About Hesperion
What is Hesperion?
Hesperion is a research‑driven initiative using AI to learn from large, consented, de‑identified datasets of patients with autoimmune diseases. The goal is to uncover patterns that can help earlier detection, more precise subtyping, and better treatment matching in the future.
Where is Hesperion based?
Norway. The project begins with Norwegian partners and aims to collaborate internationally as it matures.
Who founded Hesperion?
Hesperion was initiated by Rafał Wróblewski. Visual identity and early product design are developed in‑house; clinical and data science work will be done with partners and advisors.
Is Hesperion a medical device or a replacement for doctors?
No. Hesperion is not a medical device at this stage. It does not provide diagnoses or treatment instructions and does not replace clinical judgment.
Which diseases are in scope?
The starting point is autoimmune diseases, with an initial pilot in Rheumatoid Arthritis (RA). The scope will expand to additional conditions over time.
Who is Hesperion for right now?
Primarily researchers, clinicians, healthcare organizations and patient organizations interested in building ethically‑governed datasets and evaluating AI models. It is not a direct‑to‑patient product yet.
How it works
What kinds of data will Hesperion use?
De‑identified clinical data (e.g., demographics, labs, medications, diagnosis codes, procedures, encounter timelines, limited imaging metadata), and later—where approvals and consent permit—patient‑reported outcomes and optional omics. Exact data types depend on partner agreements and ethics approvals.
How will the data be protected?
We design for GDPR and applicable Norwegian law. Identifiers are kept separate by data controllers; Hesperion receives de‑identified research copies under Data Processing Agreements (DPAs). Access is role‑based and auditable; data are stored within the EEA with encryption at rest and in transit. Final technical details are agreed case‑by‑case with partners and their compliance teams.
Where will the data be stored?
Within the European Economic Area (EEA) on vetted infrastructure. Specific providers and regions are chosen with each partner to meet their security and compliance requirements.
Who can access the data?
Only authorized personnel bound by contracts and confidentiality, and only for approved research purposes. All access is logged and auditable. Partners retain control over their contributions per the governing agreements.
Will you ever link back to identifiable patient records?
No. Hesperion does not receive direct identifiers. Re‑identification is neither permitted nor attempted. Any optional re‑contact (e.g., for prospective studies) would be handled by the original data controller under new consents and approvals.
Do you take data from the website chatbot?
No. The chatbot is for general information only. Please do not share personal health information there. If a user provides sensitive details by mistake, we will delete them as soon as feasible and ask the user to contact their clinician.
What AI methods will you use?
We focus on multimodal learning across structured EHR, timelines and unstructured notes (via privacy‑preserving NLP), with methods for missing data, uncertainty estimation, and subgroup discovery. Models are benchmarked and stress‑tested for bias and drift.
What will the first outputs look like?
Research dashboards and reports for partners: cohorts, outcome trajectories, candidate subtypes, risk scores (for research use only), and hypotheses for prospective validation.
Will Hesperion publish results?
We plan to publish de‑identified, aggregated findings with partners and present at scientific venues, subject to agreements and approvals.
Ethics, safety & compliance
Are you compliant with GDPR?
Yes—GDPR compliance is a foundational requirement. We operate under data processing agreements, minimize data, and implement data subject rights workflows with partners.
What ethical approvals are required?
Studies using health data require approvals from relevant ethics bodies (e.g., Norwegian committees) and, where applicable, registry/IRB permissions. Hesperion runs projects only after required approvals are in place.
How do you manage model bias and fairness?
We pre‑specify fairness metrics, stratify performance across subgroups, monitor drift, and include clinicians and patient advisors in reviewing outputs. If performance is inadequate or unfair, the model is not deployed.
What are the limits of the models?
Predictions are probabilistic and depend on the data and clinical context. Outputs are intended for research and hypothesis‑generation until clinically validated. They must not be used as sole evidence for diagnosis or treatment decisions.
Will patient consent be required?
Yes—either explicit consent or another valid legal basis determined with the data controller and ethics bodies (e.g., approved registry research). Consent/opt‑out mechanisms are respected per law.
Who is Hesperion for right now?
Primarily researchers, clinicians, healthcare organizations and patient organizations interested in building ethically‑governed datasets and evaluating AI models. It is not a direct‑to‑patient product yet.
Partnerships & participation
How can a hospital or clinic partner with Hesperion?
Contact us to discuss scope, governance, security reviews and contracts. We co‑design data flows that keep identifiers with the controller and provide Hesperion with de‑identified research copies.
What do partners receive?
Joint research plans, access to aggregated insights, co‑authorship opportunities, and (later) pilot access to research tools that may inform care pathways once validated.
How do patient organizations get involved?
We welcome patient groups as advisors on priorities, consent language, and communication. This improves relevance and trust.
Can individuals upload their own medical data now?
No, not at this stage. We will announce if/when a secure, consented patient portal becomes available.
Funding & governance
How is Hesperion funded?
We are preparing grant applications in Norway, exploring collaborations, and welcoming philanthropic support. We are also open to aligned investors who share the mission and standards.
Are donations tax‑deductible?
Tax status depends on the receiving legal entity and jurisdiction. Please contact us for the current status and documentation before donating.
Who owns the data and the IP?
Each data partner retains control of its original data. Hesperion derives aggregated insights and models under the agreements. Intellectual property and publication rights are defined in partner contracts to ensure fair and ethical use.
Do you have an ethics or advisory board?
An advisory structure is being assembled, including clinicians, data scientists, legal/privacy experts and patient representatives. We will publish members once finalized.
Product & roadmap
When will there be a tool doctors can use?
We are working toward validated research tools first. Any clinical decision support would follow successful prospective validation and, if applicable, a regulated pathway.
Will there be a patient app?
Possibly in a later phase, to collect patient‑reported outcomes and share educational content—only with proper consent and security.
What is the timeline?
Timelines depend on partnerships and approvals. The near‑term focus is the RA pilot (data architecture + model baselines), followed by expansion to additional autoimmune diseases.
Website, chatbot & communications
What languages do you support?
English initially; Norwegian is planned; Polish support is available for communications. We will add more languages as resources allow.
What should I not share with the chatbot?
Please do not share personal health information, identifiers, or anything you consider confidential. The chatbot is for general information only.
How can I contact Hesperion?
General: office@hesperion.no
Partnerships: work@hesperion.no
Media: media@hesperion.no
Donations/Grants: grants@hesperion.no (or contact General if unsure)
