Innovative Ways AI is Changing Real-World Data Analysis
In the life sciences, the sources and volume of real-world data (RWD) have exploded. With new sources like wearable devices, social media, socioeconomic data, EHRs, and patient profiles combined with traditional sources such as clinical trials, claims, and diagnostics, the amount of data the average life sciences firm has to parse through has grown exponentially. Some estimates suggest that by 2030, the size of the market for RWD could more than double from 2019 numbers.i
GROWTH OF RWE MARKET
The global real world evidence solutions market size was estimated at USD 2.6 billion in 2023 and is expected to grow at a CAGR of 8.4% from 2024 to 2030.ⁱ
This wealth of new data offers new opportunities to identify patterns across broader and more complete data sources. Numerous new sources of data (some even directly from patients) provide new sources of analyses and new patterns to be identified. However, that comes at an increasingly expensive, timely, and resource-intensive data management effort. Further, many traditional solutions are not optimized to use this extensive amount of data (especially unstructured data).
AI and their powerful large language models (LLMs) offer an unprecedented opportunity to generate insights across this extensive data, transforming healthcare and drug development. However, handling this vast amount of unstructured data is challenging. This is where large LLMs like OpenAI’s GPT and others come in. With their ability to process and analyze natural language at scale, LLMs are revolutionizing how life sciences companies utilize RWD to improve patient outcomes and accelerate innovation.
Here, we’ll explore five innovative ways life sciences use LLMs for real-world data analysis.
1. ACCELERATING DRUG DISCOVERY AND DEVELOPMENT
AI and LLMs are proving instrumental in early-stage drug research and discovery, where analyzing large volumes of complex and expansive datasets like medical literature and clinical trial data is critical. One of the most valuable untapped resources of RWD is the unstructured data included with electronic health records (EHRs), clinical trials, and patient data. However, these tools’ ability to extract meaningful insights from unstructured data sources has unlocked this valuable data source for researchers. They now allow researchers to identify novel drug candidates faster.
Additionally, these LLMs have the ability to conduct data analyses comparing disparate studies, identifying novel biomarkers, and even generating hypotheses that aid in designing new clinical trials, shortening the drug development timeline and reducing costs.
Example: AstraZeneca uses AI-driven tools to scan millions of scientific papers to find relationships between genes, diseases, and drugs, helping identify new targets for drug discovery. ii
ACCELERATING DRUG DISCOVERY
Companies like AstraZeneca use AI-driven tools to scan millions of scientific papers to find relationships between genes, diseases, and drugs, helping identify new targets for drug discovery.
2. IMPROVING PATIENT RECRUITMENT FOR CLINICAL TRIALS
Recruiting patients who meet specific eligibility criteria for clinical trials is time-intensive and costly. Identifying, recruiting, and enrolling patients that fit the necessary trial parameters can be a significant expense in the early stages of clinical trials. This is significantly higher when we look at rare diseases.
LLMs are being used to reduce that time and expense while improving the quality of patient recruits. These tools can analyze electronic health records (EHRs) and patient data to identify suitable candidates based on multiple factors like demographics, medical history, co-morbidities, and more. By matching eligible patients with relevant trials, LLMs improve recruitment efficiency and ensure trials reach the required enrollment numbers.
LLMs are also beginning to play a valuable role in patient retention during trials. By analyzing feedback from patient feedback, they can identify potential issues that may impact a patient’s willingness to continue. This allows for a far more proactive approach to timely interventions, helping to keep patients engaged throughout the study.
PATIENT RECRUITMENT
Companies like IQVIA use AI-driven recruitment tools to quickly find eligible patients and match them to trials using their medical and demographic information.
3. INCREASING ADVERSE EVENT DETECTION
Monitoring and assessing adverse events (AEs) related to drugs and medical devices is a critical aspect of pharmacovigilance (the ongoing evaluation of the safety of drugs and vaccines after they are on the market).
LLMs help streamline this process by analyzing patient data, EHRs, social media feeds, and even doctor’s notes to identify potential adverse events and side effects. By automating the detection of these events, LLMs help companies take quicker action to ensure patient safety, patient engagement, and regulatory compliance.
Some models have been trained to recognize specific types of adverse events (e.g., allergic reactions), even in free-text patient complaints or online forums. This proactive identification provides earlier risk mitigation, the creation of safer therapies, and stronger patient adherence.
Example: Pfizer uses LLMs to process patient feedback from various sources, detect emerging safety signals, and improve the efficiency of their pharmacovigilance operations.v
ADVERSE EVENT DETECTION
Pfizer uses LLMs to process patient feedback from various sources, detect emerging safety signals, and improve the efficiency of their pharmacovigilance operations.
4. ANALYZING SOCIAL DETERMINANTS OF HEALTH
One of the newest sources of real world data is socioeconomic data collectively referred to as Social Determinants of Health (SDoH). Life sciences companies increasingly recognize the importance of social determinants of health, such as socioeconomic status, education, and access to healthcare, in understanding health outcomes.
LLMs are adept at analyzing the unstructured social data from diverse sources (e.g., community health reports, social media feeds, and patient surveys) to capture insights about these factors. Further, being able to integrate SDoH data with clinical and molecular data offers a more comprehensive understanding of patient populations, helping companies create more effective therapies and outreach strategies.
Example: Health tech companies use LLMs to analyze SDoH data and identify trends affecting specific demographics, enabling life sciences companies to address health disparities and improve access to care.vi
TRACKING SDOH DATA
Health tech companies use LLMs to analyze SDoH data and identify trends affecting specific demographics, enabling life sciences companies to address health disparities and improve access to care.
5. ENHANCING REAL-WORLD EVIDENCE GENERATION
Real-world evidence (RWE) is the insights derived from real-world data (RWD) analyses. It is an increasingly important part of the development process for new therapies. Companies leverage RWE insights to demonstrate the safety and efficacy of therapies outside clinical settings.
Purpose-built LLMs can sift through common RWD sources (e.g., EHRs, claims, and diagnostics) to uncover patterns that help validate patients' real-world experiences. This is especially valuable in demonstrating long-term safety, effectiveness, and quality-of-life improvements for patients, a key aspect of regulatory approval and payer negotiations.
By automating this data analysis, LLMs enable life sciences companies to generate significantly more RWE more efficiently. This results in faster product approvals, better reimbursements, and improved post-market safety monitoring.
Example: Amgen has used LLMs to extract RWE from unstructured clinical and patient data, providing insights into patient outcomes and treatment adherence, aiding in value-based care discussions with payers vii
RWE GENERATION
Amgen has used LLMs to extract RWE from unstructured clinical and patient data, providing insights into patient outcomes and treatment adherence, aiding in value-based care discussions with payers.
THE FUTURE OF LLMS IN LIFE SCIENCES AND RWD
Integrating LLMs into real-world data analysis in life sciences is just the beginning. As models become more sophisticated, their applications continue to broaden, allowing companies to generate actionable insights from even more complex and diverse data sources. From drug discovery and patient recruitment to personalized medicine and pharmacovigilance, LLMs are transforming the life sciences industry, paving the way for better healthcare outcomes and innovations.
By leveraging these AI-driven advancements, life sciences companies can continue to deliver cutting-edge therapies to patients, reducing the cost of development while increasing the safety and efficacy of these new therapies.
In the following articles, we will delve deeper into each of these initiatives to examine the processes, tools, and methodologies used to make them successful.
References:
i Coherent Market Insights, Real-world Data (RWD) Market… [Link]
ii AstraZeneca, Data Science & Artificial Intelligence [Link]
iii IQVIA: Acelerating AI and ML Adoption in Biopharma [Link]
iv Veeva, New Artificial Intelligence Application, Veeva Andi… [Link]
v Pfizer, Meet the New Digital Assistants Transforming Patient Medical Information [Link]
vi Large Language Models for Social Determinants of Health Information Extraction from Clinical Notes... [Link]
vii PitchGrade.com, Amgen: AI Use Cases 2024 [Link]
