About MetaMed
Two short marketing drafts written for MetaMed Research, the personalized-medicine consultancy Alyssa co-founded in 2012 with the futurist Michael Vassar and the Skype co-founder Jaan Tallinn, with backing from venture capitalist Peter Thiel. The company offered patients custom medical research reports drawing on Bayesian evidence evaluation, in the spirit of the LessWrong rationalist community from which several of its founders came. The second piece is largely adapted from work by Zvi Mowshowitz, MetaMed’s SVP for research.
About MetaMed Which foods are good for you? It’s a question we’ve asked for centuries. It’s a question that, after countless studies, we still don’t know the answer to.
This group says one thing. That group says another. Meat causes cancer. Meat prevents cancer. Diets should be low-fat — no, low-protein — no, low-carb. Today, everyone must take this mess of confusing, contradictory advice, and try to figure healthy eating out themselves.
If a pipe breaks, you don’t fix it on your own. You call a plumber. So, in 2010, Michael Vassar asked: Why should people be their own scientists? If we don’t fix our own pipes, and we don’t build our own cars, why should every person do their own research on nutrition? Instead of expecting everyone to be their own expert, we should have a group of real scientists, trained in evaluating evidence, handling the food problem.
After pitching his idea, Michael teamed up with famous computer scientist Jaan Tallinn to start a nutrition research company. But, it soon turned out, the problem was far larger than just food. Virtually every branch of medicine suffered from the same issue. Patients, afraid of rare diseases, tried to diagnose themselves online. Doctors had no time to read the latest research, and would recommend treatments found ineffective ten or twenty years ago. There were no scientists in the system. Everyone, from movie stars to truck drivers, had to do medical science on their own — sometimes with deadly results.
We decided to bring more science to medicine. But it wouldn’t be easy. We’d need specialists in many different fields — biologists, doctors, physicists, mathematicians, programmers. We’d need new ways of looking at health research. We’d have to root out countless biases — drug research funded by drug companies, diet studies done by diet gurus hawking their own snake oil. And, of course, we’d have the problems faced by any new company.
It took some time, but we built a large, highly skilled, specialized research team. We got investment from Peter Thiel, one of the most successful businessmen in the world. And, ever since our launch to the public, we’ve been helping people improve their health with personal medical breakthroughs. At MetaMed, science isn’t just a news story. For our patients, we make science real.
Gathering evidence to support better decisions Is a drug effective? Does gluten cause cancer? How can we know? At MetaMed, we use some core principles to figure things out rationally and scientifically. What we offer our patients are Actionable Options. We state our findings, in the most honest and useful way possible, and let doctors and patients draw their own conclusions about what action to take.
We provide the best actionable options available, even if there is uncertainty surrounding them. The act of not deciding is itself a choice, and we recognize the patient must make a decision — to take the pill, to change their diet, to undergo the therapy, or not. Sometimes, things are simple and clear-cut. But when they aren’t, we still do our best — we realize that not researching something until evidence for it is “bulletproof” harms through inaction. Which actionable options are best also must include a patient’s personal circumstances; it does no good, for example, to discuss exercise regimens too strenuous for patients to practically do.
We believe that negative actionable options are as important as positive actionable options. Patients need to know what not to do, as well as what to do. This includes treatments that have been evaluated and rejected — especially those which are commonly believed in, but don’t really work.
We offer numbers. Numbers allow patients to make good decisions. At MetaMed, numbers are mandatory. We always strive to give our best estimate, given what we know. Although there is sometimes a great reluctance to state numbers — be they absolute risks, relative risks, effect of interventions, the size of interventions, the costs and benefits of actions — for fear of being wrong or, worse, sued, better decisions are impossible without numbers. A patient needs to know how much their risk can be increased or decreased, what the costs are in terms of time, money and discomfort, and all the side effects, both good and bad. Only then can the patient make a wise decision.
When considering a medical question, every bit of data is evidence. Evidence is anything that changes how likely something is to be true. Stronger evidence is better. There is no switch where something passes from ‘not evidence’ into ‘evidence’. Not all evidence takes the form of a clinical trial or other formal experiment. Likewise, many clinical trials are right, but many are also flawed. Many findings in medical journals can and should be disregarded, if there was a severe enough flaw in the experiment — one study found that flaws are so common, 80% of medical papers turn out to be false. We always consider the prior probability — how likely we think something was before we begin research — then adjust our beliefs and findings from that prior, based on the evidence we have. (We use what is known as the Bayesian method in mathematical statistics.)
Almost all research must also deal with bias. Humans have systematic biases that distort their beliefs. While the medical research system helps to combat some of them, it exacerbates others, and introduces new biases. A strong understanding of the role of bias is necessary to evaluate evidence.
The most important bias in medical research is publication bias, or the “file-drawer effect”. Studies that don’t show positive results are far less likely to be published. This is true even if no one intended to bias the results; no journal wants to publish a finding that two things aren’t linked, or that a proposed treatment does not work.
This effect is smaller when sample sizes are large — one sees a steady decline in average effect size as sample size increases. This is the result of publication bias: scientists report the results they found, not the results they failed to find. For this reason, one must be very careful not to overreact to a consensus of findings, and adding together different studies to form a meta-study will usually overstate how strong an effect is.
Every study is undertaken for a reason. If we don’t understand that reason, we don’t understand the study. In the study, a question was asked, and a theory was tested. But why was this hypothesis chosen? Sometimes, those funding the study are trying prove something that will be to their advantage, financial or otherwise. Drug companies fund many of the studies showing their drugs work. These studies are designed from the beginning to have the highest chance of producing results the drug companies want. Such studies can be done ethically, they can be done with publication bias, or they can be done with outright fraud and manipulation of results.