Linkage 2/17: Metaknowledge, iResidents, and Baldness

What the science of science looks like. (From Evans & Foster, Science, 2011)

What the science of science looks like. (From Evans & Foster, Science, 2011)

Perhaps the biggest science story of the week took place, oddly enough, on a game show. The victory of an IBM supercomputer named Watson over human contestants on Jeopardy burned up the Internet, launching a million jokes about impending robot enslavement of humans and comparisons to 2001’s HAL. Now attention is starting to turn to how the best question-answering computer yet invented can next be applied to targets more meaningful than trivia, including helping doctors make medical diagnoses. But the computational methods behind Watson – essentially a giant word-association machine – might also help the world of science take a hard look at its own biases and flaws, according to an editorial in Science last week by two University of Chicago sociologists.

The key word is “metaknowledge,” James Evans and Jacob Foster write, meaning the assembly of knowledge about knowledge. Though the term is a bit on the Orwellian side – “Metaknowledge results from the critical scrutiny of what is known, how, and by whom” – it’s a name for the acquired instincts used by experienced scientists to read between the lines of scientific research articles. A newcomer to the field may only see the methods and results written on the page, but an experienced reader perceives additional information: the reputation of the author, the institution, and the journal, the history of the subject, and the biases and assumptions inherent to any scientific study.

Evans and Foster propose that the shift toward electronic publication and the growing ability of computers to find meaning in massive amounts of data could enable a formal study of this unwritten metaknowledge – and potentially make science more accurate and efficient. A machine trained to detect patterns in scientific literature could help sniff out a multitude of known issues that distort or impede scientific results. Many of the phenomena listed by Evans and Foster have colorful names, such as:

  • the “file-drawer problem” – the tendency for experiments with negative results to go unpublished, biasing the literature toward the experiments that showed an effect
  • the “Proteus phenomenon” – when scientists flock to a high-profile finding to gain attention by extending or debunking the original research.
  • “ghost theories” – when unspoken assumptions of a field (i.e. the use of undergraduates in most psychology studies) influence the results.

Instead of slowly learning these house rules the hard way through the frustrating and slow process of accumulating scientific wisdom, a metaknowledge machine might make the implicit aspects of science explicit. That could help a graduate student avoid wasting time on experiments that have already been done, or help the government route funding to scientific areas that are truly promising, instead of just popular.

“Metaknowledge could inform individual strategies about research investment, pointing out overgrazed fields where herding leads to diminishing returns as well as lush ranges where premature certainty has halted promising investigation,” Evans and Foster write.

Elsewhere…

The crossover of technology into science and medicine doesn’t have to happen at the level of supercomputers – consumer electronics are also making an impact in the hospital wards. Since last fall, medical residents at the University of Chicago Medical Center have been using iPads on their daily rounds to check test results, view X-Rays and MRIs, and order medications for patients at the bedside. Nesita Kwan from NBC News came out a couple weeks ago to report on how these devices are making medical care more efficient, and how Bill Gates himself responded to one resident’s e-mail.

Isaac Asimov famously said that “The most exciting phrase to hear in science, the one that heralds new discoveries, is not ‘Eureka!,’ but ‘That’s funny…” Many famous scientific achievements are accidental, serendipitous findings that weren’t quite predicted by the researcher’s hypothesis. A good example popped up this week when UCLA scientists studying stress and digestion stumbled upon a treatment that regrew hair. Curing baldness, of course, is one of the great scientific pursuits of our time (tongue firmly in cheek), so the accidental result drew tons of media attention, as aggregated by 80 Beats. But don’t throw away the Rogaine yet; scientists still need to confirm that the hair growth reactivation is translatable from mice to men. (Another cool story profiled by 80 Beats this week: bear hibernation and body temperature).

Melvin Griem, a pioneer of radiation oncology research and retired University of Chicago professor, passed away earlier this month. By studying ways of using radiation to treat cancer, including neutron therapy and the implantation of radioactive seeds, Griem helped distinguish radiation oncology as a separate field from traditional diagnostic radiology.

ScienceLife tweets, and doctors tweet too. A letter published in JAMA last week looked at whether physicians on twitter met professional standards on issues of patient privacy and use of profanity – most, but not all, did. NPR’s Scott Hensley blogged about his mixed reaction to the study, a post found via Vineet Arora, one of the social-media-friendly physicians included in the study!

About Rob Mitchum (526 Articles)
Rob Mitchum is communications manager at the Computation Institute, a joint initiative between The University of Chicago and Argonne National Laboratory.
%d bloggers like this: