AI and Innovation in Clinical Trials: A Point-by-Point Analysis of Trially’s LLM-Powered Enrollment and Optimization Strategy
The recently published expert analysis titled, "AI and Innovation in Clinical Trials," outlines several challenges facing traditional clinical research, including high costs, enrollment difficulties, and issues with generalizability. It proposes that leveraging artificial intelligence (AI), Large Language Models (LLMs), and adaptive designs is necessary to modernize trial execution and solve these persistent problems. Trially directly addresses these challenges by leveraging LLMs and AI agents throughout the patient recruitment and feasibility lifecycle. This document provides a point-by-point analysis of how Trially addresses the challenges identified in the expert analysis (referred to hereafter as the 'study').
1. Addressing Enrollment Challenges and Operational Costs
The study notes successful implementation of clinical trials faces several fundamental hurdles, primarily due to substantial operational costs and persistent challenges in enrollment or accrual. The difficulty in identifying and recruiting participants is exacerbated by the use of elaborate protocols and stringent eligibility criteria, which impede the efficiency of locating and securing suitable candidates. As a result of widespread under-enrollment, approximately 85% of clinical studies fall behind schedule, incurring average daily costs of $500,000 (according to Tufts CSDD).
Trially’s core objective is to enroll qualified patients faster. Versus conventional strategies, Trially projects a 2x to 6x boost in monthly enrollment rates for complex clinical trials. Trially's success in increasing enrollment is evidenced by a case study where a site enrolled 6 patients in 4 months after previously only enrolling one patient in 1.5 years. Furthermore, Trially offers a significant reduction in operational burden by reducing manual EHR chart review hours by 91% (or hundreds of hours per month) for Clinical Research Coordinators (CRCs). This reduction in manual labor saves time and money, addressing the persistent challenge of high operational costs mentioned in the study.
2. AI and Eligibility Optimization
The study argues that there is a defined need to use Artificial Intelligence (AI) and machine learning (ML) to refine the rules for who can join a study (AI-driven eligibility optimization). Overly rigid qualification rules often disqualify large segments of the patient population that could otherwise profit from the treatment. Training ML models on historical data and using LLMs to parse unstructured clinical data can highlight which criteria are truly necessary and improve patient matching.
Trially directly aligns with this goal through its proprietary AI matching engine, Trially Match, which uses LLM agents to instantly match patients to trials. Trially achieves a high accuracy rate of approximately 95% in screening patients against eligibility criteria. Additionally, Trially synthesizes 100% of EHR data—including both structured data (like clinical history, medications, labs) and unstructured data (like notes, imaging, free text)—to identify high-quality candidates. This ability to analyze all clinical context facilitates better pre-qualification and leads to lower screen failure rates. Trially has demonstrated a 73% reduction in screen failure rates due to the higher quality of patients prescreened. Furthermore, the platform rapidly parses complex protocols into Inclusion/Exclusion (I/E) criteria cards in under 5 to 10 minutes using AI.
3. AI Agents for Autonomous Coordination and Patient Engagement
The study introduces the concept of AI agents, which are systems able to perform autonomous, goal-directed actions across the clinical trial lifecycle. These agents mimic the roles of human personnel, such as coordinators or analysts, by executing steps based on the trial protocol and patient data. The study also highlights that Large Language Model (LLM) systems are already proven to make the process of matching patients to the right trial much more accurate.
Beyond the technical aspects, the study emphasizes that patient engagement is still extremely important for a trial's success and how LLMs can take complicated, unstructured clinical data (like doctor's notes) and automatically convert it into formats that are easier to read and understand. The study suggests that LLMs can be used to make communication between research teams and potential participants much better by:
Tailoring Communication: LLMs can customize the content of trial materials to match the specific language preferences and community background of prospective participants. This is known as cultural adaptation and goes beyond simple translation.
Simplifying Documents: They can simplify dense documents, such as consent forms.
Ensuring Understanding: This process helps participants to fully grasp the study's procedures, risks, and benefits.
Trially is already offering the necessary tools to meet these challenges by using its Trially Connect platform and Margo AI Agent to automate and enhance the entire patient workflow, maximizing patient conversion, and replacing time-intensive manual lead follow-up which is a root cause of trial problems. This comprehensive solution ensures patients are engaged and qualified for enrollment efficiently, moving the trial from matching leads to confirmed enrollments.
Margo is designed to perform specific, automated actions, such as pre-screening patients based on upcoming visits, scheduling appointments and sending reminders (via SMS and Voice), and actively re-engaging dormant patients to build a trial-ready population. This automated process efficiently pre-screens and schedules the warmest leads, replacing the manual, time-intensive "Patient-Physician Handoff" process and thereby streamlining site operations. To enhance patient communication and transparency, Margo also generates real-time patient qualification summaries and call summaries with documentation, ensuring the clinical context is fully captured and reducing the overall complexity of the trial experience for prospective participants.
4. Enhancing Feasibility and Optimizing Trial Planning
The study highlights the need for advanced methods to refine clinical research, such as using simulations to identify likely sources of failure and refine protocols before trials launch. Data-driven decision making earlier in the process is essential to reduce costly failures.
Trially’s platform enables early, data-driven decisions through Trially Intelligence and Feasibility Analytics. The Feasibility Explorer provides real-time insights across sites for planning and business development. Trially allows sites to analyze their patient population based on demographics, medications, and conditions. By utilizing the AI matching engine, Trially can generate Exact Patient Counts for sponsors, moving beyond guesswork and manual reviews to provide proof of feasibility. This capability has proven successful in helping site networks win multiple studies and awards, addressing the issue of protocol mismatch and improving site selection. Trially is also working on a Protocol Refiner function, directly supporting the study’s objective of optimizing and refining I/E criteria.
5. Infrastructure, Data Security, and Transparency
The study dedicates a section to outlining the infrastructure and ethical challenges that must be overcome for widespread AI adoption in clinical research. Specifically, the study argues that successful AI integration requires sophisticated infrastructure, high-performance computing to perform complex simulations at scale, and a critical need for robust data management. Emphasized is that:
transparency and rigorous validation are essential to build trust among clinicians, researchers, and participants;
regulatory frameworks must evolve alongside these technological advancements to establish rigorous standards for validating, interpreting, and approving AI technologies, thereby ensuring their reliability and clinical efficacy.
the “black-box” nature of AI systems (complexity making it difficult to understand the rationale behind decisions made by the AI) be resolved.
Trially emphasizes superior safety, data security, and privacy protections, maintaining full compliance with stringent international and U.S. standards, including HIPAA, SOC 2, FDA CFR 21 Part 11, and ISO 27001. To address concerns about data quality and provenance, Trially ensures PHI protection through AES-256 and TLS encryption and employs Data Isolation techniques, such as isolated storage and private servers per client, adhering to regulatory mandates. Lastly, Trially combats the "black-box" challenge by promoting transparency, allowing users to "triple click" into patient matching criteria to discern physician notes and understand the exact rationale for why a patient qualifies.
Trially's use of LLMs and AI agents to parse complex EHR data, automate eligibility matching with high accuracy, and autonomously engage patients addresses the primary challenges of cost, enrollment, and efficiency outlined in "AI and Innovation in Clinical Trials". By providing validated, scalable frameworks that prioritize data security and transparency, Trially is working toward the study’s vision of shifting clinical trials from static, rigid processes to responsive, individualized systems. Trially acts as a digital accelerator for recruitment, much like a modern GPS system instantly computes the fastest route by analyzing real-time traffic (EHR data) and road closures (I/E criteria), ensuring the trial reaches its destination (enrollment goal) quickly and efficiently.
