Quaesia

A Bibliographic Platform for Biomedical Literature Enriched by AI.

Unlike classical search engines, which primarily index keywords in abstracts, we've developed AI models to detect specific biomedical concepts and determine their potential interactions.
Thanks to these advanced models, we have constructed a massive network of interaction between biomedical concepts, comprising more than 500 million links. This platform is the direct result of our efforts to provide you with an intuitive interface to query this immense network.

Why is Quaesia More Powerful?

Enhanced Query Building

Our search engine is superior to conventional tools because it allows for more precise and comprehensive queries based on established biomedical concepts.

MeSH Thesaurus Integration: You define your workspace using biomedical concepts rather than just keywords. By integrating the MeSH thesaurus (Medical Subject Headings), the system automatically accounts for the various names and acronyms of a biolmedical entity.
Infinite Precision: You can combine MeSH terms in countless ways to create new, highly precise composite entities.
Weighted Searches: Use weighting factors to specify the importance of different entities within your query.
Keyword Fallback: If you can't find a MeSH concept that perfectly matches your idea, you can still use keywords, just like in a regular search engine.

Advanced Ranking and Results

Once you define your workspace, our search engine ranks the 40 million abstracts based on the biological entities and keywords you specified.
Evidence-Based Context: You can simply click on a link within your interaction network to retrieve the most relevant article sentences that explain the relationship between two entities, all within the context of your specific query.
Experimental Filters: We've added modules to filter sentences based on 39 different types of relationships (e.g., increase, advantage, molecular interaction, long-range interaction, etc.).

AI-Driven Prediction Module

Leveraging our enormous relationship graph, we trained an AI model to predict potential unknown links.

Custom Prediction: You can select from a list of predefined targets or build your own list from the MeSH tree to find which MeSH targets are most relevant to your query.
Evidence Included: Each prediction includes a link to the original supporting sentences as evidence available in our database, which potentially explain the predicted relationship.

AI-Augmented Abstract & Author Information

Abstract Highlighting: Thanks to manual annotations, we can significantly accelerate your reading of an abstract by clearly highlighting the entities relevant to your specific query.
Entity Discovery: Other identified entities are underlined. A simple click allows you to add new entities to your query, further refining your focus.
Relationship Visualization: We also display a graph of the relationships between biomedical concept, enabling you to access relevant supporting sentences with a single click.
Author Profile: The dedicated author page features three insightful histograms: one detailing the author's publications, one showing publications that cite the author, and one displaying all references utilized. Additionally, a table provides a complete list of MeSH terms used by the author.

Your Personalized Data

Study Preservation: Once you connect using a LinkedIn or Patreon account, you can save your studies and lists of articles.
Annotation History: You can also review all the annotations you have personally added to help correct and improve our automated AI systems.

Help Improve Our Models

We've developed a set of tools that allow you to help us correct errors made by the AI systems, enabling us to build even better models.

Support and Sustainability

The database currently contains approximately 40 million abstracts of publications referenced on PubMed and will integrate 1.5 million new ones each year.

Cost of Curation: It is time-consuming and expensive to automatically annotate all these texts using our AI models. Despite extensive optimizations, the initial calculation took several months on a GPU-equipped server, and it will take more than three months of annual calculation just to keep the database up to date.
Funding: To finance this platform, we created a Patreon page: www.patreon.com/Quaesia.
Your Impact: With your generous donations, we can continue to maintain this platform and develop new, cutting-edge tools for the entire scientific community.

My Workspace

C	N	Entity	Weight	R

Help sustain our platform by subscribing to our Patreon page

My Interaction Map

To get started, define your workspace in the sidebar

AI-Discovered Relationships

Clic on a link in your Study Interaction Map to show relathionship evidence

Score	PMID	Sentence	Edit

To limit resource costs, the table is restricted to the first 1,000 publications most relevant to your workspace. The more entities it contains, the better the algorithm ranks the sentences. You can add new entities, even with a low weighting factor.

The system for relationship extraction is currently experimental. Since we lack sufficient manual annotations to train more efficient AI models, we invite your help. Please contribute annotations by clicking a button at the end of each line.

AI-Predicted Interactions

To get started, add entities to your workspace and click on

Click a publication link or enter a PMID in the field above

Sentence

All Occurrences of entity pairs

Click an entity pair to edit or add an annotation

1. Valid Entity Detections

Text	MeSH

Are the entity detections correct (precise position and MeSH terms)?

3. Select Interaction Type

Benefite
Scale
Flow
Evolution
Mesurable
Type
Action

4. Add Custom Interaction Terms

5. Rate the Relationship

Is this relationship is important in this sentence ?

1. Define Entity Concept

Specify the MeSH terms that constitute the target entity:

Correct existing annotation by click an entity below
Copy existing MeSHs of an entity using a button
Edit list of MeSH using button

Selected MeSHs

Automatic Entity Detection

2. Select Occurrences

Mark all words in the text that correspond to the defined entity:

Click an underlined text to select an existing occurrence
Click a boxed text to deselect an existing occurrence
To add a new occurrence, use your mouse to select the text precisely, excluding any leading/trailing spaces or punctuation

3. Identify Pronouns & Acronyms

Select any occurrences that match acronyms or pronouns referring to the entity:

You must fill your workspace before saving

This name is already in use

Rename ""?

