Focus Challenge: Cognitive Computing

Don't wait, apply now!

ITServices_logo_plain_email-2eyx5ev

UChicago and Chicago Medicine students, faculty, and staff: If you could train a computer to answer questions or develop solutions, what questions would you ask or problems would you pose?

To enter, draw upon your research or interests (something you’re passionate about), and come up with a problem or scenario where finding a solution would have a great impact on an industry/society/etc.

You should consider entering this competition if — based on your current research, interests, or expertise — you can:

  1. Describe a scenario in which someone can ask a question or series of questions; or provide information that can be construed as question.
  2. Explain how people currently get answers to these questions (i.e. from a human expert or a dataset).
  3. And the answer to these questions is human-intelligible (seeing, hearing, talking, tasting, feeling) so that a person can say “that’s right!” or “that’s wrong!”

No programming or computer science experience necessary.  UChicago and Chicago Medicine students, faculty, and staff are eligible — with a few exceptions.  If you’re passionate about something in your research or day-to-day life, consider entering here.

 

Focus Challenge: Cognitive Computing

HOW EASY IS IT TO ENTER?

It's as simple as filling out a series of human-understandable questions below. Nothing else is required, except your good idea. Your first submission might take all of half an hour, if you're already well versed in the problem and solution! You can submit as many ideas as you'd like.

PART ONE OF TWO: ANSWER QUESTIONS ABOUT YOUR IDEA

For all open-answer questions, your answers should be between (200-300 words). Include references when they’ll help, especially when providing examples of data.
For more detailed information and an example, look here.
For more detailed information and an example, look here.
For more detailed information and an example, look here.
For more detailed information and an example, look here.
For more detailed information and an example, look here.
For more detailed information and an example, look here.
For more detailed information and an example, look here.
Consider a system that could recognize text, maybe speech or pictures, generate a set of possible answers from available data, provide a reasonable response, and learn from user feedback. For more detailed information and an example, look here.
Focus on the cognitive behaviors associated with your application e.g., is someone asking a question (or some computer application providing information that can be construed as a question) and getting an answer or follow-up dialogue in response. Perhaps your application finds hidden or interesting relationships and patterns in a body of information. For more detailed information and an example, look here.
For more detailed information and an example, look here.

Optional

For more detailed information and an example, look here.
For more detailed information and an example, look here.
For more detailed information and an example, look here.

PART TWO: DETAILS ABOUT YOU

Sending

Submission form: Frequently asked questions

1. What problem are you trying to solve OR the subject area you’re interested in exploring? What kind of questions are getting asked? *

In health care, for example, determining an optimal course of treatment that takes into consideration the vast amount of information that appears in research journals, patient records, and various other data sources is an intractable task for practitioners e.g., doctors and nurses, to undertake manually. It is also problematic to automate with a conventional, deterministic, fixed-rule approach as the bulk of information may be unstructured text, ambiguous, and subject to interpretation. As new information becomes available whether from new research or feedback from practitioners, conventional approaches tend not to keep up.

2. How do people currently approach the problem? *

Practitioners rely on experiencial knowledge, summary reports, heuristics, and human centric processes designed to fill knowledge gaps.

2a. Who are the experts that weigh in to answer questions or uncover relevant information from available information? * 

Continuing with the health care example, the experts might include researchers, doctors, nurses, hospital administrators, compliance officers, etc.

2b. What computer systems/tools, if any, are currently being used to help address the problem? *

Electronic medical record systems, scientific/medical journal survey services,…

2c. What data underpins the expertise or system/tool capabilities? *

Our healthcare example might include refereed medical journals, patient record notes, clinical test data, monitoring device data, etc. used to provide responses to input, answers to questions, or supporting evidence. Not all sources of information need be identified up front.

Such information is characterized by vocabularies, ontologies, and other sources of information that domain experts e.g., doctors, nurses, researchers in the healthcare example, use to learn terminology and concepts. Examples of domains are “health care”, “finance”, and “musicology”.

2d. Is the data structured e.g., tabular data, unstructured e.g., free form text, or both? *

Healthcare data occurs in all the aforementioned formats.

2e. What makes the problem especially difficult to address? *

The shear volume of free form text is a major factor here.

3. How might this problem area be improved by a cognitive approach? *

Consider a system that could recognize text, maybe speech or pictures, generate a set of possible answers from available data, provide a reasonable response, and learn from user feedback.

4. Describe one or more real-life scenarios for your application. Be expansive.

In our health care example, Doctors and nurses will interact directly with the application using a tablet or laptop device to submit a question, receive treatment options, be able to review the supporting evidence sources for a diagnoses and the top three treatment options, and then be able to provide more information that is specific to a patient in order to receive a potentially different diagnosis or treatment advice.

5. Who will benefit from a solution to the problem? *

An application that can address these issues would benefit not only the practitioners who would interact with application, but more importantly the patients who receive improved care.


Optional, for now…

6a. List up to three sources of data (be as specific as possible) from which answers and insights can be obtained: *

6b. Identify a pool of users who would be the direct users of your application. In a Q&A scenario these are the people asking the questions: *

 

6c. List the name(s) of any experts that could help validate the correctness of your application’s responses i.e., vouch that an answer is right or wrong: *