Biomarker Discovery

Wearable sensor data and multi-omics profiles predict disease biomarkers with improved accuracy and validation rigor.

Business impact

• ΔR² = 0.040 for depression prediction — Improved model fit increases accuracy in predicting depression severity
• ΔR² = 0.021 for insulin resistance prediction — Better biomarker features enhance metabolic disease risk prediction
• Reduced AI development time from weeks to hours — Faster experimentation accelerates biomarker discovery and validation

Data requirements

• Wearable sensor time-series data (Numeric) — Captures continuous physiological signals for biomarker candidate generation
• Whole-slide pathology images (Image) — Provides spatial and morphological features for cancer biomarker discovery
• Proteomics liquid biopsy data (Structured) — Enables molecular signature identification for early cancer detection
• RNA sequencing and circulating tumor DNA (Structured) — Integrates multi-omics data to enhance biomarker validation accuracy
• Scientific literature and clinical trial data (Text) — Supports literature-grounded reasoning and hypothesis validation

AI methods and techniques

• Agentic AI — Orchestrates multi-agent workflows for hypothesis generation and validation
• Predictive AI — Models biomarker-disease associations to improve prediction accuracy
• Generative AI — Forms novel biomarker hypotheses and interprets complex biological data

AI models and model families

GPT-4o, Claude, Llama, Custom deep learning models

Industries

Real-world evidence

8 documented case studies on record.

Companies using this: BioAI, Google, MultivisionDx, NEC OncoImmunity AS, Nautilus Biotechnology, Oxford Cancer Analytics, PathAI, École polytechnique fédérale de Lausanne (EPFL).