The birth of a startup idea
My father and I argue quite often what’s the best thing’s to get things done.
I like to think in the abstract about ideas and how things potentially could be, I think my dad is more explaining how the world is as of now.
He’s a Cardiovascular surgeon and does research. A while ago he told me about some publication he was working on.
He’s looking for a bio marker to see if someone have had a blood clot formed in their blood vessel.
For some reason I asked if there was any chance to access the raw data linked to similar studies.
I was imagine that there would exist some kind of “map” of all the collected data related to blood clots formed in blood vessels. But apparently it didn’t exist. Instead it seems to be common practice among researchers to keep their data by themselves and occasionally work together with a few researchers to put together a smaller pool of data.
Discovering a existing research method.
I discussed further with my father what I was trying to explain, he then told me that a research method called meta-analysis might resemble what I wanted to do, and this was quite spot on.
Meta-analysis search new knowledge from existing data.
It’s done through combining big amounts on data to find new correlations or building stronger proof for existing hypothesis.
If you think about the other word for meta analysis, quantitative synthesis. That really explains the concept in just two words.
The benefits from meta analysis is quite self-evident.
You build strong support for current hypothesis. This leads to quicker implementation of new treatment techniques.
The other benefit is that you can find completely new correlations, which leads to new knowledge and possibly new practical use.
One example of a recent meta analysis that successfully reaped knowledge was a study who found a genetic marker for breast cancer.
This opens the possibility to take precautions before actual development of cancer.
Meta analysis takes a lot of time.
Let’s go back to what I originally was looking for. “Some kind of map that showed all the existing data on a specific medical condition”.
I believed that it existed, cause it seemed like it just had to.
But from what I’ve found there aren’t.
Most of today’s meta-analysis are done through manually collecting raw data from individual studies.
There are no “Google” for raw medical data. Some people will probably tell that yes there is, Pubmed. But that’s not entirely true, because that’s mostly separate studies that are published, without easy accessibly raw data from the studies. (Pubmed is still a great organization and probably will lead further development of accesability and presentation of to research material) The closest you get are some bigger data pools that are being created by individual governments. But they still lack a lot.
So today when meta-analysis is done, researchers have to:
- Locate potential articles
- Contact primary authors
- Manually calculate effect sizes
Collecting one large data pool.
Through creating one collective data pool, it’ll be possible to look into data that simply isn’t possible through individual studies or smaller meta-analysis. It will be like one big meta-analysis where you simply decide the filters on what to look at and what to include or exclude.
This will let researchers to focus on other parts of the meta-analysis
than the actual gathering of data, like:
- Deciding Inclusion and Exclusion criteria
- Identifying outcomes
- Analyze the effect size (Correlations found in empirical data)
- Forming new hypothesis or building stronger proof from questions that one had prior to the meta-analysis.
- Initially not all available data will be collected in our database. Which may lead to a weak meta-analysis.
- The key difficulty in deciding which sets of studies are ‘combinable’.
Question every researcher have to take into consideration.
- Will the scraped data be structured in a way so it’s able to represent?
- How will one go about creating a standard structure to new submitted raw data?