#FutureReadyHealthcare
Share this blog
06 Sep 2023
Decoding Generative AI and ChatGPT in the life sciences
Generative AI (Gen AI) holds promise in various pharmaceutical applications, including drug development, clinical trials, diagnosis, and medical imaging. Categorized into artificial superintelligence (ASI), artificial general intelligence (AGI), and artificial narrow intelligence, AI is already making strides in the life sciences. Machine learning (ML), a specific AI type, includes Gen AI, capable of creating text, video, and images, presenting opportunities in writing, translation, and editing. However, ethical, and legal concerns accompany its potential. In life sciences, Gen AI contributes to drug information, patient support, adverse event reporting, and medical query resolution. ChatGPT, an AI chatbot utilizing natural language processing, engages in conversational dialogue and generates content like articles and code. Yet, caution is essential when incorporating AI, especially with sensitive data.
Transformative Impact of AI on Regulatory Compliance in Life Sciences
In the highly regulated pharmaceutical landscape, compliance with diverse laws and regulations, from clinical trials to production and distribution, is imperative. AI, employing advanced technologies like machine learning and natural language processing, addresses the challenges associated with vast datasets and intricate regulatory requirements. AI automates regulatory submissions through Natural Language Processing, ensuring consistency and error-free processes. Predictive analytics anticipates compliance issues proactively, and AI's rule-based learning aids in interpreting complex regulations. The technology's adaptability to different regulatory regimes globally mitigates risks and ensures compliance across jurisdictions. While the future promises more sophisticated models and tools, addressing concerns like data privacy and the need for human oversight is crucial for responsible AI integration into life sciences regulatory compliance, marking a transformative era in navigating this intricate field.
The emergence of Generative Artificial Intelligence (Gen AI) has created a focused interest within the industry to incubate and explore novel ways of working in only a few short months. This advancement is especially recognized in Regulatory Affairs. As a way to connect and hear from the industry, Indegene recently conducted a webinar on "Applying Generative AI in Regulatory Affairs: From Hype to Practical Applications." As I look to lead Indegene's Regulatory Affairs practice to new ways of working, I had the honor of moderating this insightful session, featuring distinguished industry experts: David Berglund: Executive Director, Global Head of Regulatory Operations at AstraZeneca, Sudheer Somineni: Head of Regulatory Solutions at Pfizer, and Ritesh Dogra: AVP - Medical Technology, at Indegene.
In an effort to ascertain our audience's familiarity with Gen AI, we polled attendees during the webinar with interesting findings: Approximately one-third of the attendees have already initiated adoption and experimentation within their organizations. The emergence and potential of this technology have created an opportunity for end users to truly experiment, and due to the accessibility of Gen AI, it is not surprising to see this happen in just a few short months. (Figure 1).
Three Use Cases of Generative AI in Regulatory Affairs
The discussion with the panelists explored the convergence of technology and regulatory affairs to discuss potential use cases and applicability of generative AI in regulatory affairs. An overarching sentiment is the opportunity to simplify business processes and assist the regulatory expertise in making decisions to enhance processes in select areas, including submission management, health authority query management, and medical writing. While harnessing Gen AI for drug discovery remains a distant aspiration, it's crucial to understand the complexities. The opportunity to adopt repeatable, systematic methods is a practical initiation for gen ai in regulatory affairs. Cost considerations, like training and document redaction, are also key factors to address.
When we design our solutions (using Gen AI), I think we need to take into consideration the possibilities which may not be possible today, but in a year's time, it may very well be possible.
says David Berglund, Executive Director, Global Head of Regulatory Operations, AstraZeneca.
At Indegene, the CTO established an incubation hub for every functional area to develop use cases, employing a blend of Gen AI and other models. Crafted and trained by our expert Regulatory domain team, these models yielded notably accurate outcomes in simulated real-world scenarios. Use cases encompassed: (A) Entity extraction for document assessment, (B) Document summarization, (C) Content authoring, (D) Generating data-driven insights, and ensuring regulatory compliance.
