Lunit and Agilent Partner on AI-Based Companion Diagnostics to Advance Precision Oncology

Lunit and Agilent partner to advance AI-based companion diagnostics for precision medicine

Lunit (KRX:328130.KQ) and Agilent Technologies (NYSE: A) announced a nonexclusive collaboration to co-develop AI-based companion diagnostic (CDx) solutions for research and clinical trials. The partnership combines Lunit's AI for biomarker analysis with Agilent's leadership in tissue-based diagnostics to improve diagnostic accuracy and quantify therapeutic response.

The initial work will apply Lunit's algorithms to Agilent assays to evaluate biomarkers critical to new drug programs. The goal: help pharma teams accelerate CDx development, standardize reads across sites, and increase confidence in endpoint measurement.

What product teams should know

• Scope: Co-development of AI-based assays and software that score biomarkers on digitized tissue slides (e.g., IHC/ISH) and output standardized readouts for trials.
• Primary users: Pathologists, biomarker leads, clinical ops, and CRO partners needing consistent, reproducible scoring across global sites.
• Expected outcomes: Higher concordance with expert reads, reduced inter-reader variability, faster enrollment decisions, and clearer CDx acceptance criteria.
• Deliverables: Validated algorithms, predefined scoring modules, QC dashboards, and APIs for LIS/LIMS and image management systems.

Proposed workflow

• Input: Whole-slide images from partner sites; ingestion via image management platform.
• Inference: Algorithm computes biomarker scores with confidence intervals and QC flags.
• Output: Structured results (score, cutoff status, QC metrics) exported to EDC/LIMS and trial databases.
• Deployment: On-prem or secure cloud; integration with site scanners and existing validation processes.

Validation and regulatory path (CDx)

• Analytical validation: Accuracy, precision, repeatability/reproducibility (site, operator, instrument), LoB/LoD where applicable, lot-to-lot stability.
• Clinical validation: Association with clinical benefit; pre-specified cutoffs; bridging to final assay configuration; reader studies against gold standards.
• Guidance: See FDA resources on companion diagnostics (FDA CDx overview).

KPIs for product leads

• Pathologist concordance (vs. consensus panel) and variance reduction across sites
• Turnaround time per case and percent of auto-scored cases without pathologist override
• QC failure rate, retraining frequency, and model drift alerts by site/scanner/stain
• Sensitivity/specificity vs. clinical outcomes; lot-to-lot and site-to-site reproducibility
• Onboarding time for new sites and training completion for end users

Key risks and mitigations

• Stain/scanner variability: Use diversified training sets, rigid QC gates, and site-specific calibration.
• Data drift: Continuous monitoring, drift dashboards, and controlled model updates.
• Human factors: Clear UX for QC flags, audit trails, and fail-safes to manual review.
• Interoperability: Standards-based APIs, validated integrations with LIS/LIMS and EDC.
• Regulatory readiness: Documentation mapped to FDA/CE expectations and pharma submission timelines.

Executive perspective

"Biomarker testing is at the heart of precision oncology, but today it is still largely dependent on manual interpretation," said Brandon Suh, CEO of Lunit. "By combining Agilent's global leadership in tissue-based diagnostics with Lunit's proven AI algorithms, we can help pharma partners bring biomarker-driven therapies to market faster and with greater confidence - ultimately ensuring patients receive the right treatment at the right time."

According to Nina Green, vice-president and general manager of Agilent's Clinical Diagnostics Division, the collaboration strengthens Agilent's ability to deliver advanced companion diagnostic solutions to pharma partners and patients worldwide.

What to watch next (6-12 months)

• Early-access programs with pharma and CROs; assay readouts used as exploratory or secondary endpoints
• Multi-site analytical validation with standardized SOPs and scanner panels
• Prospective clinical studies and joint presentations at major oncology and imaging conferences
• APIs and reference integrations with leading image management and LIMS vendors

Action list for product development teams

• Map the end-to-end workflow (scanner → image store → algorithm → QC → report → EDC/LIMS) and define interface contracts.
• Predefine analytical acceptance criteria and reader study designs with biostats and pathology leads.
• Stand up MLOps for regulated environments: versioning, traceability, locked models, and change control.
• Plan site enablement: training, QC procedures, and periodic proficiency testing.
• Align with pharma partners on labeling intent and submission strategy for future CDx claims.

About Lunit

Founded in 2013, Lunit is a global leader in AI for cancer diagnostics and therapeutics. Lunit develops AI solutions for medical imaging and biomarker analysis to enable precise diagnosis and personalized treatment. The FDA-cleared Lunit INSIGHT suite supports cancer screening at over 7,000 medical institutions in 65+ countries, while Lunit SCOPE is used in research partnerships with global pharma focused on biomarker development and companion diagnostics. Lunit studies have appeared in leading journals and at major conferences such as ASCO and RSNA. Learn more at lunit.io.

Further resources

• Biomarker testing basics and patient context: NCI biomarker testing
• Upskilling your team for AI-enabled product work by role: Complete AI Training - courses by job