We can turn the clock back about 100 years to see what doctors were using then that was effective. That'll be fine for a large percentage of cases, and then the time lag between now and when a database of cures for common ailments is compiled won't be that long. Hey, what if there's one out already? What if a groups of scientists did a case study of these cures one by one to confirm efficacy? No one said potential patients couldn't network, either, including the smart ones ;-).
Then all the GP would have to remember is how to diagnose and then look it up in the database, and prescribe treatment.
