Discoveries change the world. We accelerate discovery.

ConceptualEyes has its roots at Oak Ridge National Laboratory, where in collaboration with the National Library of Medicine, tools and infrastructure were developed to address rapidly growing Big Data challenges with massively scalable graph analytics and machine learning algorithms on some of the world's most powerful supercomputers. The technology behind the ConceptualEyes platform then called ORIGAMI (Oak Ridge Graph Analytics for Medical Innovation) demonstrated its application to diagnosing mystery illnesses, rare disease discovery, hypothesis generation towards the risk of retinopathy in diabetic patients, understanding fraud in healthcare and many more use-cases and success stories. ORIGAMI was a proud recipient of the R&D 100 award  (also referred to as the Oscar's of Invention) for 2016.

Today, ConceptualEyes is built on top of the strong foundation of ORIGAMI and is being built to drive discovery and innovation in a wide variety of industries: health (e.g., drug discovery, disease target identification), finance (e.g., risk management), energy (e.g., oil exploration), and manufacturing (e.g., product design).  Significant R&D investment from the ConceptualEyes leadership has allowed ConceptualEyes to leapfrog its competition as a discovery engine with the edge in leveraging artificial intelligence algorithms for machine reading, natural language understanding, scalable performance and a superior user-experience. Deployable both on hosted servers and commodity cloud-computing services, ConceptualEyes is offered both as a software-as-a-service and platform-as-a-service solution for the individual researcher, mid-size and large enterprise with privacy and security requirements. 

Our vision is to build Knowledge Universes that ingest, integrate, interact, imagine and innovate based on data-driven discoveries at the speed of thought. 

Toward this goal, our first product "SEEKER" is focused on health and medical research and here we build on our past successes and through continued support and interest from users across medical research organizations, pharmaceutical companies, and global health organizations. Sign-up to try our product at http://planet.conceptualeyes.com.

 

The Leadership Team

Frequently Asked Questions

The Engine: What, Who, Why

What does the ConceptualEyes platform do?

ConceptualEyes applies semantics, statistics and logic to explore the what-is, investigate the what-else, examine the what-if, and hypothesize the what-could-be on domain-specific knowledge graphs. It is the first of its kind explainable artificial intelligence (AI) platform that is able to back the results with peer-reviewed evidence and provenance. Although agnostic to underlying data, the tool in its current form is hosting a knowledge graph of medical facts extracted from millions of abstracts in the medical literature since 1865 that was made available by the National Library of Medicine.

As a user trying to navigate the medical maze of literature, the tool leverages AI algorithms to filter noise from signal and get to facts, connections and concepts of value that would otherwise take several months of research and review.

How is ConceptualEyes different from PubMed, Google Scholar and other search engines?

ConceptualEyes is more than a document-retrieval platform. It is a knowledge-discovery platform with intelligence implemented for knowledge reasoning and language understanding. ConceptualEyes is an AI platform akin to a voracious reader that has learned how to write relevant sentences by connecting the appropriate terms and understanding semantic pattern structures of grammar in language. By virtue of that artificial intelligence, ConceptualEyes goes beyond retrieving documents from repositories based on occurrence of keyword, or a predefined set of rules, to allow users interactively discover relevance and context.

What is the underlying dataset? How up-to date is it? How often is the data updated?

The Knowledge Planet interface is using the publicly available PubMed abstracts. Our update cadence is currently 6 months and in a few weeks, we will be updating daily.

I am a scientist, medical researcher or health innovator. How will ConceptualEyes help me?

ConceptualEyes is designed to do what computers do best to help researchers do their best. The tool helps researchers by: (a) reducing time for literature survey from months to days (b) shortening time for hypothesis generation, design/discovery that would otherwise take years (c) revealing associations and patterns that fuel the gestalt/intuition to ask the right questions.

For instance, a principal investigator in medicine today is sifting through 25000+ article titles, 2500 abstracts and 250 papers, and on average takes 8 months to design a hypothesis and write a grant proposal for funding. Preliminary case-studies are showing that our tool significantly increases productivity for both experiment design and grant writing.

