Artificial Intelligence (AI) Services

Machine Learning and Data Analytics

Since 2016, T&T has supported the Dept. of Defense (DoD), Defense Health Agency (DHA) by developing Natural Language Processing (NLP) algorithms and innovative Machine Learning solutions.


Specifically, we provide the DHA Vision Center of Excellence (VCE) with NLP development for the Defense and Veterans Eye Injury and Vision Registry (DVEIVR) database.


DVIEVR is a joint initiative between the DoD and Veterans Affairs (VA) to manage an ocular care clinical data repository, to allow for DoD and VA clinical providers and researchers to review and research data collected on eye care diagnosis, surgical intervention, operative procedures, visual acuity, and treatment of our service members and veterans.

Prior to T&T’s efforts on the DVEIVR project, data was pulled from records of various healthcare systems and data stores, and data abstractors would utilize structured text records, as well as unstructured text fields, to fill in patient records for the DVEIVR database. They would scour thousands of records per month looking for every instance of an eye injury or treatment, to log the data in the appropriate screen/field in DVEIVR. This approach is very timely and prone to errors, especially as new abstractors are becoming familiar with appropriate vision terminology.

In an effort to improve accuracy and to help abstractors find potential injury and treatment data faster, T&T was contracted to design NLP application program interface (API) tools. The intent of these tools was to analyze DVEIVR free text data fields using NLP over the network and produce de-identified enriched data that could be used by programmers at any point in the system. The APIs were to be developed based on a common framework acceptable between the T&T developers and the DVEIVR master system integrator.


T&T developed a Machine Learning Engine to analyze highly jargonized medical text. Using complex NLP algorithms, the engine interprets words and data in context and can then feed webforms that can be used for data analysis, trend analysis, and even auditing of records. As NLP interprets clinical data to assist abstractors, those abstractors can approve or disapprove NLP’s work when documenting medical providers’ notes into data and information for researchers. Our Machine Learning engine yields excellent results with accuracy that our team is continuously improving during this ongoing project.

T&T’s NLP Team has made considerable strides in implementing advanced technology on DVEIVR where other contractors failed. Our team of data scientists, NLP developers, and functional analysts laid the groundwork for developing and integrating NLP algorithms and machine learning web services with the DVEIVR database, which will allow for more efficient processing of medical record free text data, to improve the capacity of our clinicians and researchers to locate and compile a greater amount of relevant data to their database queries. This advanced technology not only reduces human error introduced to the database through the use of Clinical Data Abstractors, but can also be used in the future to perform automated data cleansing and data quality activities.