Date: November 06, 2003 Time: (All day)
Event Type: Lecture
Cancer research data is piling ever higher and quicker (69020 Pubmed pubs in 2002). A large and growing community of very smart people is struggling to store it, manage it, connect it, and serve it out again. Data miners mine highly suggestive correlations in huge numbers, so the number of new questions explodes too. But the throughput of clinical trials is not growing; it the rate-limiting step. This argues that we had better choose and design our new, expensive, lengthy clinical trials with extreme care.To do this, clinical trialists gather information, use it all to make mental pictures, imagine how their pictured world would behave, and draw educated guesses that guide them. Oncology Thinking Cap is software designed to make each of these steps easier and more accurate. It integrates knowledge gathering with comprehensive mathematical cancer modeling. The design goals include babying the user by removing tedium, and calculating the models accurately, while maintaining model flexibility with a conceptual box big enough for novel future biology and for paradigm-smashing thinkers.I will identify essential elements of the software architecture, demonstrate their implementation in the newest Oncology Thinking Cap, reveal the project as shocking long-term goals, and pick the audience brains for good ways to utilize natural language processing technology.