Dialogue for Reverse Engineering Assessments and Methods (DREAM) Challenges pose scientific questions with great impact on human health to the crowd, collect solutions from the participants and validate the statistical significance of the top-performing solutions. This year researchers from the University of Illinois at Urbana-Champaign (UIUC) participated in two DREAM challenges and won both.
In the recent DREAM ALS Stratification Prize4Life Challenge, Professor Jian Peng from Computer Science and Graduate student Jinfeng Xiao from Biophysics & Quantitative Biology developed a new computational approach to predict the progression of Amyotrophic Lateral Sclerosis (ALS) based on patients’ clinical history. ALS is typically a rapidly progressing neurodegenerative disease. In most cases, patients can survive between within 3-5 years from onset of symptoms, but the rate of progression in the patient population can vary by an order of magnitude. Unwinding such underlying heterogeneity can hopefully shed light on disease mechanisms and drug development, and reliable prediction on progression rate can assist clinical decisions. Advised by Professor Jian Peng, Jinfeng Xiao has developed a novel algorithm to identify the most important clinical signals to predict ALS progression. The method was ranked as the best performer on predicting disease progression of patients from two national registries, namely the Irish National ALS Register and the Piemonte and Valle d’Aosta Register in Italy.
This summer, a team of graduate students, including Jinfeng Xiao, Sheng Wang, Jingbo Shang, Xiang Ren, Doris Xin and Henry Lin, advised by Professor Jian Peng and KnowEnG Project Director Professor Jiawei Han, participated the DREAM Prostate Cancer Challenge with the goal to predict the survival of patients with metastatic castrate resistant prostate cancer (mCRPC), an advanced form of the disease whose outcomes are poor and treatment remains unclear. They have designed several machine learning algorithms to predict survival times and risks from patients’ health records. Finally, the team is ranked 2nd and also recognized as a co-winner of the challenge.
Inspired by the results, the teams are currently further developing and testing their methods. On Nov 16, Xiao will present the work on ALS progression in more details at the RECOMB/ISCB Conference on Regulatory and Systems Genomics with DREAM Challenges. The prostate cancer challenge team is now working with the challenge organizers and other winners for a paper submission to Nature Biotechnology. The participants are also trying to make their methods more systematic, rigorous and widely applicable.