Search the KHIT Blog

Tuesday, January 9, 2018

Blockchain and health care?

Ran across a post over on Medium, "Can Blockchain Help Us Improve Health?"
“There’s great excitement and huge investment around what blockchain technology can do to improve the delivery of health care...” BUT “...not everyone is equally optimistic that this technology is more than a passing fad”
I commented thereunder:

My concern goes to Gartner Hype Cycle conflation. Crypto financial transactions need, minimally, only four data elements, short and sweet:

  1. Payor_ID;
  2. Payee_ID;
  3. Amount;
  4. DateTimeStamp.
(And, of course, implicit in the foregoing are the (often pseudonymous) “Private Key” ID linkages to the transactors’ “Public Keys” that are used to validate the transaction, reflected via the DateTimeStamp. (Oh, and, gotta mention the “security hash” code linking the blocks.)

In contrast,
  1. The typical ambulatory EHR houses between 3,500 and 4,000 variables in its RDBMS schema (to say nothing of huge inpatient systems). A typical patient encounter may reflect hundreds of them (or more, comprising structured alphanumeric data, imaging, and open-ended text narratives). There were 60 vars in my most recent bloodwork alone. A typical “ROS” (Review of Systems data) houses 140 or so variables. And, all of these are typically “1 to n,” i.e. one-to-many longitudinal per patient (i.e., “progress note” stuff);
  2. Given that the ostensible “virtue” of the blockchaining architecture is its “immutability” (i.e., validated blocks cannot be modified once added), how do we handle the inevitable errors that plague all database systems? Find mistake(s)/omission(s), have to “append” execute a new block transaction (which is again “distributed” to everyone in the aggregate “ledger”/”wallet” population)?
  3. HIPAA 45.CFR.164.3,4,5 et seq data security, breach notification, and privacy requirements governing CE’s and their BA transactors (and also 42.CFR.2)?
  4. What of the continuing (lack of) “interoperability” problem (which I often irascibly refer to as “interoperababble”)? What of the clinical workflow implications? Clinicians are already drowning in data as they traverse the never-ending “productivity treadmill.”. How will the data variables in the validated new ‘blocks” be viewed by recipients? Will we need new, custom “download APIs?”
  5. Blockchain transactions are not “free,” they require payment of “transaction fees.” Crypto transaction fees have been on a steady, significant rise. Will they be reasonable (and stable) for health data transactions? Will such fees be bandwidth-consumption based?
Dunno. I have concerns. More inefficient IT Geek playground stuff w/respect to Health IT?


Currently, there are some 20 million crypto "wallets/accounts" (many of them anonymous duplicates), each a recipient of every accruing block addition transaction (the core peer-to-peer "distributed ledger" concept). I have to question the "bandwidth and data footprint efficiency" of this model -- beyond other reservations (some of them related).

Just some initial reactions.



apropos of the overall "crypto" tech topic, from NPR's Science Friday:


From Naked Capitalism -- to my foregoing points:
"Without belaboring the topic further, what amounts to a newfangled way to do a database does not solve the problem of data integrity. In fact, that is almost certainly more easily addressed (as in cleaned up over time) with traditional databases. Yet many of the hoped-for solutions, like in trade, act as if blockchain can solve the data accuracy problem, which exists independent of tampering risk..."
Jus' sayin'.

More to come...

No comments:

Post a Comment