# Keiji AI — Comprehensive Guide

> This document provides an in-depth overview of Keiji AI, the TrialMind platform, product capabilities, research, team, blog content, compliance, and academic programs. For a concise version, see https://keiji.ai/llms.txt

---

## 1. Company Overview

**Keiji AI** (https://keiji.ai) builds TrialMind, an agentic AI platform that accelerates clinical trials across the entire lifecycle — from drug discovery and trial design to patient recruitment, data monitoring, biostatistics, and regulatory reporting.

**Founded by:** Jimeng Sun, Professor of Computer Science at the University of Illinois Urbana-Champaign (UIUC), former Global Head of AI Research at IQVIA, named Top-100 AI Leaders in Drug Discovery.

**Mission:** Accelerate and improve clinical research through the power of artificial intelligence.

**Key facts:**
- Peer-reviewed publications in Nature, Nature BME, Nature Communications, NeurIPS, EMNLP, ICLR, KDD, and other top venues
- Trusted by AbbVie, Regeneron, Takeda, Guardant Health, Medidata, and Beth Israel Deaconess Medical Center
- SOC 2 Type 2, HIPAA, and ISO 27001 compliant
- Headquartered in the United States

---

## 2. TrialMind Platform

TrialMind is a B2B SaaS platform built on specialized AI agents and proprietary MCP (Model Context Protocol) tools purpose-built for clinical research.

**Architecture:**
- Agentic AI system with specialized agents for each clinical trial phase
- Model Context Protocol (MCP) tools for standardized AI integration
- Agent Skills architecture for domain-specific capabilities

**Integrations:**
- PubMed, Embase, Cochrane, and other biomedical literature databases
- ClinicalTrials.gov, EudraCT, and global trial registries
- EHR systems for patient matching and real-world data
- CDISC (SDTM, ADaM) and OMOP data standards
- MedDRA coding for adverse events
- Genomic repositories for multi-omics analysis

**Data standards supported:** CDISC SDTM, CDISC ADaM, OMOP CDM, MedDRA, ICH E6(R2), PRISMA

---

## 3. Product: Use Cases by Phase

### 3.1 Discovery

#### Target & Mechanism Discovery
- **URL:** https://keiji.ai/product/discovery/target-mechanism-discovery
- **Description:** AI-powered identification of novel therapeutic targets and mechanism of action analysis.
- **Capabilities:**
  - Multi-omics data integration for target identification
  - Pathway analysis and mechanism of action modeling
  - Target druggability assessment and prioritization
  - Disease-target association mining from literature
  - Genetic evidence scoring and validation
  - Competitive target landscape analysis
- **Typical users:** Discovery Scientists, Computational Biologists, Research Directors, Target Assessment Teams

### 3.2 Planning

#### Real-World Data Analysis
- **URL:** https://keiji.ai/product/planning/real-world-data-analysis
- **Description:** End-to-end RWD analytics for cohort feasibility, survival analysis, treatment patterns, and safety signals.
- **Capabilities:**
  - Patient Feasibility Analysis: Quantify eligible patient populations under protocol-like inclusion/exclusion criteria
  - Line of Therapy Reconstruction: Reconstruct treatment lines, sequencing, and adherence from claims or EHR data
  - Mutation Detection Rate: Measure biomarker prevalence and testing rates across cancer types and methodologies
  - RWD Time-to-Event & Outcomes: Estimate survival, progression, and utilization using validated definitions
  - Clinical Outcomes Analysis: Compare outcomes across treatments, biomarkers, and patient subgroups
  - Temporal Trend Analysis: Characterize how diagnostics and treatments evolve over time
  - Treatment Pattern Analysis: Analyze treatment adoption, switching, and sequencing by disease and biomarker
  - RWD Safety & Pharmacovigilance: Detect and quantify safety signals using longitudinal and disproportionality analyses
- **Typical users:** Clinical Scientists, Epidemiologists, HEOR Scientists, Data Scientists, Medical Directors

#### Trial Design
- **URL:** https://keiji.ai/product/planning/trial-design
- **Description:** AI-assisted protocol development and optimization for efficient trial execution.
- **Capabilities:**
  - Protocol synopsis generation from study objectives
  - Eligibility criteria optimization based on historical data
  - Endpoint selection informed by regulatory precedents
  - Sample size estimation with adaptive design support
  - Competitive trial landscape analysis
  - Risk assessment and mitigation planning
- **Typical users:** Clinical Scientists, Medical Directors, Regulatory Affairs, Biostatisticians

