I am pleasantly surprised. Always nice to see people who have made such a transition.
I agree with you on all points here. Once in production, stability is achieved and car companies certainly do not hold all the secrets to resource allocation and administrative approaches to healthcare.
The degree of unpredictability is higher in software and medical procedures. The supply chain - production approach can only get so far.
That being said, the production approach can work splendidly well for low risk diagnoses. The production approach has already been adopted by some software companies. Especially contract based companies that involve a well defined repetitive grind and set of tools. Think developing a Cash-of-clans like mobile game or making websites for certain organizations. Both industries are quite mature. The software problem isn't interesting, but it also rarely throws curve balls.
_______
I am going on a slight tangent, but here goes.
I think there are some medical low hanging fruits. This is especially true in the US, where the escalation of service is extreme. Either you get no service or a $200/hr medical expert. If the disease is going to be a low risk common illness 95% of the time, then maybe the patient could go through a $40/hr medical practitioner whose job would be to attends to 'easy' cases and only escalates it to the $200/hr doctor if the case is serious enough. A similar case applies to the ER and ambulances too. There is, "I am about to die" emergency and then there is the "I am hurt, but 20 more minutes to the hospital in car won't change much" type of emergency.
Off topic: But, This approach is quite similar to a popular ML approach called Cascade Classifiers used for resource efficient applications. It is apt, that it would come up in a discussion about how to make Healthcare more resource efficient.
Lastly, I find the requirement of 4 year college to enter medical school to be preposterous. It is a waste of money and time for applicants and serves only to line pockets of academic institutions.