“Insert marvellous new medical breakthrough discovered by science” is going to change millions of lives based on research … in mice?
This is actually the background of nearly all those breaking headlines. Almost none of the exciting clickbait actually includes research and clinical trials performed on humans that’s ready to enter mainstream medicine (and makes for great humour for this Twitter account, @justsayinmice).
Which is also why none of the marvellous medicine of clickbait headlines seem to exist in your local doctor’s office. The simple truth is that the research was probably performed in mice, then the journalists took the publication and and hyped it up.
Wait for another decade and it might be approaching the first human trials … if, and only if, we’re just about certain that it won’t hurt anyone or have nasty side effects that didn’t show up in the mice. Give it two more decades of clinical trials and assuming all of that’s worked, and there weren’t any differences in the physiology of mice and humans, then perhaps the marvellous discovery will have made it as an option for the very rich who can afford it.
That’s a fairly standard progression of medical research. Technologies, drugs and methods are tested in either cell cultures or in mice in an attempt to replicate the human physiology and response to the procedure or medicine. It means that the majority of headline breaking research is still in the stages of mouse models and no-one is really certain of whether it’ll behave the same way in humans.
An unfortunate example of this issue was with the drug Thalidomide in the 1950s. It was a drug shown to alleviate morning sickness and had been extensively tested in mice and rapidly entered mainstream usage by many expecting mothers. Sadly, the mice had not shown any side effects due to them producing higher quantities of antioxidants in the womb which protected their offspring, whereas in humans, many children were born with severe deformities, over 10,000 affected cases were directly attributed before 1962.
We use mice because they do share many physiological similarities to humans in the way that they respond to different medicines and their lifespans are shorter so we can get data on the long-term effects fairly rapidly … and remarkably, humans are more keen to experiment on tiny white cheese-lovers than to put their on lives and health on the line without much idea of whether what they’re doing is about to work.
So mice it is.
Some scientists are then responsible for studying the differences between mice and humans to have a better idea of which “mouse medicines” will work effectively in humans, so that again, we can avoid testing unsafe medical technologies on living human patients.
And those scientists have just gotten a pay rise. A group in the Technion-Israel Institute of Technology have recently used machine learning to develop a computer model that helps them to better predict which mouse models may translate to humans.
Mice and humans share 78.5% of the same genes and this new model, named Found in Translation (FIT), helps to predict whether the genes responsible for a certain effect in the mice are likely to be found in the same role in humans and the target drugs will affect those genes similarly.
FIT was developed via machine learning, meaning that the researchers took data from 170 previous mouse-human medical trials and the related genetic information and fed it to FIT to sort through and determine which genes are similar in mice and humans and which aren’t. Then, to test how well it was working, they took some more already-published data and gave it to FIT to see if it would reproduce other well-studied phenomena. It performed with an accuracy of 88%. Not perfect, but good enough to easily veto completely unrelated trials without having to progress to a stage of human clinical trials.
FIT works by first evaluating the data set (genetic and health information) to see whether it is similar to the data sets that it learnt from during the machine learning phase. Obviously, if the genetic information is relatively new and untested, FIT has had no experience and cannot provide accurate indicators.
If the data passes this test, FIT then evaluates it and matches each gene to its database (garnered from the previous studies) and gives each gene a numerical value. The higher the numerical value assigned to the gene, the more likely it is that the medical finding will translate from mouse to human.
It sounds quite promising to more rapidly sort through large datasets and discard irrelevant genetic information that won’t translate and focussing medical research efforts and funding into clinical trials for better candidate drugs and therapies.
Best of all, it’s a free online tool called Mouse2Man. You can explore more about it, how it works, or even try uploading data (if you’re a real nerd) on their website here.