#### Digital Twin
- **URL:** https://keiji.ai/product/planning/digital-twin
- **Description:** Simulate trial outcomes and patient responses using advanced predictive models.
- **Capabilities:**
  - Virtual patient population generation
  - Trial outcome simulation with confidence intervals
  - Arm comparison and effect size estimation
  - Dropout and enrollment rate prediction
  - Sensitivity analysis for design parameters
  - What-if scenario modeling
- **Typical users:** Biostatisticians, Clinical Scientists, Data Scientists, Medical Directors

### 3.3 Initiation

#### Site Selection
- **URL:** https://keiji.ai/product/initiation/site-selection
- **Description:** Identify and rank optimal trial sites based on performance data and patient access.
- **Capabilities:**
  - Historical site performance scoring and ranking
  - Patient catchment area analysis
  - Investigator experience and expertise matching
  - Geographic and demographic optimization
  - Enrollment velocity prediction by site
  - Site risk assessment and mitigation flags
- **Typical users:** Clinical Operations, Feasibility Managers, Site Relationship Managers, Project Managers

#### Patient Recruitment
- **URL:** https://keiji.ai/product/initiation/patient-recruitment
- **Description:** Precision patient matching and recruitment optimization powered by AI.
- **Capabilities:**
  - Automated patient-trial matching against eligibility criteria
  - EHR-based patient identification and pre-screening
  - Recruitment channel optimization and targeting
  - Diversity and inclusion planning tools
  - Enrollment projection and bottleneck identification
  - Patient journey mapping and engagement optimization
- **Typical users:** Patient Recruitment Specialists, Clinical Operations, Site Coordinators, Diversity Officers

### 3.4 Execution

#### Data Monitoring
- **URL:** https://keiji.ai/product/execution/data-monitoring
- **Description:** Risk-based monitoring with AI-powered data quality surveillance and site risk assessment.
- **Capabilities:**
  - RBM risk framework development per ICH E6(R2)
  - Risk identification and prioritization by severity
  - Monitoring plan and governance structure generation
  - Site risk scoring and heatmap visualization
  - Targeted site visit agenda generation
  - Data quality and consistency checks
  - Lab value distribution and outlier detection
  - Outlier patient identification and flagging
  - Automatic query generation and recommendations
  - Key Risk Indicator (KRI) configuration and tracking
- **Typical users:** Clinical Operations, Data Managers, Central Monitors, Medical Monitors, Quality Assurance

#### Safety Signal Analysis
- **URL:** https://keiji.ai/product/execution/safety-signal-analysis
- **Description:** Proactive adverse event detection and safety signal management.
- **Capabilities:**
  - Automated adverse event coding and classification
  - Signal detection using statistical and ML methods
  - Causality assessment support
  - Aggregate safety reporting and trend analysis
  - SUSAR identification and expedited reporting
  - Benefit-risk assessment dashboards
- **Typical users:** Pharmacovigilance Scientists, Safety Physicians, Drug Safety Associates, Medical Monitors

### 3.5 Analysis

#### Biostat Analysis & Programming
- **URL:** https://keiji.ai/product/analysis/biostat-analysis-programming
- **Description:** Automated SDTM/ADaM construction, validation, and statistical analysis.
- **Capabilities:**
  - Automated SDTM dataset generation from raw data
  - ADaM dataset construction with full traceability
  - Conformance validation against CDISC standards
  - Statistical analysis program generation (SAS, R)
  - TLF (Tables, Listings, Figures) automation
  - Define.xml and reviewer's guide generation
- **Typical users:** Biostatisticians, Statistical Programmers, Data Standards Leads, CDISC Specialists

#### Outcome Prediction
- **URL:** https://keiji.ai/product/analysis/outcome-prediction
- **Description:** ML-powered prediction of trial outcomes, success probability, and timelines.
- **Capabilities:**
  - Trial success probability estimation
  - Primary endpoint outcome prediction
  - Timeline and milestone forecasting
  - Comparative effectiveness modeling
  - Subgroup response prediction
  - Go/no-go decision support analytics
- **Typical users:** Clinical Scientists, Data Scientists, Portfolio Managers, Medical Directors

