Exact Sciences OCR Hack

828 Registered Allowed team size: 1 - 4
828 Registered Allowed team size: 1 - 4

Winners are announced.

Ideation Phase
Online
starts on:
May 08, 2020, 10:00 AM ()
ends on:
Jun 01, 2020, 09:55 AM ()
Prototype Phase
Online
starts on:
Jun 08, 2020, 04:00 AM ()
ends on:
Jun 30, 2020, 03:55 AM ()

Winners

Overview

ANNOUNCEMENT:

The tentative dates for result announcement are between Aug 9th - Aug 11th


Exact Sciences Corp. is committed to playing a role in the eradication of cancer. As part of this mission, we offer screening and diagnostics tests for detection of cancer at various stages.

The US Healthcare IT systems consist of a variety of heterogeneous systems but a large percentage of them that do not support simple interfaces that allow for electronic order submissions. This environment results in Exact Sciences needing to manually process a large number of typed and handwritten order forms submitted by fax.

Themes

We would like to challenge developers to explore and analyze the current state-of-art, commercially-available, off-the-shelf Optical Character Recognition software system COTS OCR.

The challenge being offered to the community is to identify available products and demonstrate their effectiveness in automating the extraction of patient, provider, and payer data from the structured fields of a standard fax order form. Teams are encouraged to augment the COTS OCR solutions for maximum processing efficiency.

Optical Character Recognition

Hospital Management systems have evolved a lot through the years:

  • You need to fill a form that registers you as a new patient or a returning patient.
  • Once your handwritten details are updated into the system, you receive a token ID that queues you up to see the doctor.

While the creation and maintenance of a user profile is handled efficiently—thanks to automation—there is still a dependency on the staff for updating all the handwritten entries of each user into the system manually. Needless to say, this method is highly prone to human error.

Your task is to build a highly accurate OCR solution that takes a manually filled form as input and updates the corresponding entries into a digital form. You can build your solution using existing frameworks. Your solution must be ready-to-use and meet industrial coding standards.

Resource references:

You can use the blank form as an example of the best case scenario and the sample form as an example of the worst case scenario that you can run into while scanning the forms.

Prizes USD 5000 in prizes

Exact Sciences reserves the right to withhold the prize money from any team that does not provide the deliverables required to complete submissions and enable the judging of solutions.

Main Prizes
First prize

USD 2500

Second prize

USD 1500

Third prize

USD 1000

Social Share

Help & Support

Please contact event admin
dfuhrman at dfuhrman@exactsciences.com
Notifications
View All Notifications

?