Share4Rare

EU Horizon 2020 platform connecting rare disease patients, caregivers, and researchers across 50+ countries. B2SLab designed and built the entire data science backend — and continues to maintain it.

Keywords

rare diseases, patient data, Human Phenotype Ontology, HPO, EU Horizon 2020, Sant Joan de Déu, UPC, bioinformatics, people like me

Status	Project concluded · Platform still active
Funding	EU Horizon 2020 · Agreement #780262
Coordinator	Sant Joan de Déu Foundation, Barcelona
B2SLab role	Data science platform design, build, and maintenance
Duration	2018 – 2022 (EU project)
Platform	share4rare.org

What is Share4Rare?

Share4Rare is a European collective awareness platform built around rare disease patients, caregivers, researchers, and patient associations. Funded under the EU Horizon 2020 programme (Agreement #780262), the project set out to transform the fragmented landscape of rare disease knowledge by creating a single place where patients can learn, connect with others who share their condition, and donate their clinical data to advance research.

The platform rests on three pillars:

• Learn — access reliable, understandable medical information on rare diseases, compiled and curated by clinical experts.
• Connect — find and interact with patients and families facing the same diagnosis, anywhere in the world.
• Share & Research — donate clinical and symptom data in a secure, GDPR-compliant environment, contributing to real scientific studies and enabling patient-driven research.

The EU project ran from 2018 to 2022. The platform, however, never stopped: it continues to operate, onboard new research studies, and serve the rare disease community. B2SLab maintains the data science backend to this day.

The Share4Rare platform community — connecting patients, caregivers, and researchers globally.

---

B2SLab’s Role: the Data Science Engine

B2SLab was the data science and biostatistics partner within the consortium. Our contribution went far beyond analysis — we designed, built, and deployed the entire data science infrastructure that powers the platform’s intelligence layer.

The data pipeline

When a patient donates clinical information on Share4Rare, that data does not simply sit in a database. B2SLab built the pipeline that:

1. Standardises donated phenotypic data using the Human Phenotype Ontology (HPO) — a controlled vocabulary that allows clinical terms reported in plain language to be mapped to formal, interoperable medical concepts. This is essential in rare diseases, where patients and clinicians often use different words to describe the same finding.

2. Processes the harmonised data mathematically to extract meaningful patterns — identifying which patients share phenotypic profiles, clustering conditions by symptom overlap, and surfacing structure in data that would otherwise be invisible.

3. Feeds results back into the platform as actionable features. The most visible of these is “Personas Como Yo” (People Like Me) — a recommender system that connects patients with similar clinical profiles, letting someone with an undiagnosed condition or a rare syndrome find others who may share their journey.

NoteTechnical highlights of the B2SLab data science platform

• HPO-based phenotype standardisation: donated clinical data is mapped to a formal ontology, enabling machine-readable comparison across patients with heterogeneous reported terms
• Phenotypic similarity engine: the backbone of the “People Like Me” feature, computing similarity scores between patient profiles at scale
• Secure architecture: all data processing complies with European GDPR requirements; the system works on anonymised representations, never exposing individual records
• Ongoing maintenance: B2SLab continues to maintain and evolve the platform’s data science layer beyond the formal EU project period

The data processing approach was presented publicly by Prof. Alexandre Perera Lluna in the Share4Rare Webinar VI (How your data is processed in Share4Rare, 2019), one of a series of community-facing educational sessions on the platform’s workings.

Tip“Personas Como Yo” — People Like Me

Rare diseases are, by definition, rare — finding another patient with the same diagnosis in your city or country is often impossible. The People Like Me feature, built by B2SLab, identifies patients with matching phenotypic profiles across the entire platform community. By comparing the clinical features each patient has reported — mapped to HPO terms — the system surfaces meaningful connections that geography and chance never could.

---

The Platform Today

The EU project may have concluded, but Share4Rare keeps growing:

Active research studies and disease focus areas on the Share4Rare platform.

Platform scale (2026)

• 5,000+ patient and caregiver users
• 50+ countries represented
• Active research studies across multiple rare disease categories
• Ongoing open calls for academic research proposals

Active disease areas: - Neuromuscular disorders (incl. Duchenne MD) - Rare paediatric tumours - Undiagnosed conditions - Epidermolysis Bullosa - Phenylketonuria (PKU) - Rett syndrome - Rare cardiac, renal, and vascular diseases - Alpha-1 antitrypsin deficiency

The platform’s research programme remains open: academic centres and patient associations can submit study proposals. B2SLab continues to provide the data science infrastructure that makes these studies possible.

---

Consortium

The Share4Rare consortium brought together eight European partners:

Partner	Country	Role
Sant Joan de Déu Foundation	Spain	Project coordinator; clinical leadership
Universitat Politècnica de Catalunya / B2SLab	Spain	Data science platform; biostatistics
Omada Interactiva	Spain	Technology development
World Duchenne Organization (UPPMD)	Netherlands	Patient organisation (neuromuscular)
Melanoma Patient Network Europe (MPNE)	Belgium	Patient organisation (rare tumours)
The Synergist	UK	Social innovation
Newcastle University	UK	Clinical expertise (neuromuscular)
Asserta	Spain	Socioeconomic impact

---

Related Publications

B2SLab’s work on Share4Rare produced research contributions at the intersection of natural language processing, clinical ontologies, and rare disease informatics:

NoteKey papers

Mapping layperson medical terminology into the Human Phenotype Ontology using neural machine translation models Manzini E, Pérez Díez I, Pérez González M, Caballero E, Perera-Lluna A. Expert Systems with Applications, 204, 117446, 2022. https://doi.org/10.1016/j.eswa.2022.117446

This paper directly addresses one of the core data science challenges in Share4Rare: how to bridge the gap between the informal language used by patients and caregivers and the formal HPO terms required for computational phenotype comparison.

See also the full group publication list for additional related work on phenotypic similarity, network diffusion methods (diffuStats, FELLA), and rare disease biomarker discovery.

---

Links

• Platform: share4rare.org
• Webinar VI — How your data is processed: share4rare.org/es/news/share4rare-webinar-vi-…
• EU project record: Horizon 2020 Agreement #780262
• B2SLab blog post (2018 kick-off): B2SLab Takes Part In Share4Rare