Tag: big data
Personal experiments
During my medical training, we were taught that if a patient responds to a treatment, it doesn’t necessarily mean that every patient will respond in the same way. The results in a single patient might be due to chance so it was important to look at the results of well-designed research studies before we could conclude that the treatment really worked. In statistics, “N” refers to the sample size in an experiment so we referred to these individual observations as “N-of-1” experiments (and we did not look at them very favorably).
Unfortunately, there are lots of problems with research studies – they take a long time to complete, the patients in the studies are very carefully selected and may be very different from you, there is often bias in the way the results are interpreted, etc.
What if doctors and patients had tools that allowed them to design high-quality experiments specifically for the individual patient?
Research in real time
This past week, I was reminded of the problem of fraud in medical research when the British courts ruled that the mother of a child with autism had lied about her child’s symptoms. The woman was a supporter of Dr. Andrew Wakefield, the British doctor who published studies connecting the Measles Mumps and Rubella – MMR – vaccine to autism based on “data” he made up.
Published medical studies, especially randomized controlled trials, remain the most reliable way for doctors to make treatment decisions. But, in addition to fraud, there are many other problems with medical research studies including the fact that they cost a lot of money to perform and take a long time to complete. An even bigger problem is that research studies have not been done for most of the questions that doctors and patients want answered. And even if there is a relevant published study, the people in the study may be very different from you.