Privacy Policy
Introduction
TrialFinder provides general information about clinical trials. Data collection is kept to a minimum. There are no accounts, and no personal data is tied to you.
Resources & educational content
The Resources section does not collect personal data beyond what is described elsewhere in this policy. Individual articles are not tracked. Page views in this section are included only in the same privacy-oriented, aggregate analytics (Umami) used across the site.
What we do not collect
There are no user accounts, and there are no marketing or advertising cookies. Google Analytics and similar advertising or ad-tech platforms are not used.
Analytics (Umami)
TrialFinder uses Umami (hosted at cloud.umami.is) for anonymous, aggregate usage information, including page views and optional custom events. There are no accounts. Umami does not rely on advertising cookies. Search terms can appear in anonymous usage events. Uploaded document contents are not sent to Umami—only non-sensitive metadata such as language, filter settings, upload flow stage, input mode (file or pasted text), and field names that were edited. Technical processing may occur on Umami’s infrastructure; see https://umami.is/privacy
Cookies and browser storage
Language preference: a small first-party cookie (name: locale) stores Hebrew or English for the next visit. It is not used for advertising or cross-site tracking (SameSite=Lax, long-lived path=/).
Recent searches: the last few unique queries (typically up to five) are kept in the browser’s sessionStorage by default, so they can be tapped again during the visit. If “Remember searches on this device” is turned on, that list is stored in localStorage instead. The list itself is not uploaded to TrialFinder servers; only the active query is sent when a search runs. The list can be cleared at any time using “Clear” on the home page.
The search term appears in the page URL as the q parameter, so it can show up in browser history and in shared links.
Theme (light/dark): the choice may be stored in the browser (typically localStorage by the theme library) and is not used to identify anyone.
This browsing session only: an optional search snapshot may be kept in session storage so “back to search results” can restore the query. It is cleared when the tab is closed (and is not a name or email).
Anonymous data
Search terms are stored in aggregate and anonymous form—keywords of the resolved search term and a result count—to understand what people look for. There are no accounts, and no IP address or session id is stored with those terms. Anonymous click events (for example, a click on a contact link) may also be stored. These records are not linked to the local “Recent searches” list. Hebrew phrases that needed translation are kept in a cache for 60 days so the translation is not repeated.
See also “Cookies and browser storage” above for how recent searches are kept on the device.
Medical document upload
A PDF or image of a medical summary can optionally be uploaded, or clinical text pasted, to help pre-fill the search. Text extraction (including OCR for scans) and removal of common personal identifiers run in the browser before anything is sent to TrialFinder servers. Original files are not uploaded or stored on TrialFinder servers.
Only scrubbed text is sent for category classification and structured clinical profile extraction using Google Gemini. Automated redaction is not perfect and may miss some identifiers; review the extracted details and scrubbed preview before confirming. Do not upload another person’s records without proper authorization.
Artificial intelligence
Search text is sent to Google’s Gemini models for translation, spelling correction, and for classifying the condition and extracting a profile that personalizes ranking. Plain-language summaries (Hebrew or English, depending on site language) are generated using Google Gemini based on public trial protocol text. When medical document upload is used, scrubbed clinical text (not the original file) may also be sent to Gemini for extraction.
AI-generated summaries (via Gemini) are experimental and for informational purposes only. Information may be inaccurate or incomplete. Always verify details with the official trial listing and a qualified clinician.
Storage and security
Data is stored in Supabase (PostgreSQL) according to the provider’s security practices. API keys are kept in server-side environment variables only.
External links
Links to ClinicalTrials.gov, Buy Me a Coffee, and other sites are subject to those sites’ privacy policies.
TrialFinder is not responsible for the content, availability, privacy practices, or conduct of third-party trial coordinators, sponsors, or external medical or research websites. Use of those services is at your own risk.