Leveraging AI for Clinical Trial Excellence: How Trially and Margo Complement Phesi's Global Data Strategy
The modernization of clinical trials hinges on leveraging vast patient data and advanced artificial intelligence (AI) to shift toward a more patient-centric approach. Phesi is a company that uses AI and real-world data to accelerate clinical trials. Gen Li, PhD, MBA, president of Phesi, emphasized this need for a more patient-centric approach in an interview with Applied Clinical Trials, detailing how global patient data and AI analytics accelerate site activation, strengthen recruitment, and improve trial design. While Phesi utilizes the world’s largest clinical trials database, covering over 300 million patients, Trially and its AI agent, Margo, translate this global, data-driven vision into immediate, site-level operational success through precision matching and patient engagement driven by Large Language Models (LLMs).
Phesi’s core objective is the strategic design and planning of smarter trials by leveraging its massive patient database. Dr. Li explains that Phesi can receive a synopsis from a client and, guided by those documents, recruit patients from that pool of 300 million patients, enabling sponsors to "meet your patients before you even start the trial". This insight is then used to improve trial design and avoid amendments.
Trially complements Phesi's macro-level strategy by excelling at the micro-level execution. Where Phesi provides the foundational global patient intelligence for trial design, Trially focuses on the challenge of operationalizing recruitment at the site, specifically addressing the root cause problem of Trial/Site Mismatch based on guesswork. Trially instantly matches patients to trials and trials to sites using its proprietary AI matching engine and feasibility analytics, achieving approximately 95% accuracy. This precision is achieved by leveraging LLM-assisted prescreening to parse clinical trial criteria and patient data from any source, including EHRs, CTMS, PDFs, and XML.
AI and Digital Twins: Forecasting Trajectories and Accelerating Feasibility
A significant shared objective is using advanced AI to model and predict clinical outcomes, moving clinical development toward science. Phesi generates Digital Twins, Digital Patient Profiles, and Digital Trial Arms to simulate clinical trials with patient-centric data analytics. This capability allows Phesi to foresee how the control arm is going to behave—including safety and efficacy profiles—potentially replacing part or all of the control arm or guiding future development.
This objective is supported by recent research on LLMs and digital twins, which confirms that generative AI is revolutionizing digital twin development, enabling virtual patient representations that predict health trajectories. The research highlights that LLMs, such as the Digital Twin—Generative Pretrained Transformer (DT-GPT), can forecast patient health trajectories and showcase potential advantages as clinical forecasting platforms, proposing a path toward digital twin applications in clinical trials, treatment selection, and adverse event mitigation. Importantly, these models leverage EHRs from real-world data and observational studies, overcoming common challenges like missing data and sparsity.
Trially enhances Phesi’s design stage by providing real-time, granular feasibility data. Trially’s "Pipeline Radar" proactively alerts sites to best-fit studies for their unique patient population, allowing sites to instantly send sponsors exact patient counts. This ability to provide precise feasibility assessments is critical, as traditional methods relying on manual EHR reviews are time-consuming and often imprecise. By providing this proof of feasibility with enriched clinical data, Trially helps sponsors de-risk studies—a local validation that immediately backs Phesi’s global, design-level insights. Trially Match's capability to instantly parse a complex protocol using NLP and stack-rank candidates saves valuable time, showing a 91% reduction in EHR chart review hours per month.
Margo: Translating Matches into Enrollment
Dr. Li emphasized the need to redistribute trial workloads and improve recruitment by using AI to identify and activate new sites. Trially achieves this outcome through Margo, the AI agent that bridges the gap between patient matching and final enrollment. Trially Connect, powered by Margo, enrolls high-quality patients faster by engaging, prescreening, scheduling, and following up.
Qualified candidates often slip away due to insufficient follow-up and site staff burden. Margo solves this "last mile" recruitment challenge by acting as an automated Clinical Research Coordinator (CRC) assistant. Margo prescreens patients based on upcoming visits and schedules appointments via SMS, voice, and email. Furthermore, Margo can re-engage dormant patient populations to build a trial-ready pool. This dramatically reduces the burden on site staff, freeing up CRCs who typically spend over 250 hours per month on manual screening.
LLM-powered recruitment shows statistically significant success, according to market research cited by Trially. The AI-assisted approach yielded a 20.4% eligibility rate (35 enrollments), far exceeding the manual process's 12.7% eligibility rate (19 enrollments). This superior matching quality translates to a 73% reduction in screen failure rates, fulfilling Phesi’s patient-centric goal of improving outcomes.
In summary, Phesi defines the data-driven "what and why" of smarter, more efficient trials using global data for design optimization and digital twin generation. Trially, utilizing LLMs like the DT-GPT research proposes, provides the "how and where". Trially and Margo apply precision LLM technology at the site level, leveraging local EHRs for high-accuracy patient matching and automated engagement, effectively realizing Phesi's objective to accelerate site activation and strengthen recruitment without compromising efficiency. By marrying macro-level strategic planning (Phesi) with micro-level operational execution (Trially), the industry can achieve smarter trials and faster cures.
