Some of the most important questions in medicine aren't being asked because they fall between disciplines, affect too few patients, or challenge the models we already have. Bloomed Health exists to build what's missing: open research frameworks for mechanistic paradoxes, clinical tools for conditions that lack them, and agentic AI that makes expert-level reasoning accessible to anyone who needs it.

We work across genomics, cancer predisposition, endometriosis and adenomyosis, stroke and vision loss, comparative oncology, and chronic illness, wherever the gap between what patients experience and what the research infrastructure provides is widest.

Everything we build is open source. If it helps one researcher ask a better question or one patient understand what's happening to them, it was worth making.

POLE Frameshift — Research framework for POLE c.138del (p.Leu46Phefs*8), an ultra-rare frameshift variant that truncates the POLE catalytic subunit at residue 54 yet produces a clinical phenotype indistinguishable from classical PPAP (TMB >100 mut/Mb). Five candidate mechanistic models, prioritized experimental approaches, therapeutic strategy, and full literature synthesis. Resolving this paradox could restructure how POLE truncation variants are classified across clinical genetics.

Cancer Predisposition Variant Analyst — Claude Code skill for ultra-rare variant interpretation in cancer predisposition genes. Resolves genotype–phenotype paradoxes, assembles ACMG/ClinGen evidence, and maps mechanism to therapy. 72 curated references (2024–2026). Includes a worked example: POLE truncation producing PPAP against consensus.

Rate of Growth Estimator — Evidence-based web application for estimating growth patterns of benign gynecologic masses using validated clinical models and current research data. Designed for patients and clinicians navigating conditions like fibroids and endometriomas where growth trajectory directly informs treatment decisions.

Blindness Visualizer — A comprehensive tool that interprets multiple visual impairments as layered visual overlays, making the subjective experience of vision loss communicable to clinicians, caregivers, and the general public. (In progress.)

• Comparative oncology — cancer biology and shared predisposition pathways
• Agentic research tools — AI workflows for literature synthesis, variant interpretation, and experimental design
• Endometriosis/adenomyosis research frameworks

Build what doesn't exist. If a tool, framework, or dataset would help patients or researchers but nobody has made it, that's what we work on.

Open by default. Every project is shared under permissive licenses. Science that stays behind closed doors helps no one.

Patient-driven, research-grade. The people closest to a disease often understand its gaps better than anyone. We translate that understanding into rigorous, citable work.

Bridge disciplines. The most important discoveries happen at the boundaries — between genomics and immunology, between human and veterinary oncology, between clinical observation and mechanistic biology.

We welcome contributions across all projects. Whether you're a bioinformatician, clinician, patient-researcher, or developer, there's work to be done. See individual repository READMEs for specific contribution guidelines, or open an issue to start a conversation.

Individual projects specify their own licenses. Research frameworks are typically shared under CC BY 4.0. Software tools use MIT or equivalent permissive licenses.

_{Bloomed Health is an open-source research organization. Nothing here constitutes medical advice.}