decisions. Whether in finance, where AI detects fraud; the legal profession, where AI analyzes contracts, or medicine, where AI accelerates drug discovery and diagnostic improvement, AI technologies have permeated every aspect of human existence.

As AI technologies continue to evolve, they are poised to expand their influence, presenting both new opportunities and challenges. This presents an exciting moment to harness these technological advancements and streamline our approach to AE reporting.

Embracing AI: Four Ways to Address AE Reporting Challenges

We foresee four major avenues where AI adaptation promises to improve the speed and precision of AE reporting:

1. Automated Text Analysis

NLP algorithms have the capability to streamline the examination of market research interview transcripts or recordings in near real-time. These algorithms can be trained to swiftly identify keywords and phrases associated with potential AEs. For example, if a participant mentions "experiencing cardiac events after taking Drug A," NLP can promptly flag "cardiac events" as a potential AE-related remark.

AI can take this a step further by creating a visual representation of the transcript as a sentiment map, categorizing user-selected key terms into positive, negative, and neutral themes (refer to Figure 2). With a simple click on an adverse event, users can access associated phrases from the transcript, facilitating quick reference and enhancing the overall research experience.

Figure 2: A screenshot of a sentiment map of a transcript depicting the list of cardiac events mentioned and grouped under negative sentiment

2. Contextual Analysis

NLP algorithms excel in contextual analysis when identifying AEs, effectively distinguishing between casual references and potential adverse events. For instance, they can differentiate between phrases like "cardiac events are scored based on a certain framework" and "Drug A is associated with the risk of cardiac events."

Figure 3: A screenshot of a sentiment map depicting mentions of 'cardiac events' categorized into a Neutral theme considering the context of the discussion

3. Enhanced Standardization

NLP can play a crucial role in standardizing and structuring extracted information, ensuring consistency in reporting. This standardization is vital for meaningful comparisons and analyses. For instance, NLP can efficiently transform different expressions of the same AE (e.g., "heart failure," "congestive heart failure," "HF," "CHF") into a standardized terminology, simplifying data aggregation and analysis.

Figure 4: Screenshots displaying the prompts with alternative terms of a specific AE mentioned

Notice how AI could take it a step further (refer to Figure 4) by suggesting common synonyms for a specific AE, empowering users to tag and standardize the data for further analysis.

4. Automated Reporting

Leveraging AI-driven parsing tools and pre-configured AE report templates, generating automated AE reports becomes highly efficient. These advanced tools can also extract, organize, and structure relevant data from various sources, ensuring that AE reports adhere to predetermined formats and standards. This expedites the report creation process and minimizes errors, enhancing the quality and consistency of AE reporting.

decodeMR’s Tryst with AI-driven AE-reporting Process

At decodeMR, we believe embracing innovative ideas and technologies is the most rewarding approach to creating transformative solutions for our stakeholders. Our ongoing endeavor involves the seamless integration of capabilities of AI/NLP with our proprietary qualitative research tool, which we have affectionately codenamed 'QAT.'

We have had our fair share of ups and downs in this exploration. On the positive side, we could harness AI to accurately categorize the term' cardiac events' within a neutral theme (refer to figure 3), all while considering the surrounding discourse. On the downside, we encountered challenges related to AI-induced hallucinations. Despite the hurdles, this collective experience forms a robust foundation for our future advancements.

Bringing Everything Together and the Way Forward

The four potential use cases we have explored – 'Automated Text Analysis,' 'Contextual Analysis', 'Enhanced Standardization,' and 'Automated Reporting' - represent just a glimpse of what is achievable. Given the extensive availability of various DIY AI tools, it must be noted that there are limitless possibilities for crafting solutions.

As we wrap up, let us consider the wisdom from a Chinese proverb,

"The best time to plant a tree was 20 years ago. The second-best time is now."

By embracing AI and ML now, we do not just alleviate these challenges; we unlock a future of precision, efficiency, and, most importantly, improved patient care.


bottom of page