#### Trial Intelligence
- **URL:** https://keiji.ai/product/analysis/trial-intelligence
- **Description:** Comprehensive trial landscape analysis, eligibility comparison, and competitive intelligence.
- **Capabilities:**
  - Global trial search across registries (ClinicalTrials.gov, EudraCT, etc.)
  - Competitive landscape mapping and visualization
  - Eligibility criteria comparison and benchmarking
  - Trial outcome prediction based on historical data
  - Enrollment trend analysis and forecasting
  - Therapeutic area and indication tracking
- **Typical users:** Competitive Intelligence Analysts, Clinical Scientists, Business Development, Strategy Teams

### 3.6 Reporting

#### Regulatory Writing
- **URL:** https://keiji.ai/product/reporting/regulatory-writing
- **Description:** AI-assisted generation of regulatory documents and submission materials.
- **Capabilities:**
  - Clinical study report (CSR) drafting and assembly
  - Protocol and amendment generation
  - Investigator brochure updates
  - Regulatory response document preparation
  - Safety narrative writing
  - Submission-ready formatting and QC
- **Typical users:** Medical Writers, Regulatory Affairs, Clinical Scientists, Submission Managers

#### Literature Review
- **URL:** https://keiji.ai/product/reporting/literature-review
- **Description:** Systematic literature review automation and evidence synthesis.
- **Capabilities:**
  - Automated literature search across multiple databases
  - AI-powered abstract screening and full-text review
  - Data extraction with customizable templates
  - Risk of bias assessment automation
  - PRISMA-compliant reporting
  - Evidence synthesis and meta-analysis support
- **Typical users:** Medical Writers, HEOR Scientists, Systematic Review Specialists, Clinical Scientists

---

## 4. Research

Keiji AI's research program spans three pillars, each grounded in peer-reviewed publications.

**URL:** https://keiji.ai/research

### 4.1 Drug Discovery
- Multi-modal Representation Learning: MedCLIP (EMNLP 2022), BioBridge (ICLR 2024), TransTab (NeurIPS 2022)
- AI Data Science Co-scientists: DSWizard (Nature Biomedical Engineering 2025), BioDSA (2025)

### 4.2 Clinical Development
- Trial Design Optimization: Trial2Vec (EMNLP 2022), AutoTrial (EMNLP 2023), SPOT (ACM-BCB 2023), TrialPanorama (2025)
- Patient Recruitment: TrialGPT (Nature Communications 2024), InformGen (JAMIA 2025)
- Digital Twins: TWIN (KDD 2023), PromptEHR (EMNLP 2022)

### 4.3 Evidence-based Medicine
- Clinical Predictive Modeling: MediTab (IJCAI 2024), SurvTrace (ACM-BCB 2022), TransTab (NeurIPS 2022)
- Deep Research Systems: TrialMind-SLR (npj Digital Medicine 2025), LEADS (Nature Communications 2025), DeepEvidence (2025)

### TrialPanorama
- **URL:** https://keiji.ai/research/trialpanorama
- Large-scale clinical trial benchmark with 1.6M records and fine-tuned 8B parameter LLM
- Designed for trial outcome prediction and clinical trial understanding tasks

### Publications
- **URL:** https://keiji.ai/research/publications
- Full list of peer-reviewed publications across Nature, Nature BME, Nature Communications, NeurIPS, EMNLP, ICLR, KDD, IJCAI, ACM, JAMIA, npj Digital Medicine

---

## 5. Team

### Jimeng Sun — CEO & Co-founder
Professor of Computer Science at UIUC. Former Global Head of AI Research at IQVIA. Named Top-100 AI Leaders in Drug Discovery and Healthcare. AI research deployed at Massachusetts General Hospital, IQVIA, and UCB. Author of peer-reviewed publications across Nature and top AI venues.

### Zifeng Wang — CTO & Co-founder
PhD in Computer Science from UIUC. Experienced researcher for clinical trial applications. Former engineer at AWS, Tencent, Amplitude, and Medidata. Lead author on key platform papers including LEADS, TrialMind-SLR, and DSWizard.

### Brandon Theodorou — Co-founder
BS from Harvey Mudd College, PhD from UIUC. Research experience at NIH and IQVIA. Former software engineer at ServiceNow and Microsoft. Research focus on clinical trial prediction and digital twins.

### Venugopal Thati — VP Engineering
Former VP and Global Head of Platform Engineering and Enterprise Architect at CEVA. Former VP at Goldman Sachs with experience in development of financial services at scale. Brings enterprise platform engineering expertise.