 

Engaging with us

What are some examples of successful use of ConceptualEyes?

Several organizations have used the tool to design workflows for mystery-illness diagnosis, rare-disease discovery, association of diabetic retinopathy and beta-blocker treatment of hypertension and risk. The emerging application of this tool is in literature summarization in specific areas of research – such as schizophrenia and autism, protein storyboarding for muscular dystrophy and discovery of nutraceuticals.

Who do I contact for help with the tool?

A.      Troubleshooting / Glitches / Bugs

We can be contacted in many ways. For interface related feedback, one can submit bugs and glitches using the feedback button. We are also constantly monitoring emails at contact@conceptualeyes.com.

 

B. Services work such as doing a case-study, delivering a report, designing specific workflows, access to the underlying APIs

You can also email us at info@conceptualeyes.com. We proud ourselves in responding by the next business day.

 

Can I host the tool on premise? On my own infrastructure? How?

Yes, you can host the tool on premise. We provide the interface as a software-service offering, but if you have infrastructure that can host the tool and the associated data sets, we are happy to set-up and configure the environment within your organization’s firewalls. Please email us at contact@conceptualeyes.net

Do I have ownership of the results I produce?

 

Yes. You will retain ownership of your data, any algorithms and software that you may have used, and all outputs.

 

Search Interface

I see four tabs on the interface – Explore, Connect, Conceptualize, Hypothesize. What do they do? Do I have to use them in a specific order?

The Explore, Connect, Conceptualize and Hypothesize are four cognitive functions implemented in our tool to help you navigate the medical literature. A researcher can use them in any order and can even submit multiple search queries simultaneously. The functionality offered with each of the tabs is described below.

Explore: Search functionality in this tab allows you to type and choose a medical term of interest, to retrieve knowledge-facts associated with the term. Explore allows you to learn about the “what-is” around a term or concept. The explore tab presents results from navigating the literature to provide a list of terms that (i) co-occur (mentioned or referenced together often), (ii) share similar facts and properties and (iii) define context around the search term. 

Example:  If you enter the term “Nexium” and press the “Search” button, facts such as “Nexium interacts with acids”, “Nexium treats Gastroesophageal reflux disease”, etc. will be retrieved along with the list of associated terms (such as Omeprazole, Proton-pump inhibitors), similar terms (such as Esomeprazole, Famotidine, Ranitidine, Antacids, etc.) and context terms (such as Pharmaceutical Preparations, Therapeutic procedure, etc.)

 Connect: This search tab allows a researcher to quickly evaluate a “what-if” intuition that two terms may be related in a meaningful way. The artificial intelligence algorithm automatically fills-in-the-blanks to connect the two terms with a sense of semantics and logic. This capability can be used to validate known associations such as Nexium and Heartburn or used to validate hunches such as the potential of treating Ebola virus with Chloroquine (a malarial drug).

 Example: For Nexium and Heartburn – ConceptualEyes will present an association as follows: “Nexium is an Esomeprezole. Esomeprezoles affect heartburn.”  With Ebola virus and Chloroquine, the result could look like “Ebola virus location_of Epitopes (B lymphocyte). Immunotoxins disrupt Epitopes. Chloroquine negatively_stimulates immunotoxins.

Conceptualize: The Conceptualize tab allows a researcher to quickly organize and investigate the “what-else” around a search term of interest.  The results summarize the research activity around the term of interest at a meta-level to augment an individual’s expertise from an adjacent domain

Example: In the Ebola virus example, the results would show that Biologic, Genetic, Pathological, Physiological functions are processes associated with Ebola virus and Ebola virus is also part of a gene or genome, immunologic factor, and a nucleotide sequence.

Hypothesize: The Hypothesize tab is the gateway to ask the “what-could-be” questions. The search tab takes multiple terms of interest as input and does a simulation to predict potential non-obvious associations.