Confirm deletion of ""

This name is already in use

This publication is already included in the list

Rename ""?

Confirm deletion of ""

Num	PMID	Title	Delete

Date	Annotation	Edit	Cancel

Confirm deletion of ""

To define entities we relied on the “MeSH thesaurus”. Create high-precision concepts through MeSH composition. For example, combining 'Breast Neoplasms' and 'Therapeutics' allows our ranking model to target “breast cancer treatment” specifically. Use the search bar or browse the MeSH tree to start your composition. When you find an interesting term, simply click 'Add MeSH' to include it in your current selection.

Already in our target list

MeSH Ontological Relationships

Hue:

Saturation:

Lightness:

Quaesia Database Statistics

The Quaesia database is available for download in JSON format. This database includes

#Entry	Type
39,536,075	PMIDs
286,629,063	Sentences
864,041,973	Entity Detections
676,636,631	Relation Extraction
11,136,741	Lemmatized words
22,754,241	Authors
355,035	MeSH terms
72,514,681	MeSH pairs in relation extraction
508,106,655	MeSH pairs in relation extraction (typed by 39 interaction terms)
1,986,710	MeSH mappings to external database ID

Database Architecture and Tools

This database was built using a MongoDB system. A full description of the architecture and data model is available here: Database Description.
In addition to the base data, we offer the following resources for download:

Python Utilities: A Python script to load the database into a MongoDB system, along with a set of example queries and model scripts.
Information Retrieval Model: A pre-trained model designed to rank PMIDs based on a query that includes MeSH terms and lemmatized words.
Prediction Interaction Model:
A pre-trained ProNE model available for predicting affinity between entities.

FTP Server Connection Details

Host
Port
User
Password

My Licences

Invoice	Date	Amount	Open

Licensing and Pricing Plans

To cover the extensive costs associated with generating and maintaining the database, models, and scripts, a license fee is required. All licenses include unlimited downloads for one year of the full database, models, and scripts.

Organization Type	Employee Size	Annual Fee (EUR)
Nonprofit	All sizes	€200
Company or for-profit structure	<10 employees	€300
	10 to 100 employees	€600
	100 to 250 employees	€1,000
	>250 employees	€2,500

Payment of the license fee grants you the right to download and use the entire database, including all updates, for a period of one year. Please read the full Licence carefully to understand all rights and prohibitions. In summary:

Permitted Uses (Rights)

You may use the database to support your activities, including training AI models, building search engines, or generating therapeutic insights.
Commercial use is permitted, except under the Nonprofit license.

Prohibited Uses (Restrictions)

You may not resell the database, either in its original form or modified.
You must respect the individual licenses of any external databases included within this dataset.

Required Attribution (Obligation)

You must cite the Quaesia website for any public use or dissemination of the database or its derived data.

To access the download form, log in using your LinkedIn or Patreon account. We utilize these platforms as trusted third-party providers via OpenID Connect to quickly and securely verify your identity.
By signing up, you agree to our End-User License Agreement and Privacy Policy

Database Download and Licensing Steps

To complete your license and access the database download, please follow the steps below:

Complete the Registration Form: Provide the necessary information required for the license agreement.
Access Payment: You will be redirected to the payment module, processed securely via Stripe.
Confirmation and Access: Once payment is validated, you will receive an email notification that your account on Quaesia has been updated.

Key Account Deliverables

Within your account, you will find:

Your personalized License Agreement.
The unique connection details for the FTP download server.

License Restriction

Type of structure

Name

Address

Country

E-mail

E-mail (Repeat)

Company ID (DUNS, CRN, NEQ, HRB, SIRET…)

Number of employees

Please provide the following details to help us investigate the issue:

The Issue: Clearly describe the problem you encountered.
Steps to Reproduce: Detail the exact actions you took before the issue occurred.
Expected Result: Describe what you anticipated would happen.
Actual Result: Describe the specific error or results you observed.

You can support our ongoing work and development by subscribing to a membership on our Patreon page. https://www.patreon.com/Quaesia

Membership Benefits

A Premium subscription provides access to additional features, including:

Unlimited Study Size: Remove limits on your research parameters.
Advanced Querying: Access typed relationship queries.
Predictive Analysis: Utilize relationship prediction models.

If you are already a member, please log in to access your benefits.

We use Patreon as trusted third parties, utilizing the OpenID Connect protocol to securely verify your identity. By logging in, you gain the ability to:

Save your customized studies and article lists.
Correct automatic annotations, which directly helps us build better AI mdels.

By signing up, you agree to our End-User License Agreement and Privacy Policy.

