Download presentation
Presentation is loading. Please wait.
Published byEdgar Cobb Modified over 9 years ago
1
Bringing the technology of FedEx parcel tracking to the Electronic Health Record (EHR) 1/2 0
2
Barry Smith Director, (US) National Center for Ontological Research – leader on ontology projects for US Army Founding Coordinating Editor of the OBO (Open Biomedical Ontologies) Foundry project Consultant to German Federal Health Ministry on ontology projects 2
3
Principal Investigator ◦ Protein Ontology ◦ Infectious Disease Ontology Scientific Advisor ◦ Gene Ontology (world’s most successful ontology) ◦ Ontology for Biomedical Investigations (OBI) ◦ Cleveland Clinic Semantic Database in Cardiothoracic Surgery 3 Barry Smith
4
Barry Smith: Funding Work funded by European Union, US, Austrian and Swiss National Science Foundations, Volkswagen Foundation, Humbolt Foundation Winner of $3 million Wolfgang Paul Prize from German Government National Center for Biomedical Ontology $18 million collaboration of Stanford University, The Mayo Clinic and Smith’s group at University at Buffalo 4/2 0
5
What is tracking technology? The use of unique IDs (e.g. bar codes) in order to track items as they move through a system. ◦ Logistics ◦ Manufacturing ◦ Transportation of merchandise ◦ Handling of evidence in a criminal investigation ◦ Etc. 5/2 0
6
What does tracking technology do for the businesses who use it? FEDEX ◦ Coordination: With parcel tracking FEDEX links different activities to the parcel itself as it moves through the system. FEDEX activities: ◦ Receiving ◦ Addressing ◦ Billing ◦ Scheduling ◦ Delivering ◦ Locating (in transit) ◦ Relocating (if lost) ◦ Restituton (in case of loss) 6/2 0
7
Tracking technology Allows the customer (sender and recipient) ◦ to inform himself of the process of a parcel through the system ◦ to influence this process in case of need Tracking allows coordination of many different tasks without mixing up information about one parcel and another. 7/2 0
8
Tracking technology is already used as a matter of course for bank and credit card transactions cargo and logistics conveyor belt manufacturing mp3 files (for Digital Rights Management) archeological specimens forensic evidence tracking 8/2 0
9
Why should medicine adopt it? Patients ◦ grow older or change domicile ◦ change health care provider ◦ need specialist care. Because there is no unified ID system in the various health care institutions, all kinds of opportunities to ensure continuity of care are lost. Even where a reliable system of patient IDs exists, there is no ID system for ◦ the patient’s disorders ◦ the patient’s documents ◦ the many other items relevant to patient health, recorded always in a general way 9/2 0
10
Two levels to this problem the level of what is general ◦ the patient record systems used by different healthcare providers use different terms for the same general kinds of entities on the side of the patient (here ontology comes in) the level of what is particular ◦ tracking technology is not employed to link the different patient record systems together 10/ 20
11
Real world example (true story) Person P with complaint X ◦ Went out of town and became sick. ◦ Attended a series of different healthcare providers ◦ Informed each in turn that he had a quite specific ailment which needed urgently a quite specific kind of treatment. ◦ In part, at least, because the tracking mentality is so alien to the medical world, each new set of healthcare providers insisted on re-diagnosing from scratch. ◦ The process wasted time and almost killed him. 11/ 20 http://www.syleum.com/2009/03/17/healthcare-data-model/
12
The Electronic Health Record Currently the Electronic Health Record is standardly conceived as a digitalized record of what the doctor thought, saw or did ◦ what tests he did ◦ what medicines he prescribed ◦ what hypotheses he formulated In the US, the record is primarily used to support hospital billing needs In other countries it is used as an aid to diagnosis 12/ 20
13
Currently the Electronic Health Record is focused on the doctor Data is organized according to which doctors saw the patient, not according to the problems which were treated, creating data clumps which are very often isolated from each other 13/ 20
14
This blocks continuity of care Different hospitals used different record systems If, in contrast, you have a health record which ◦ tracks the patient, ◦ and the patients different problems, ◦ and treatments through time, then you can ◦ glue together the records of the patient at different times ◦ create an easily accessible and re-usable history. 14/ 20
15
The Electronic Health Record Still works with general terms for almost all the entities recorded 15/ 20
16
The story of Jane Smith 16/ 20
17
Jane’s favourite supermarket July 4th, 1990 Jane goes shopping 17/ 20 The freezer section of Jane’s favourite supermarket The only available warning sign used outside An injured upper leg
18
A visit to the hospital 18/ 20
19
Diagnosis: severe spiral fracture of the femur 19/ 20
20
General types and particular instances General type ◦ Fracture ◦ Headache ◦ human being ◦ Death ◦ fall ◦ accident Particular instances ◦ Mary’s fracture ◦ my current headache ◦ Me ◦ Reagan’s death ◦ Mary’s fall ◦ Jim’s accident last Wednesday 20/ 20
21
