Informatics students compete in international competition
Two Arizona State University biomedical informatics (BMI) Ph.D. students are competing in the internationally recognized Medical Natural Language Processing Challenge supported by the National Institutes of Health (NIH).
Under the direction of BMI professor Graciela Gonzalez, Ehsan Emadzadeh and Azadeh Nikfarjam are in the final stages of a 5 month-long competition in the fifth i2b2/VA challenge on sentiment classification of suicide notes.
Informatics for Integrating Biology and the Bedside (i2b2) and the Cincinnati Children’s Hospital Medical Center along with the Veterans Administration (VA) have partnered to propose the challenge task to participants. Composed of two tracks and an objective and subjective analysis, the challenge has a 2-fold rational: to encourage collaboration between the two research communities and to create an annual calendar that optimizes the research centers’ mission to create, disseminate and teach new knowledge.
Past challenges included a relations challenge (2010), medication extraction challenge (2009) and obesity challenge (2008).
This year’s task was to extract emotions in the sentiment classification of 1,000 hand-annotated suicide notes, written by people who have passed away as a result of suicide. Categories of 15 emotions included hopelessness, fear, happiness and pride. The extraction of emotions enabled the student researchers to find preventive solutions and to investigate the mental health of people prone to suicide or special groups.
Emadzadeh and Nikfarjam participated in the second track of the challenge, where they received training data of 600 notes to prepare them for the next stage of official testing in June. On Aug. 1, official test data of 300 new notes was released, with system outputs on the data due a mere two days later.
Participants were then asked to submit a short paper describing their system and analyzing their performance in September.
“A person at the risk of suicide most likely talks or writes about his or her feelings and this highlights the importance of natural language processing for automated monitoring and preventive purposes,” Emadzadeh and Nikfarjam wrote in their final paper.
The top-performing systems of particularly novel approaches will be invited to present or demo their systems at the workshop in October.
Since competition results have yet to be announced, the students’ findings are being withheld. Nevertheless, Emadzadeh and Nikfarjam’s competitive project stands as a strong example of the many exciting opportunities BMI students at ASU actively participate in.
Informatics for Integrating Biology and the Bedside (i2b2) is an NIH-funded National Center for Biomedical Computing based at Partners HealthCare System. The i2b2 Center develops a scalable informatics framework to enable clinical researchers to use existing clinical data for discovery research. Part of the United States Department of Health and Human Services, NIH is the primary agency of the government responsible for biomedical and health-related research.