Speed in toxic chemical identification is critical for saving lives during chemical incidents. Unfortunately, current systems designed to access toxic chemical databases (e.g., WISER, the NIOSH Pocket Guide, and Haz- Map) use very basic search algorithms, which require too many user inputs to identify a chemical. This R21 proposal is a feasibility study to design, implement, and evaluate algorithms and interfaces to significantly expedite the identification of toxic chemical from current databases, resulting in faster responses during toxic chemical emergencies. The faster response time should help to reduce injury and death that continue to plague the population of first-responders and chemical plant workers in their line of duty. The above goals require a systematic understanding of the structure underlying current toxic chemical databases, the needs of first responders, and the design of algorithms and interfaces that leverage the database structure, and address first responder needs. We therefore propose a three step feasibility study: 1. Quantify the complex relationship between chemicals and their symptoms and properties in three large public health databases (WISER, the NIOSH Pocket Guide, and Haz-Map). This will be achieved by using novel graphical network analyses to reveal and quantify regularities such as how symptoms co-occur across chemicals. 2. Identify the needs of first-responders through interviews and surveys. The results will enable us to design algorithms and interfaces that ensure practical value to end users. 3. Develop, and evaluate algorithms and interfaces that rapidly identify toxic chemicals based on symptoms. This will be achieved by developing robust algorithms that exploit structures in toxic chemical databases, and practical interfaces grounded in the needs of end users. The prototype system will be iteratively evaluated and enhanced through user studies and made available on the Web and on mobile devices. The results of the above feasibility study to develop algorithms and interfaces will be followed by a larger study through an RO1, which will address the inaccuracies and redundancies in current toxic chemical databases, in addition to a randomized speed/accuracy study that compares the performance of the new databases and algorithms to traditional methods. The arc of this research starting from this R21, and proceeding to an RO1 requires an interdisciplinary approach that combines human-computer interaction, data mining, and the domain knowledge of first responders and toxicologists. This proposal brings together just such an interdisciplinary team, who offer a unique opportunity to develop powerful and general methods that improve the speed and accuracy of toxic chemical identification during emergencies, resulting in an impact on the lives of millions of workers exposed to toxic chemicals each year.

Public Health Relevance

Although rapid identification of toxic chemicals during chemical emergencies is critical for reducing injury and saving lives, current systems for searching toxic chemical databases use only very basic search algorithms and interfaces that require too many user inputs. This proposal is a feasibility study to design, implement, and evaluate novel algorithms and interfaces to help first responders significantly expedite how they search current toxic chemical databases during emergencies. This increased speed in toxic chemical identification has the potential of reducing injury and saving lives in the population of more than two million first-responders and chemical workers who have relatively high incidences of toxic chemical exposures.

National Institute of Health (NIH)
National Institute for Occupational Safety and Health (NIOSH)
Exploratory/Developmental Grants (R21)
Project #
Application #
Study Section
Safety and Occupational Health Study Section (SOH)
Program Officer
Inserra, Steve
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of Texas Medical Br Galveston
United States
Zip Code
Bhavnani, Suresh K; Dang, Bryant; Bellala, Gowtham et al. (2015) Unlocking proteomic heterogeneity in complex diseases through visual analytics. Proteomics 15:1405-18
Bellala, Gowtham; Stanley, Jason; Bhavnani, Suresh K et al. (2013) A rank-based approach to active diagnosis. IEEE Trans Pattern Anal Mach Intell 35:2078-90
Bhavnani, Suresh K; Bellala, Gowtham; Victor, Sundar et al. (2012) The role of complementary bipartite visual analytical representations in the analysis of SNPs: a case study in ancestral informative markers. J Am Med Inform Assoc 19:e5-e12