Education and Trump

PTC (Positive Train Control) Another broken system?

I have no idea how PTC works but when you listen to all the excuses for why it has not been rolled out everywhere and the general technical specs you get the sense that it is another outdated approach that should be scrapped.

We now have self driving cars and we can't control a train that is moving on a fixed track??

Seems to me that the solution should be pretty simple. Outfit each train engine with a system that constantly uses GPS to know it's exact location. Then, based on the location, use the control application to know what the maximum speed limit is for that location and force the train to slow down to that level if necessary and report the over speed condition to train management on a real time basis so they can decide it it's necessary to contact the engineer. (Being 2 miles over not so big a deal, 25 over, red flag)

Mapping all the train routes with the max speed would seem to be a trivial task by having a Google type train that travels each route and establishes the maximum speed. In fact I suspect you could define that simply by monitoring radius of a turn and pitch of the tracks.

Think about it... you can buy a Garmon GPS for under $100 that shows you the speed limit for almost every road you travel on. Mapping that for train tracks should be easy.

Congress could fix this by simply mandating the solution by 2020 but of course our dysfunctional government will not do anything because they want the big contributions (bribes) from the railroad industry.

Medicare Fraud

Medicare fraud is estimated to be from $60 to $100 BILLION a year which is an astounding amount of fraud. And in today's NYT we learn...

Last week, when the Department of Justice charged three people in Miami with fraud and other crimes in a $1 billion scheme to bilk Medicare, it was the single largest criminal case in the nine-year history of the Medicare Fraud Strike Force, a coalition of federal, state and local law enforcement agencies. A month earlier, a crackdown by the strike force led to civil and criminal charges against 301 people in dozens of schemes totaling $900 million in allegedly false billings."

To me the most frustrating thing about this is the inept way fraud is investigated. Everything is done way after the fact probably using ridicules outdated tools.  How could we let someone run one billion of false charges before being stopped?  I think there is a much better approach to fix this once and for all. The key is early detection and utilization of current technology.


PART ONE  (Early Detection)

1) Require all providers to submit their claims within 14 days of the service or have a 20% penalty assessed for late submission. 7 days would even be better.

2) On a very timely basis enter the data into the system the day it is received.

3) Immediately after data entry send an email to the patient asking them to confirm that they in fact did received the service. Send snail mail for those who don't have email.

Today I get documents in the mail from Medicare showing the doctor visits and procedures that have been billed and paid by medicare. They typically come months after the event, so it's hard to remember what happened 3 months ago,  and there is nothing indicating that I should review these and report fraud.  I once discovered obvious fraud from a supplier of my INR test strips. They were billing for twice the amount that I actually received. I attempted to alert medicare but as I remember the process was cumbersome and after filling my report I never heard a thing about what, if anything, happened.

If the patients could simple click a box saying yes or no you can quickly start to see a pattern on a given provider AND someone could immediately call the patient to clarify what has or has not happened in cases where they clicked on the NO option.

4) Using very simple statistical analysis we should constantly compare a given doctors submissions against normal and average charges for similar procedures so you can quickly discover anomalies.

This is a no brainer approach.  How can one Doctor submit 15 claims a day for motorized scooters. The minute you see a statistical anomaly you flag the doctor for investigation and hold up payment.

PART TWO ( Investigation)

Investigation of potential fraud must be very timely so it can be nipped in the bud. 

Here is how it should work....

1) All payments that are flagged for investigation because of statistical anomalies, patient feedback and other causes, (like first submission from a new clinic or doctor) , are added to the investigation queue with step one actions.

2)  iPads or other tablets are used to direct field investigators on what they are to do. This would be a rich and very intelligent application that manages the investigators day. Here is how it would work.

a) Investigator opens the app at his home for the start of the day. Obviously the app captures his start time, etc.

b) App gives him or her their  first assignment. Assignment would typically start with a map showing them to go to a doctor/clinic address and the app would give clear directions and act as a GPS for the agent. 

c) Once they have arrived at the address they click on ARRIVED. and then the app gives them the next instruction. Things like...

-Verify that the office actually exists.  (Does the office exist, yes or no)
  If yes, show next step. If No ask agent to describe what they see at the address if it even exists.
-If yes then various types of instructions could be given such as:
  
Go in and attempt to make an appointment. Observe what you see and answer these questions:
      Did the office look official and normal for the type of service.
      Was there a receptionist who could allow you to make an appointment
      Were you asked about the current provider and insurance.
      Etc.
or

Wait for n hours and observe patient activity.
     How many patent's came and left
     And perhaps other questions that relate to the type of service, etc.

- Once done with this office, present the investigator with the next address, etc.

Now the obvious advantage of this relatively simple process is that you are getting timely feedback about a given provider that is very actionable. If the office does not exist you shut them down, hold payment and refer to next step potential criminal processes. If it exists but there was no lobby or receptionist you go to the next steps, etc.  If no patients came and went in the normal business hours when the office was supposedly open then it's probably a scam requiring further investigation. You could loop back the next day and have the agent observe things again, etc.

There are all kinds of thing one could do with this approach. 

The cost to develop this app is literally floor sweepings compared to level of fraud. The tasks for the agent are pretty low level so the cost per agent, including salary, travel and and I.T. support should not exceed $85k per year. At that rate you could hire 20,000 agents for about $1.7 billion a year.  I doubt you would need more than 2,000 to have a significant impact. I guarantee that fraud would drop by 75% with this approach. Will medicare do anything like this? Sadly I suspect nothing will change.