Medical Data Sharing

How do we know how to treat patients?
If we have a method that has been consistently successful in the past, then we can reasonably apply that treatment to patients with the same condition.

In oncology (cancer) this paradigm is often lacking. The probability of successful treatment may be low. The criteria for success may be distorted.

The selection of proper treatment involves matching the patient to be treated with those that have benefit ted from the treatment in the past.

Currently, the matching process is limited by the published Thus, matching is possible to the extent that one uses the criteria that the authors of a report decide to publish. If the reported criteria are the determinants of response, then the patient has a high probability of responding well to the treatment. If not, the chance of response is only vaguely related to the criteria used. When a cancer is uniform and there is a clear treatment, patients do well. When we are basing treatment on 30% response rates and we cannot identify who the responders will be, the patients chances are not good.

However,the current system may sell alot of drug! If we can't identify the responders but we know that they exist, there is an argument for treating every patient with the therapy that works in a fraction of them. That system sells much more drug than one in which only the most appropriate patients are treated!

All of this is leading to the proposal that data about patient response to treatment be made available in a minable, shared way. All of the data should be available to anyone that is treating that disease so that the data can be mined and the best possible matches made.

The technology for this is clearly available. It has been used for sharing music and entertainment. This is an extension of the Wikipedia idea. It is not unlike the resources that Google makes available to all comers.

I am recruiting people with the talents, skill and idealism to undertake this important ( possibly Quixotic) project.

Please help!