If you watch the webinar recording embedded in this blog, you will see Ritesh Dogra from Indegene demonstrate the three use cases mentioned below.
HA Query Similarity: Recommending similar queries based on historical data, improving query prediction accuracy.
Protocol to Informed Consent Form Generation: Transforming complex protocol documents into patient-friendly Informed Consent forms.
Document Search and Summarization: Uploading, searching, and generating summaries based on keywords, streamlining document processing.
Challenges Involved in Adopting Generative AI in Life Science
To set the context and understand more about the concerns of the industry, we asked the audience about their ease with adopting Gen AI. We found nearly half worried about its adoption in Regulatory Affairs, citing concerns like data privacy and security, particularly due to the involvement of patient data and the need for consistent output from Gen AI (Figure 2).
Segueing from there, the conversation transitioned toward the possible obstacles as the panel delved into the audience's apprehensions regarding the integration of Gen AI in Regulatory Affairs. In response, the panelists tackled these concerns by presenting approaches professionals can take to address concerns. Sudheer underscored key strategies: (1) Ensuring sufficient literacy of the technology, (2) Applying a systematic and responsible deployment, (3) Human-in-the-loop validation, and (4) Making use of robust APIs tailored for privacy, security, and seamless integration.
David emphasized the importance of having sufficient data, particularly highlighting the challenges smaller companies might face in terms of HAQ (Health Authority Queries). He mentioned the significance of setting confidence and learning thresholds, envisioning a scenario where the industry collaboratively addresses this issue. Additionally, David pointed out that Accumulus might serve as a potential bridge in overcoming these challenges of generative AI in regulatory affairs.
Here are a handful of gen AI solutions for regulatory affairs that will enable you to maximize the potential of Gen AI.
Concerns and Challenges
Solution
Quality and consistency
Training and refining using prompt engineering
Creating application wrappers
Trust and explainability
Implementing fail-safe quality and trust mechanisms
Process model checks and validation
Human in-the-loop validation methods for enhanced explainability
Data privacy, security, regulatory compliance, stability
Utilizing enterprise-grade APIs (e.g., Azure APIs, AWS) that address privacy, security, and stability concerns
Ensuring easy integration
Data Bias
Curating and evaluating training data to minimize bias
Ensuring fairness in generated content
Lack of control on output
Ensuring fairness in generated content and validate generated output before external use or sharing
Regulatory Compliance
Generating content in compliance with regulatory guidelines, data standards, and specific reporting formats
Providing information about the literacy of these models (like Gen AI) and allowing experimentation and control settings within small groups would go a long way.
says Sudheer Somineni, Head of Regulatory Solutions, Pfizer.
Embracing a Thoughtful Approach
As a final topic, we discussed actionable steps organizations should consider when evaluating and embracing Gen AI within Regulatory Affairs. The panel suggested the following strategic measures: enhancing organizational awareness, identifying and supporting early adopters, and creating conducive opportunities for people to drive the integration process. There was an emphasis on the significance of taking the time to understand the technology, acknowledging that we are still in the early phases of its potential, and underscoring the need to ensure compliance and repeatability of systems before deploying into production.
Where the webinar outlined a path from exploration of Gen AI in life sciences to its practical applications within Regulatory Affairs, the industry needs to take a more aggressive stance to apply a data standard which will improve confidence in the usage of systems that depend on repeatability and simplicity. Once more advance use cases start demonstrating competitive advantage through the usage of Gen AI to speed time to market, increase communication, reduce complexity in the presentation of submissions, and increasing scientific value, then we will quickly find that Gen AI becomes a necessity for those looking to stay at the forefront of innovation in Regulatory Affairs.
For further discussions and exploration of this transformative journey, feel free to connect with us at GenerativeAI@Indegene.com. Let's embark on this exciting path together.
Explore how Indegene’s AI solution cuts down time to market by 90% and improved content discoverability for a global pharma company
Share this blog
Get exclusive pharma
insights delivered to your inbox
Latest
Latest
