Generative artificial intelligence (AI) and large language models (LLM) are on a path to revolutionize healthcare in a myriad of ways, from facilitating drug discovery, through to Chatbots providing first-port-of-call health advice to patients. Generative AI refers to a subset of AI models capable of generating new content, including text, images, audio, and video, which is similar to human-generated content.

In health economics and outcomes research (HEOR), generative AI can enhance the efficiency of research processes, generate deeper insights from healthcare data, and improve clinical decision-making. This in turn can make the clinical development process and ultimately patient management more efficient, accurate, and patient centered.

At the recent International Society for Pharmacoeconomics and Outcomes Research, Inc (ISPOR) congress, there was a real focus on the application of generative AI in HEOR1. Here are some of the top use cases presented:

Innovative AI for Biomedical Text Analysis: The landscape of generative AI is rapidly evolving, with a particular emphasis on creating biomedical LLMs that allow us to better understand and process medical texts. By pre-training LLMs with biomedical text, these sophisticated models can complete tasks such as deciphering clinical notes, creating clinical narratives, and systematically reviewing and analyzing clinical evidence on a large scale.

Transforming Real-World Data into Actionable Evidence: AI technologies, including natural language processing and LLMs, are at the forefront of converting the unstructured data found in electronic health records into structured, usable information. This pivotal transformation enables the expansion of their use in real-world evidence studies, allowing for the assessment of burden of disease, comparative effectiveness and safety, and unmet need in a real-world population.

Harnessing AI to Tap into Patient Experiences on Social Media: AI is unlocking the potential of "social listening" by systematically capturing and summarizing self-reported outcomes on social media platforms. This innovative approach is instrumental in creating dynamic cohorts that capture the holistic value of an intervention, something that is especially important for chronic disorders.

Impact of LLMs on Clinical Trial Programs: From the vantage point of the pharmaceutical industry, the integration of generative AI into clinical trials is a game-changer. This includes applications as broad as study endpoint determination, streamlined patient identification and screening, and advanced data analysis and evidence synthesis.

Despite the promise of generative AI, challenges such as computational demands, data quality, privacy issues, and the need for transparent AI methodologies remain. However, these do not seem to be dampening the enthusiasm for the transformation that generative AI is promising. Many pharmaceutical companies are deploying generative AI across their research & development and commercial divisions.

“Like every industry, HEOR functions need to assess their workflows and understand where deploying generative AI will have the most impact. Resources are always limited, and successful deployment of an AI solution often takes considerable investment, so it is important to focus on the workflows with the greatest opportunity for transformation” Shanida Nataraja, Senior Director, Real World.

In HEOR, and across healthcare more broadly, generative AI has the potential to transform the way we conduct research and allow us to provide more precise, efficient, and evidence-based insights that inform healthcare policy, clinical practice, and patient care. So, are you embracing AI in your HEOR programs? If not, now is the time to join the journey; after all, by streamlining research, and the processes through which we generate real-world clinical insights, we can hopefully bring innovative medicines to patients faster than ever before!

References

  1. ISPOR 2024 Conference Session: The Future of Data-Driven HEOR Decision-Making Powered by Generative AI: How Soon is Now?