Tuesday, November 13, 2012

Army's Work with Language Barriers

The Army realizes that language barriers can make communicating a challenge and can hinder the delivery of effective medical care. Without the assistance of a human translator, a medic may not be able to accurately capture enough information to fully address a patient’s needs.

Researchers from RDECOM’s “Communications Electronics Research Development and Engineering Center” or CERDEC have spent the past year developing strategies and methods to use to improve machine foreign language translation software in support of military medical translation needs.

After hearing reports about translation challenges in Haiti’s 2010 earthquake, TATRC recognized that the need for translation was pressing for the medical community and that the Army could assist with the engineering research and development to help improve medical capabilities.

According to Ray Schulze, Chief, Information Management Branch for CERDEC’s Command, Power and Integration (CP&I) Directorate, the Army is specializing in language translation in the disconnected environment and is working on various mobile platforms.

Schulze reports that using the combination of automatic speech recognition which takes the spoken word and converts it into text along with foreign language translation technologies does already exist, but the accuracy of those technologies is mediocre at best when used within the medical domain.

In some locations, soldiers currently have foreign language translation technologies like the new Phraselator, initially developed by DARPA developed in partnership with CERDEC in 2001. These systems were developed for particular soldier scenarios using one-way translation and generic phrases that require the user to stick to a script. Commercial translation applications also exist, but they are made for tourists in foreign countries.

The “Medical Application of Speech Translation” or MAST is a collaborative research project between CERDEC, TATRC, and the Army’s Southern Command called SOUTHCOM. TATRC is the sponsor and program manager for MAST while CERDEC provides the engineering support and SOUTHCOM facilitates access to the operational environment and end user base.

“Medics are seeking a small portable solution to basically house translation software that runs on a mobile device to be used without an internet connection,” said Cynthia Barrigan, MAST Program Manager and Portfolio Manager for global health engagement at TATRC.

She said, “In addition, we know that while users are interested in using translation technology, they are concerned with how to integrate it into their clinical routine in the field and how a patient will react to it. They are also aware that the translation needs to be very accurate to be useful.”

The MAST project team went into the field armed with a variety of mobile devices loaded with a commercial translation software application called “Jibbigo”. The MAST team gathered useful operational data and demonstrated the technology to users.

Daniel Yaeger, a CACI contractor supporting CERDEC CP&I is a subject matter expert in the area of machine language translation. He has traveled to Honduras three times in the past six months performing technology demonstrations and collecting simulated data for the MAST project. Each trip to Honduras used different hardware, microphones, and devices to gain feedback from doctors.

 He reports that the more data you give the program, the better it becomes and that translation software becomes more accurate in much the same way a small child learns. He said,” A child learns by listening to conversations, and then they absorb the information. This is similar to statistical machine language translation because essentially, you tell the program what sound means for a specific text. When you do this enough times, the algorithms behind the machines translation software actually learns these new phrases.

CERDEC CP&I engineers followed SOUTHCOM medics on several medical readiness training exercises to observe how medics interact and communicate with patients. This included observing the process of registration, triage, and observing the types of conversations between doctors and patients. By understanding how medics interact with patients, engineers can determine the specific requirements that are needed to develop an effective machine translation system.