Yes that would count as evidence but only if you modified your experiment slightly:
Don't tell anyone that you will pray for them.
Instead of personally praying for each person, give the list of participant names to someone you trust.
This person can then pray for a subset of the people listed on random days, recording the person they prayed for and the day.
You conduct interviews with the people as you suggested.
After you record the results of the interviews, you then look at the data from the person who prayed and see where things matched up. You can then observe if there are any statistically significant differences between those who were prayed for and those who were not
The reason this counts as evidence is because it's not eyewitness testimony, it's a controlled experiment which should be reproducible by anyone. By itself it doesn't prove anything but it would help to start building a body of evidence that prayer can work, or not depending on your results.