Do not ask again

You have reached the maximum number of entities allowed for your current Quaesia Patreon subscription level. To continue adding entities, you can:

Upgrade Your Plan: Increase your entity limit by subscribing to a membership on our Patreon page. https://www.patreon.com/Quaesia
Use Free Access: Utilize this feature during the Sunday free-access window to continue your work without a subscription limit.

Membership Benefits

A Premium subscription provides access to additional features, including:

Unlimited Study Size: Remove limits on your research parameters.
Advanced Querying: Access typed relationship queries.
Predictive Analysis: Utilize relationship prediction models.

If you are already a member, please log in to access your benefits.

We use Patreon as trusted third parties, utilizing the OpenID Connect protocol to securely verify your identity. By logging in, you gain the ability to:

Save your customized studies and article lists.
Correct automatic annotations, which directly helps us build better AI mdels.

By signing up, you agree to our End-User License Agreement and Privacy Policy.

This feature is available exclusively with a Quaesia Patreon Premium subscription. To enable this feature, you can:

Upgrade Your Plan: Increase your workspace limit by subscribing to a membership on our Patreon page. https://www.patreon.com/Quaesia
Use Free Access: Utilize this feature during the Sunday free-access window to continue your work without a subscription limit.

Membership Benefits

A Premium subscription provides access to additional features, including:

Unlimited Study Size: Remove limits on your research parameters.
Advanced Querying: Access typed relationship queries.
Predictive Analysis: Utilize relationship prediction models.

If you are already a member, please log in to access your benefits.

We use Patreon as trusted third parties, utilizing the OpenID Connect protocol to securely verify your identity. By logging in, you gain the ability to:

Save your customized studies and article lists.
Correct automatic annotations, which directly helps us build better AI mdels.

By signing up, you agree to our End-User License Agreement and Privacy Policy.

Experience Quaesia Premium for free today!

A Premium subscription provides access to additional features, including:

Unlimited Study Size: Remove limits on your research parameters.
Advanced Querying: Access typed relationship queries.
Predictive Analysis: Utilize relationship prediction models.

Existing members: Please log in for uninterrupted access to your premium features.

We use Patreon as trusted third parties, utilizing the OpenID Connect protocol to securely verify your identity. By logging in, you gain the ability to:

Save your customized studies and article lists.
Correct automatic annotations, which directly helps us build better AI mdels.

By signing up, you agree to our End-User License Agreement and Privacy Policy.

A Bibliographic Platform for Biomedical Literature Enriched by AI.

Why is Quaesia More Powerful?

Enhanced Query Building

Advanced Ranking and Results

AI-Driven Prediction Module

AI-Augmented Abstract & Author Information

Your Personalized Data

Help Improve Our Models

Support and Sustainability

Tutorials

Select a tutorial link from the list

My Workspace

My Interaction Map

To get started, define your workspace in the sidebar

Loading, please wait...

AI-Discovered Relationships

Your workspace have changed. As a result, the ranking below may no longer be up to date

Clic on a link in your Study Interaction Map to show relathionship evidence

Loading, please wait...

Interaction type:

AI-Predicted Interactions

Your workspace have changed. As a result, the ranking below may no longer be up to date

Available exclusively with a Quaesia Patreon Premium subscription

This feature is free today, Build your workspace and then select a MeSH target buttons.

Select a MeSH target buttons to launch prediction

To get started, add entities to your workspace and click on Find publications

Publications Per Year

Text Access

Your workspace have changed. As a result, the ranking below may no longer be up to date

No results found, change filter or retry

Click a publication link or enter a PMID in the field above

Unknown PMID

Click an author's link in the "AI-augmented abstract" tab

MeSH Usage

Text Analysis

My search Topic

Loading, please wait...

Select a Sentence to Add a Relationship Evidence

Edit Relationship Evidence

Sentence

All Occurrences of entity pairs

1. Valid Entity Detections

2. Valid Relationship

Click Submit to remove the relationship evidence for this entity pair in the text

3. Select Interaction Type

4. Add Custom Interaction Terms

5. Rate the Relationship

Edit Entity Detection

1. Define Entity Concept

Selected MeSHs

Automatic Entity Detection

2. Select Occurrences

Sentence Scope: Annotations must be confined to a single sentence

Starting Point: Begin by defining the entity you wish to annotate in the upper-left panel

3. Identify Pronouns & Acronyms

My Data

You must fill your workspace before saving

This name is already in use

Rename ""?

Confirm deletion of ""

This name is already in use

This publication is already included in the list

Rename ""?

Confirm deletion of ""

Confirm deletion of ""

Search MeSH Terms

MeSH Ontological Relationships

Select Color

The color may be too bland. Recommendation: Choose a saturation greater than 30%

The color may be too dark. Recommendation: Choose a lightness between 80% and 95%

The color may be too light. Recommendation: Choose a lightness between 80% and 95%

Download Database

Quaesia Database Statistics

Database Architecture and Tools

FTP Server Connection Details

My Licences

Licensing and Pricing Plans

Permitted Uses (Rights)

Prohibited Uses (Restrictions)

Required Attribution (Obligation)

To get started, add entities to your workspace and click on