Neurological disorders and disease have been shown to impact the speech production of affected individuals. Speech signal analysis can provide clinical information that can be used to predict the onset of certain diseases and their progression, together with the effectiveness of treatment procedures. While speech analysis has a tremendous potential as the foundation for a new generation of diagnostic tools, the development and deployment of such tools has been hindered by two closely tied problems: (1) the lack of an in-depth understanding of the relationship between neurological disorders and speech production and (2) the small and incomplete sets of speech samples (and the lack of medical context) prior studies are based on. As a consequence, the exact links between neurological conditions and speech are not understood well enough to allow us to design accurate diagnostic tools yet. Further, while pervasive and mobile technologies have made it easy to collect significant amounts of data that can be used for diagnosis, the collected data are often insufficient to perform appropriate assessment without access to historical data and careful correlation and analysis of data over extended periods of time. For example, evaluations of potential minor traumatic brain injuries (mTBI), like concussions, rely on baseline data that were collected at an earlier time (e.g., during a pre-season physical exam of an athlete). Access to historical information is required and the more data available, the more accurate the assessments can be. Access to sensor data collected from a large number of subjects, whether sick or healthy, may allow us to extract new information or to establish new types of baselines. Hence there is a need for a fully automated assessment system, that can collect, organize, process, and analyze speech from individuals with concussions and provide meaningful insights about the effects on speech production. In this work, a portable speech collection and analysis system is designed and implemented to extract acoustic metrics from speech tests developed to study various aspects of speech production affected by concussions. Data is collected and analyzed from high school and collegiate athletes for the Fall 2014 athletic season for this purpose. Statistical analysis and machine learning techniques are used to predict the most significant metrics that could be indicative for the particular case of mTBIs. The system is extensible to incorporate test designs that could be used to study the impact of other neurological conditions and diseases on speech.