A few tests here and there probably wont do much good, but an abundance of regular examinations over a longer course, starting in childhood, will lead to more effective identification models, like how people with kidney or liver issues get regular blood tests to monitor trends in function. Making regular brain exams a thing as technology improves is beyond sensible given the potential benefits to health in general. And since data science is going to be a fact of life regardless of objection, using the results of those regular exams to profile for things like psychopathy/sociopathy might as well be a thing also. To avoid corruption, all tests will have to be administered absent conflicts of interest and it must be up to the subject of the testing to demonstrate absence of conflict lest a harsh penalty be levied, like how most drivers today are obligated to carry liability insurance at the very least.

You're right about the impossibility of the model in the moment, but I'm thinking longer term. As new technology emerges to better map the brain and it's fuctions, the easier it will be to profile for negative personality traits. In that time, infrastructure policies can be implemented to provide for robust ethics training aimed at personality types who are likely, based on past evidence, to abuse any accumulated power. In the here and now, we can begin by implementing policies which promote and encourage ethical uses of emerging tech, and which set stiff penalties for unethical abuses of the same. Making ethics tuition a prime factor in education from the earliest stages is one soecific policy I would advocate for in the present. And since my ethical bearing is Secular Humanist, I would advocate for that as a basis for all ethics training models.