What is Embase AI?

Embase AI is available as an add-on subscription. Contact your organization’s administrator if you are unsure whether your organization subscribes to it.

You must also be signed in with a registered Elsevier account or through your organization.

1. Open Embase.
2. Select Embase AI from the search options or navigation menu.
3. Enter a biomedical research question in Ask for biomedical insights.

You can also select one of the example questions displayed on the starting page.

Before asking your question, select Filters to control the records used by Embase AI.

Publication year

• All
• Last 2 years
• Last 5 years
• Last 10 years
• Custom range

Publication type exclusions

• Preprints
• Conference abstracts
• ClinicalTrials.gov records
• Articles in press

Study types

• Meta-analysis
• Systematic review
• Cochrane review
• Randomized controlled trial
• Controlled clinical trial
• Observational study
• Literature review
• Case report
• Animal study
• In-vitro study

Your question may adjust filter parameters when it overlaps with a filter. Select Apply filters to use the selected settings. When available, select Save as default preferences to reuse them for future questions.

Enter your question and select Ask.

Embase AI then:

1. Translates the question into Embase query language.
2. Searches the relevant Embase content.
3. Synthesizes a cited response.
4. Creates a title for the conversation and adds it to History.

While a response is being generated, the query, references, filters, follow-up field, and export controls may be temporarily unavailable.

Expand the Searching for… section associated with a question to review the generated Embase query.

From this section, you can:

• Select Open in results page to run the query on the standard Embase Results page.
• Select Copy query to copy the search strategy.

A follow-up question may produce a refined search that combines relevant context from the earlier question with the new request. Review generated queries before relying on them for formal or systematic searching.

The answer contains numbered citations linked to supporting Embase records. Depending on the question, the response may include headings, summaries, comparisons, or tables.

After each answer, you can:

• Select Copy answer to copy the response.
• Provide positive or negative feedback.
• Select a numbered citation to inspect its supporting record.
• Select Show all references to open the complete reference list for that response.

The References panel displays bibliographic information for the records used by the response.

Depending on record and institutional availability, you can:

• Open the Embase record.
• View the abstract.
• View Emtree index terms.
• Follow an available full-text or organization-specific link.
• Find similar records.

The References panel has a separate Export action for bibliographic records. You can configure the export and choose which citation, bibliographic, abstract, index-term, identifier, and link fields to include.

After each response, Embase AI may suggest three related follow-up questions based on the conversation.

You can:

• Select a suggested question to submit it immediately.
• Enter your own question in Ask a follow up question.

Each follow-up produces its own generated search and cited response. New suggestions are displayed after the latest response.

Select Start new conversation when you want to explore an unrelated subject.

Embase AI adds completed conversations to History and assigns each conversation a title based on its subject.

Select Show conversations history to:

• Reopen an earlier conversation.
• Start a new conversation.
• Rename a conversation.
• Export a conversation.
• Delete a conversation.

Conversations are grouped by date. Select a conversation title to return to it.

Select Export this conversation after the latest response, or open the conversation’s actions in History and select Export.

Conversation export is separate from the bibliographic-record export available in the References panel.

Embase AI uses content from the Embase corpus and the Emtree thesaurus.

The How it works panel states that the large language model is hosted in Microsoft Azure and that query content is not logged or used by third-party providers.

Although Embase AI uses domain-specific biomedical content and provides links to supporting references, AI-generated responses may be incomplete, inaccurate, variable, biased, or inappropriate. Responses are generated in real time without individual human review.

Always inspect the generated query and cited records, and apply scientific judgement before acting on or applying an Embase AI response.