OncoTwin Insights

OncoTwin Insights, developed by 4baseCare, is an AI-powered comparable-patient intelligence platform built for oncology. Described as the world's first atlas dedicated to capturing cancer's genomic diversity across underrepresented populations, it helps oncologists move beyond static reports by matching each patient's genomic and clinical profile to thousands of real-world patient outcomes — delivering evidence-backed treatment insights in real time. The platform is designed for clinical teams, molecular tumour boards, and oncology decision-makers who need transparent, reproducible, and statistically grounded support at the point of care.

At its core, OncoTwin finds the most similar real-world patients — referred to as "twins" — using clinico-genomic similarity and network algorithms, then summarises treatment journeys and observed outcomes with transparent rationale. The system is built on three foundational pillars that work together to surface meaningful clinical insights.

Core Technology Pillars

• Similarity Matching: Weighted similarity is computed across clinical context, biomarkers, and genomics, with auditable drivers of match and confidence bands to ensure interpretability.
• Network Intelligence: Graph-based methods stabilise retrieval and reduce one-off coincidences, surfacing clusters of related cases rather than a single nearest neighbour for more robust comparisons.
• Peer Evidence (Statistics): Cohort-level outcome rates among similar patients are presented with sample sizes (n), covering durability, early failure, and outcome context by treatment class.

How the Matching Workflow Operates

1. Normalize: Clinical, biomarker, and genomic representations are standardised into a consistent format.
2. Constrain: Stage and driver cohort alignment is applied to avoid implausible or clinically irrelevant comparisons.
3. Score: Tuned weights, subcohort logic, and co-alteration penalties are applied to generate similarity scores.
4. Explain: Top twins are surfaced alongside match drivers and a peer evidence panel with confidence indicators.

AI-Powered Clinical Decision Support Capabilities

• Real-World Outcome Intelligence: AI trained on diverse global datasets provides treatment paths supported by real patient outcomes.
• Digital Twin Patient Models: A virtual model of each patient predicts response patterns and simulates therapy options.
• Personalized Therapy Recommendations: Insights are tailored to genomic alterations, biomarkers, comorbidities, and treatment history.
• Adaptive Evidence Engine: Insights are continuously updated to incorporate the latest studies, clinical guidelines, and real-world data.

The Twin Report

• Top-matched patient twins with quantified similarity scores and contributing match features.
• Comparative treatment trajectories, including lines of therapy, treatment duration, and progression-related signals.
• Peer evidence summary presenting outcome rates with corresponding cohort size and statistical confidence indicators.
• Risk and limitation annotations highlighting data sparsity, potential confounding factors, and data completeness considerations.

Molecular Tumour Board Enablement

• An integrated case view consolidates genomic findings, clinical history, biomarkers, and prior treatments in a single standardised format.
• An evidence contextualization layer links molecular alterations to guidelines, literature, and real-world cohort signals.
• Twin-based cohort insights provide outcome context from clinically similar patients without prescriptive inference, supporting multidisciplinary discussion.

OncoTwin Insights is designed with explicit trust and safety guardrails: it functions as decision support rather than a clinical directive, presents observational evidence without causal claims or guaranteed outcomes, provides transparent rationale for every patient match, and flags small or sparse samples with confidence-aware annotations to ensure clinicians can interpret results appropriately.