---

## 6. Blog

Technical articles on AI in clinical research, systematic literature review, MCP agents, FDA/EMA regulation, and clinical data standards.

**URL:** https://keiji.ai/blog

### Blog Post Summaries
1. **Don't Delegate, Do It Yourself — The New Way to Work in the Age of AI** (Jan 29, 2026)
   - Author: Jimeng Sun | Category: Industry
   - For fifty years, the best career advice was simple: learn to delegate. As you rise, stop doing and start managing. That advice is now wrong.
   - URL: https://keiji.ai/blog/dont-delegate-do-it-yourself-new-way-work-age-of-ai
2. **The AI Regulation Revolution: How New FDA and EMA Guidance Opens the Clinical Trials Market** (Jan 15, 2026)
   - Author: Jimeng Sun | Category: Industry
   - January 2025 marked a turning point for AI in clinical trials. The FDA and EMA released landmark guidance that legitimizes AI as a core technology in drug development.
   - URL: https://keiji.ai/blog/ai-regulation-revolution-fda-ema-guidance-clinical-trials
3. **Beyond the Prompt: The System Design Engineering Behind Production-Grade Agents** (Dec 1, 2025)
   - Author: Venugopal Thati | Category: AI Agents
   - Anyone can build an AI agent in a weekend. But then "Day Two" happens. Building a production-grade agent is a distributed systems challenge.
   - URL: https://keiji.ai/blog/beyond-the-prompt-system-design-engineering-production-grade-agents
4. **Understanding AI's New Power Duo: MCP and Agent Skills in TrialMind** (Nov 17, 2025)
    - Author: Jimeng Sun | Category: AI Agents
    - An introduction to Model Context Protocol and Agent Skills — two architectural patterns transforming isolated AI models into connected, knowledgeable agents.
    - URL: https://keiji.ai/blog/understanding-ai-new-power-duo-mcp-and-claude-skills
5. **Powering the Next Generation of Clinical Research: How MCP Unleashes TrialMind Agents** (Jun 20, 2025)
    - Author: Jimeng Sun | Category: AI Agents
    - With pharma spending $50B+ annually on clinical development with 90% failure rates, MCP standardizes AI integration for clinical research.
    - URL: https://keiji.ai/blog/powering-next-generation-clinical-research-mcp-trialmind-agents
6. **Clinical Trials Demystified Part 4: AI Models in Clinical Trials** (Mar 14, 2025)
    - Author: Jimeng Sun | Category: Clinical Trials
    - How NLP, generative AI, computer vision, and predictive modeling are transforming trial design, patient recruitment, safety monitoring, and operational efficiency.
    - URL: https://keiji.ai/blog/clinical-trials-demystified-part-4-ai-models
7. **Clinical Trials Demystified Part 3: Clinical Trial Data & Standards** (Mar 12, 2025)
    - Author: Jimeng Sun | Category: Clinical Trials
    - CDISC, MedDRA, ICH guidelines, and how standardization enables AI-driven innovation.
    - URL: https://keiji.ai/blog/clinical-trials-demystified-part-3-data-standards
8. **Clinical Trials Demystified Part 2: Clinical Trial Software & Vendors** (Mar 11, 2025)
    - Author: Jimeng Sun | Category: Clinical Trials
    - The landscape of EDC, CTMS, eTMF, and emerging AI-powered solutions transforming trial operations.
    - URL: https://keiji.ai/blog/clinical-trials-demystified-part-2-software-vendors
9. **Clinical Trials Demystified Part 1: Workflow for Clinical Trials** (Mar 10, 2025)
    - Author: Jimeng Sun | Category: Clinical Trials
    - Clinical trials are essential to translating biomedical discoveries into treatments. Discover how AI is revolutionizing this $80B+ industry.
    - URL: https://keiji.ai/blog/ai-clinical-trials-primer-for-researchers

---

## 7. Compliance & Security

### Certifications & Compliance
- **SOC 2 Type 2** — Certified. Annual audit of security, availability, and confidentiality controls.
- **ISO 27001** — Certified. Information security management system.
- **HIPAA** — Compliant. PHI safeguards for healthcare data.
- **GDPR** — Compliant. EU data protection standards.

