PRIME Biosciences Pioneers AI-Driven Clinical Research with SACRA: The 'Super-Aligned' Agent Enabling End-to-End Automation

Imagine a scenario in which AI-powered clinical research agents, empowered by advanced algorithms, medical logic, and professional databases, can detect subtle lesions in CT scans, accurately interpret patients’ dialectal chief complaints, keep pace with the latest global diagnostic and treatment guidelines in real time, and even automatically generate visualized chains of evidence when unexpected drug side effects emerge during clinical trials—thereby enabling researchers to make rapid assessments and verify consistency with source documents. This vision is being turned into reality by a Chinese technology company.

On the behind-the-scenes battlefield of pharmaceutical R&D, clinical researchers grapple daily with the “three major mountains”:

● Data Maze: Navigating the data labyrinth, electronic medical records, imaging reports, and laboratory data... Data in various formats are like scattered puzzle pieces, making manual organization time-consuming and error-prone.

● Knowledge Black Hole: Struggling to explore the edge of the "knowledge black hole," with millions of new medical publications added globally each year, traditional search methods are akin to finding a needle in a haystack. Clinical trial protocol designs are becoming increasingly complex, and the application of various knowledge standards, such as WHODrug and MedDRA, across different protocols and clinical diagnosis and treatment records lacks uniformity.

● Decision Fog: Navigating through the fog of decision-making, key decisions such as identification of adverse events/serious adverse events (AE/SAE), determination of dose-limiting toxicity (DLT), dose escalation schemes, determination of maximum tolerated dose/maximum administered dose/recommended phase 2 dose (MTD/MAD/RP2D), assessment of partial response/progressive disease/stable disease (PR/PD/SD), and adjustment of trial protocols require simultaneous consideration of medical logic, regulatory requirements, and patient safety.

SACRA, the "Super-Aligned Clinical Research Agent" developed by Purui Life Sciences, acts as an "AI brain" for clinical research operations, resolving industry bottlenecks through three core capabilities:

1. Data Alignment: Tagging Data with "Gene Labels"

● Intelligent Conversion: Automatically convert unstructured data, such as physicians’ handwritten medical records, ward-round voice consultations, and patient audio recordings, into standardized electronic data (clinical-trial-grade structured EMRs). This seamlessly integrates physicians’ familiar clinical diagnostic and treatment pathways with clinical trial methodological workflows, eliminating the need for additional procedural burdens when participating in trial projects.

● Data Provenance: Each data point comes with its own "birth certificate," offering full traceability from collection to analysis, akin to securing data with a "blockchain lock" to support CRAs in data provenance. This fundamentally eliminates data tampering and entry errors, making clinical monitoring and audits more convenient and intuitive.

● ALCOA/ACCA Global Data Standards: Purui integrates global data governance standards required by GCP into its super-aligned AI agent system designed for clinical research, thereby meeting the ACCA and ALCOA requirements for source data (Accuracy, Consistency, Traceability, Legality, Originality, Completeness, and Timeliness).

2. Knowledge Alignment: Building a "Living Clinical Trial Pharmaceutical Knowledge Base"

● Global Standards Hub: Real-time integration with 20+ international medical standard libraries, including OMOP, CTCAE, and CDISC, to automatically complete terminology mapping (e.g., aligning "rash" with CTCAE definitions and grading), ensuring data compliance with global regulatory requirements such as those of the FDA and CDE. It enables normalization of heterogeneous case and examination data from hospitals across China, documented by physicians with diverse backgrounds.

● Dynamic Knowledge Engine: Captures global medical literature, regularly updates publications from mainstream journals and authoritative medical guidelines such as NCCN and ASCO, and deeply integrates the latest global medical literature, diagnostic and treatment guidelines with localized clinical experience. The “librarian” precisely matches research needs with medical knowledge nodes, supporting the underlying logic of intelligent decision-making.

3. Decision Alignment: The "Golden Pair" in Human-Machine Collaboration

● Risk Warning Officer: In clinical drug trials, AI can automatically analyze patient data and generate visualized chains of evidence to help rapidly identify signals of adverse effects, thereby assisting researchers in quickly grading and assessing adverse events.

● Chief Decision Advisor: Establish a "human-in-the-loop" collaborative mechanism to create complementarity between AI recommendations and expert judgment, ensuring both efficiency and adherence to safety baselines.

1. Mixture of Experts (MoE): Empowering AI with Specialized Strengths

Much like specialist physicians in a hospital, each AI "expert" focuses on handling specific disease areas. When cardiovascular data is encountered, the system automatically activates the cardiology expert model; when diabetes-related data is detected, it switches to the endocrinology expert model, significantly enhancing accuracy. Purui has developed over 30 disease-specific models through its sub-specialty strategy, comprehensively covering the therapeutic areas required for clinical research.

2. Agent Engineering Pipeline: The "Smart Factory" from Idea to Implementation

● Requirements Analysis: Decompose complex requirements into atomic tasks, match optimal data resources using dynamic knowledge graphs, and provide a data/knowledge engine for sub-disease expert models.

● Model Validation: Rapidly iterate through modular testing to build a capability portfolio for AI agents.

● Industrialized Production: Automated Model Distillation Factory + Adaptive Data Routing + Knowledge Feedback Mechanism, enabling AI to efficiently process information in production environments while continuously evolving.

In practical applications, Purui’s SACRA has already demonstrated remarkable capabilities:

● CRF Form Completion: The accuracy of automatic population for relatively simple structured data, such as basic patient information and laboratory values, can reach 99%, potentially reducing manual review time for such data by 80% in real-world scenarios. SACRA also demonstrates remarkable potential in handling unstructured data; the logical reasoning capabilities and structured text generation of large language models provide robust support for scenarios such as data entry and monitoring.

● Adverse Reaction Signal Monitoring & Event Identification: Within the project, the AI agent leverages CTCAE knowledge integration to detect adverse event signals in real time, assisting Principal Investigators (PIs) in event identification and automatically annotating the results within case records. Meanwhile, it displays the reasoning chain, basis for judgment, and corresponding source data points, enabling researchers to rapidly make final determinations based on the AI-generated outputs while providing feedback to refine the results and accelerate model evolution.

● Concomitant Medications: By comparing the full dataset of patients participating in the trial, the agent rapidly identifies concomitant medication events, automatically records concomitant medication details using standard terminology (including start date, frequency, dosage, end date, and indication), performs preliminary classification of concomitant medications (CM) and links them to adverse events (AE), thereby alerting the Principal Investigator (PI) to potential protocol deviations or areas requiring improvement.

Through SACRA, Puhui is realizing its vision of “letting data speak, letting knowledge think, and infusing decisions with human warmth.” This will not only enhance the quality of clinical trials in China but also propel the entire pharmaceutical industry into an era of “automated industrialization.”

When AI truly comprehends the value of human life, and when technology and ethics are deeply integrated, what we are witnessing is not merely a technological breakthrough, but also a reverence for and commitment to life. The revolution in clinical research has only just begun.