About {Health}Re·gen
Research context, scientific motivation, methodology, and performance overview for the HealthRegen predictive oncology platform.
Global Context
According to the World Health Organization, cancer remains one of the leading causes of death worldwide. A lifetime incidence of around 40% highlights the urgent need for earlier screening and predictive assessment tools.
Our system focuses on understanding how mutations affect DNA-repair and growth-control genes, processes which, once compromised, lead to unchecked cell proliferation and malignant tumor development.
Our Vision
We believe the future of medicine is predictive. {Health}Re·gen is not just a computational tool, but a research platform that integrates genetics with environmental factors to offer a more complete picture of individual health.
Its goal is to support earlier, safer, and more actionable decisions by turning complex medical data into calibrated risk estimates that can guide prevention and follow-up.
Research Poster
Full poster overview of the project’s motivation, workflow, clinical interpretation, results, and genetic framework.
If the poster does not appear, make sure the file is placed exactly at assets/HealthRegen-1.pdf.
HealthRegen’s Accuracy Through Time
Longitudinal view of model accuracy together with upper and lower uncertainty bounds across the recorded evaluation timeline.
Genetic Framework Behind the Model
The platform is based on the idea that cancer risk emerges from the interaction between inherited genomic variation, environmental pressure, biological triggers, and probabilistic effects that shape the final phenotype.
Monogenic vs Polygenic Disease
The poster distinguishes two main categories of genetic disease. Monogenic diseases are caused by a single pathogenic sequence change in the human genome, while polygenic diseases result from multiple sequence changes across the genome acting together with environmental factors.
Cancer belongs primarily to the second category. This means that meaningful prediction cannot rely on a single mutation alone, but on broader pattern recognition across multiple biological and contextual signals.
Why Cancer Requires Pattern-Based Prediction
Because cancer is polygenic, the model must identify correlated sequence clusters and the conditions that activate them, instead of simply checking whether one mutation is present or absent.
This is why the project combines neural-network pattern detection with calibrated risk estimation: to detect likelihood and possible location of tumor development before symptoms emerge.
Clinical Meaning of the Output
The poster positions HealthRegen as clinical decision support rather than a diagnostic replacement. Integrated into doctors’ software, it can request missing data, guide the next relevant medical checks, flag possible predisposition, and help indicate which lifestyle factors should be maintained or modified.
Phenotype Formula
This formula summarizes the core logic of the project: the phenotype emerges from interaction, not from one isolated variable.
Illustrative Mutation Risk Table
The poster includes BRCA1 and BRCA2 example risk comparisons across several cancer categories, contrasted with general population risk levels.
| Cancer Type | General Population Risk | BRCA1 | BRCA2 |
|---|---|---|---|
| Breast | 12% | 50–80% | 40–70% |
| Second primary breast | 3.5% within 5 years, up to 11% | 27% within 5 years | 12% within 5 years, 40–50% at 20 years |
| Ovarian | 1–2% | 24–40% | 11–18% |
| Male breast | 0.1% | 1–2% | 5–10% |
| Prostate | 15% (N. European origin), 18% (African Americans) | <30% | <39% |
| Pancreatic | 0.50% | 1–3% | 2–7% |
Why the Poster References Huntington’s Disease
The Huntington example is used as a conceptual bridge. Once a clear inherited sequence relationship was mapped, risk became detectable much earlier.
HealthRegen extends that reasoning to cancer, a much more complex case in which the signal is distributed across multiple genomic variations and environmental interactions rather than one decisive mutation.
Sample of Patterns Found for White Females in the US, Age 30+
Illustrative distribution of selected modifiable risk contributors across cancer categories, including alcohol, weight, physical inactivity, processed meat, and low dietary fiber.
Predictive Interpretation over Irreversible Intervention
The project argues that calibrated prediction from routine data is a safer and more actionable approach for diffuse cancer-risk assessment than irreversible genome editing in context-dependent scenarios.