An Open Source Clinical Trial of Coronavirus Vaccines

As the world’s scientists race to find a cure for the Coronavirus, will they work in private labs or will they openly share data and work in an “open source” research environment.  While fame and fortune (the Nobel Prize, billions in drug patents) would suggest the private lab approach, science will make faster progress if they work as “open source hippies”.

Suppose that each research team has its own favorite drug treatment.  This treatment consists of injecting m different compounds into a person suffering from the Virus. So, if m=2 — each treatment can be described by how much of the first attribute and how much of the 2nd attribute is injected into a person to potentially cure the sick person.  Call this (x1,x2).

If each research team reports the following information;

Did the Coronavirus infected person recover within Z days?
Age of the infected person
gender of the infected person
race of the infected person
major medical pre-existing conditions of the infected person
(x1,x2) injected into the person.

Then a Machine Learning team can use these data from each research team (so if there are 20 research teams that each enroll 500 sick people) there would be 10,000 data points in total.

The ML approach would yield an estimate of what is the best treatment for different demographic groups and an efficient vaccination process could begin.

If the samples can be pooled across science teams, could the vaccine be more quickly identified and started to be mass produced?

Do science teams have ways to co-ordinate their sharing of their data or will some famous scientist start the process by sharing her data and hoping others will follow?

In the past economists have debated the merits of open-source science.    In a high stakes setting, these issues arise again.