 Example: A user can enter terms such as Distal muscle weakness, Gait, Child, Woman and then choose one among many targets (such as Disease, Gene or Drug) that the user is interested in finding a non-obvious connection. If the Disease target is chosen, the output will be a list of potential diseases diagnoses.

Why do I always have to choose from the list? Some medical terms are not being accepted as valid inputs. Why?

ConceptualEyes in its current form indexes the MeSH vocabulary.  The MeSH browser – https://meshb.nlm.nih.gov allows you to map terms to what will work in the ConceptualEyes interface. We are aware of the issue that the search box is not accepting terms with special characters. We are currently working on fixing this issue and it will be resolved soon.

I see the spinning wheel icon on some searches. What does that mean?

Please check the spelling and completeness of the search term. The spinning wheel icon typically shows up when the search term was not picked from one the list of MeSH terms. Certain times, the spinning wheel is an indicator that our server is still finishing up a previous search. If the spinning wheel persists for longer than a minute, try refreshing the page and searching again.

If the spinning wheel issue still persists, please contact us at contact@conceptualeyes.net  

How can I better manage my search and analysis?

A.      How to filter results?

ConceptualEyes offers 3 kinds of filters. By default, the filters are turned off. The first filter box appears right above the result sets. This filter box helps you sift for results that contain a specific item of interest in addition to the search term. There is an advanced filter button around the filter box that helps select a few from hundreds of possible predictions. The advanced filter also helps create a stop-list of terms that are out of context for the search. The third filter applies only to the Hypothesize tab, where you can filter by the meta-type tags.

 Example:   In a search result you see iodine and hydrogen peroxide. While you want to dive deeper on the results involving iodine, you are not interested in hydrogen peroxide. You can block the term(s) (like hydrogen peroxide) from appearing in your results using the Advanced Filter.  Alternatively, you can focus on only the terms (e.g. iodine) that you select using the filter box above the table/graph visualization of results.

B.      How do I eliminate publications from my analysis? e.g., I would like to analyze medical literature for work excluding my own research. Can I choose only the publications of interest to me?

You will be able to filter out publications by authors, institutions, and year of publication at a later date in our future product offerings.

Can I see the publications between each relationship either in the table or in the graph?

Yes. You can see the publications that resulted in the relationship between two terms. You can click on a graph edge or a table row to get the publications that support the relationship you are exploring.

 

Can I analyze or drill-down on the search results?

Yes. The tool allows the flexibility and functionality to traverse every result row from the search. The drill-down window pops up when a result or the eye icon at the end of each row is clicked.

 Note that different tabs will produce different kind of drill-down results. For example, the Conceptualize tab produces meta-relationships for the term. The drill down in that tab helps you navigate to the data tagged under the meta-data. With the Connect tab, the drill down gives you the context and topics that was used to generate that result.

Some result rows look similar and some do not. Should each row be considered independent of each other?

Every row is an independent result of a close-to-exhaustive search of possible associations. Sometimes, they may look similar because of the semantic overlap (e.g. AFFECTS and CAUSES), but each row is independent of the other.

Is a graph the only output type when I use the Connect tab?

The network graph visualization of connections and associations is the default output setting of the tool. You can switch to a table view by clicking the list/table icon that appears above the graph visualization.

Some of the predicates are not intuitive. (e.g. neg_affects, Rev_associated_with etc.). Why? How do I interpret such results?

ConceptualEyes in its current form uses the MeSH vocabulary and the Unified Medical Language System (UMLS) as the ontology to interpret the predicates. Predications such as NEG_AFFECTS are defined in UMLS by the National Institute of Health and we decided to follow their nomenclature.

Here are some quick tips to interpret the predicates

NEG_ : prefix is short for negatively.

Rev_ : The occurrence of Rev_  describes the directionality of the association. While reading a result from left to right and one counters a Rev_(predicate), that segment of the association has to be read from right to left.

We're Growing!

We are fervently looking for candidates who enjoy the startup atmosphere. Find out more about current openings below.

Contact Us

9350 NE 30th Street
Clyde Hill, WA


contact@conceptualeyes.com