WHERE MEDICAL RECORDS GO TO DIE
Bill Inmon
Founder, Chairman, CEO, Best-Selling Author, University of Denver & Scalefree Advisory Board Member
WHERE MEDICAL RECORDS GO TO DIE
By W H Inmon
Take a look at a medical record sometime. Medical professionals were sold on the proposition that if they went for EHR (electronic health records) or EMR (electronic medical records) that medicine would be brought into the 21st century. Indeed, EHR was much better than the old card system that was written manually for medical records. Now records could be exchanged across networks. Now records could be organized electronically. And this indeed was an advancement in the modernization of medicine.
No question about that.
But thinking that EHR was Nirvana was a mistake. There is a great deal that is missing from medical records. Look closely at your EHR medical record and see just how much of the medical record is in the form of text. A lot of the EHR is in the form of text.
So what is the problem here? If you want to read a person’s medical record, can’t you just start reading the text? The answer is of course you can. But what if you want to read 100,000 medical records. Can’t you do that? Of course you can, if you have nothing to do for the next year or so. Furthermore, after you have read 100,000 records, exactly how many of them can you recall when you have finished? The answer is maybe ten or twenty records, if you are lucky.
So you really can’t manually read 100,000 medical records after all.
So why would a person want to read 100,000 medical records? Well, from a research standpoint it is quite useful to see how a body of people are reacting to a drug. Or an immunization. Or a procedure. There is great value in medicine in being able to understand how medicine is practiced over a large segment of the population. Where has the drug been effective? Where has the drug not been effective? What types of people are most sensitive to the drug? Not sensitive to the drug? Under what conditions should the drug not be used? These questions are vital to the practice of medicine.
The problem is that when the effectiveness of a drug is locked up in the form of text that the drug’s secrets might as well be locked up as well. In jail.
In order to unlock the secrets of a drug, the results need to be transformed from text into a tabular form fit for a computer. Once text is translated into a tabular form, the computer can easily and handily read 100,000 records. If you don’t make this transformation, you cannot look at 100,000 records.
And your data just goes away to die. It is locked in jail. And that is good for no one.
The good news is that now we have technology – Textual ETL - that can transform medical records into a standard data base. There does not need to be a modern elephant boneyard. Unless of course that is what you want. Now – if you want – you can read and analyze 100,000 records. Easily and economically and quickly.
Bill Inmon is from Denver, Colorado. Bill has two Scottie dogs – Tipper and Jeb, and Bill feeds them faithfully every night. If Bill forgets to feed them, he is quickly reminded at 5:00 pm every evening. He may forget, but Jeb and Tipper do not forget. Ever.
We do magic with Software and Data for high tech industry since 1993
3 年Urmas Kobin have a look!
Healthcare Project Management
3 年Unstructured/semi structured text is the bane of many EHR analysts. The challenge, as you mention, is that much of critical healthcare data is captured in an unstructured manner. Additionally, with the proliferation of Telehealth, you are now faced with the additional challenge of generating information from virtual conversations.
Consultant - Enterprise Data Solutions
3 年Bell the cat!
Bid Forensics Consultant, Capture/Proposal Manager, Volcanic Geologist, US Army (MI) Veteran
3 年Bill, I told my doctor a couple of years ago that his office does not have my permission to put my medical records into any computer system that is connected to the internet. There is no negotiating that. It usually takes me a long time to read and redact HIPAA documents and I will do it only on paper.
Healthcare IT Executive
3 年I would say that a better answer rather than NLP and textual ETL (which is good, but will never correctly identify all instances) is collecting the data from the beginning correctly. I've found that this is done exceedingly well using a tool from Medicomp called Quippe.(https://medicomp.com/quippe-documentation/). We used Quippe at a multi-hospital system in Indonesia and it worked very well.