### Security Architecture
- **Infrastructure:** AWS cloud hosting, network segmentation, regular penetration testing, vulnerability scanning, secure configuration management
- **Data protection:** AES-256 encryption at rest, TLS 1.3 encryption in transit, data loss prevention controls, secure backup and retention policies
- **Access control:** Multi-factor authentication required, role-based access control (RBAC), least privilege principles, regular access reviews, privileged access management
- **Incident response:** 24/7 Security Operations Center, automated threat detection, incident response playbooks, regular drills, forensics capabilities

### Security Policies
- Information Security Policy
- Data Classification Policy
- Incident Response Policy
- Access Control Policy
- Vendor Security Policy
- Business Continuity Policy

**Security contact:** security@keiji.ai

**URLs:**
- Security: https://keiji.ai/security
- Compliance: https://keiji.ai/compliance
- Privacy Policy: https://keiji.ai/privacy
- Terms of Service: https://keiji.ai/terms

---

## 8. Academic Program

**URL:** https://keiji.ai/for-academics

TrialMind is **free for academic researchers**. The platform provides AI copilots for literature analysis, real-world data analytics, and trial design — built for clinical science.

### Active Research Camps
1. **Systematic Literature Review Camp**
   - 10+ medical schools and clinicians participating
   - Focus: AI-assisted screening & extraction, multi-center validation study, publication-ready workflows
2. **Biostatistics Programming Camp**
   - 20+ biostatisticians participating
   - Focus: Automated SAS/R code generation, TLF creation, CDISC-compliant outputs
3. **Trial Success Prediction (PTRS) Camp**
   - Pharma industry experts participating
   - Focus: Phase transition prediction, risk factor identification, decision support for Go/No-Go decisions

### Knowledge Base
- 1.3M+ publications indexed
- 850K+ protocols analyzed
- 200K+ task examples in the training corpus
- Coverage of top biomedical journals

### Eligibility
- Institutional email required (.edu, .ac.uk, etc.)
- Web-based platform, no download needed
- Contact: research@keiji.ai

---

## 9. Frequently Asked Questions

**Q: What is TrialMind?**
A: TrialMind is an agentic AI platform built by Keiji AI that accelerates clinical trials across the entire lifecycle. It includes specialized AI agents and proprietary MCP tools for tasks like literature review, trial design, patient recruitment, data monitoring, biostatistics, and regulatory writing.

**Q: Who uses TrialMind?**
A: TrialMind serves pharmaceutical companies, biotech firms, CROs (contract research organizations), and academic medical centers. Users include clinical scientists, biostatisticians, medical writers, data managers, pharmacovigilance specialists, and regulatory affairs professionals.

**Q: What data standards does TrialMind support?**
A: TrialMind supports CDISC standards (SDTM, ADaM), OMOP CDM for real-world data, MedDRA for adverse event coding, ICH E6(R2) for risk-based monitoring, and PRISMA for systematic literature reviews.

**Q: Is TrialMind compliant with healthcare regulations?**
A: Yes. TrialMind is SOC 2 Type 2, ISO 27001, HIPAA, and GDPR compliant. Data is encrypted with AES-256 at rest and TLS 1.3 in transit.

**Q: Can academic researchers use TrialMind for free?**
A: Yes. TrialMind is free for academic researchers with a valid institutional email (.edu, .ac.uk, etc.). Keiji AI also runs research camps for systematic literature review, biostatistics, and trial success prediction.

**Q: What makes TrialMind different from general-purpose AI tools?**
A: TrialMind is purpose-built for clinical research with domain-specific agents trained on biomedical data. It integrates with PubMed, ClinicalTrials.gov, EHR systems, and CDISC standards. The underlying models are backed by peer-reviewed publications from the founding team.

---

## 10. Contact & Links

- **Website:** https://keiji.ai
- **Book a Demo:** https://keiji.ai/demo
- **Blog:** https://keiji.ai/blog
- **Research:** https://keiji.ai/research
- **For Academics:** https://keiji.ai/for-academics
- **About:** https://keiji.ai/about
- **Careers:** https://keiji.ai/about/careers
- **Security:** https://keiji.ai/security
- **Compliance:** https://keiji.ai/compliance
- **Privacy Policy:** https://keiji.ai/privacy
- **Terms of Service:** https://keiji.ai/terms
- **Contact:** https://keiji.ai/contact
- **Security email:** security@keiji.ai
- **Research email:** research@keiji.ai
- **Concise LLM reference:** https://keiji.ai/llms.txt