Kinds of codes in the EHR The EHR uses specific codes (alphanumeric ID’s) for: ◦ Patients (for example your Social Security Number) ◦ Physicians (standardly assigned by the hospital) ◦ Times (date, time of day) ◦ Places (each hospital will use a different ID system e.g. for buildings) But it uses general ‘observation codes’ (‘obscodes’) for everything else. The obscodes are taken from standardized clinical terminologies The record tells us that there is some particular fracture – an instance of the general class ‘fracture’ – but which one? 21/ 20
22
Examples of observation codes used in the EHR 557204/07/199026442006 closed fracture of shaft of femur 557204/07/199081134009 Fracture, closed, spiral 557212/07/199026442006 closed fracture of shaft of femur 557212/07/19909001224 Accident in public building (supermarket) 557204/07/199079001 Essential hypertension 093924/12/1991255174002 benign polyp of biliary tract 230921/03/199226442006 closed fracture of shaft of femur 230921/03/19929001224 Accident in public building (supermarket) 4780403/04/199358298795 Other lesion on other specified region 557217/05/199379001 Essential hypertension 29822/08/19932909872 Closed fracture of radial head 29822/08/19939001224 Accident in public building (supermarket) 557201/04/199726442006 closed fracture of shaft of femur 557201/04/199779001 Essential hypertension PtIDDateObsCodeNarrative 093920/12/1998255087006 malignant polyp of biliary tract PtID = Patient ID ObsCode = Observation Code taken from SNOMED (Systematized Nomenclature of Medicine)
23
557204/07/199026442006 closed fracture of shaft of femur 557204/07/199081134009 Fracture, closed, spiral 557212/07/199026442006 closed fracture of shaft of femur 557212/07/19909001224 Accident in public building (supermarket) 557204/07/199079001 Essential hypertension 093924/12/1991255174002 benign polyp of biliary tract 230921/03/199226442006 closed fracture of shaft of femur 230921/03/19929001224 Accident in public building (supermarket) 4780403/04/199358298795 Other lesion on other specified region 557217/05/199379001 Essential hypertension 29822/08/19932909872 Closed fracture of radial head 29822/08/19939001224 Accident in public building (supermarket) 557201/04/199726442006 closed fracture of shaft of femur 557201/04/199779001 Essential hypertension PtIDDateObsCodeNarrative 093920/12/1998255087006 malignant polyp of biliary tract Different patients, same fracture codes: Same (numerically identical) fracture ? Different patients, same fracture codes: Same (numerically identical) fracture ? Same patient, different dates, same fracture codes: same (numerically identical) fracture ? Same patient, same date, 2 different fracture codes: same (numerically identical) fracture ? Same patient, same date, 2 different fracture codes: same (numerically identical) fracture ? Problems 23/ 20 Different patients. Same supermarket? Same freezer section? Same patient, different dates, Different codes. Same (numerically identical) polyp ? Same patient, different dates, Different codes. Same (numerically identical) polyp ?
24
Enormous problems with current practices It is difficult to count the number of (numerically) different diseases, disorders, medical problems in a given patient This produces bad statistics on: ◦ Incidence ◦ Prevalence ◦ Cost It is difficult to identify outcomes (and thus determine value added) ◦ Is this the same disorder now as that which was recorded 3 years ago? 24/ 20
25
Enormous problems with current practices Suppose the doctor sees several patients all of whom complain of nasty viral warts. He sees in all their records that they all ‘visited swimming pool’ ◦ but the records do not tell him that they all visited the same swimming pool ◦ swimming pools are not identified via unique IDs. 25/ 20
26
unique numerical IDs for all concrete individual entities relevant to the diagnosis and therapy of each patient 26/ 24 The solution
27
This proposal can be implemented harmlessly The Fedex-style tracking layer would in 99% of the cases work behind the scenes ◦ would not force everybody to speak a new language ◦ or to learn new technology. Only in very rare cases would it require intervention from the doctor/coder ◦ but these are precisely the cases where the referent tracking data is of most help to both the doctor and the patient 27/ 20
28
It can be implemented incrementally For example, test it first ◦ in the hospital’s blood bank services ◦ in the hospital’s organ transplant services 28/ 20
29
In medicine The treatment of general types (diabetes, fracture, death) is dealt with quite successfully by medical terminologies and ontologies The treatment of particular instances is still underdeveloped 29/ 20
30
Solution: Referent Tracking Method: introduce Unique Identifiers for each relevant particular ◦ for Mary’s fracture ◦ John’s heart ◦ Jim’s tumor Result: An ever growing map of clinical cases, and of their interrelations to other clinical cases 30/ 20
31
Services provided by our solution ◦ automatic generation of IDs for particulars ◦ automatic assignment of these IDs to the particulars which need to be referred to in the EHR ◦ repository for all the Ids ◦ Database of statements relating the corresponding particulars to the types recorded in the EHR 31/ 20
32
Advantages of the solution Referent tracking can solve a number of problems in an elegant way Existing coding systems and technologies can be used for the implementation The IDs are generated by software behind the scenes, so that normal coding practices can continue as before 32/ 20
33
Advantages of the solution Because the same patient often attends different hospitals using different coding systems the use of common ID repositories will gradually lead to automatic mappings between terminologies Thus, the big problem of continuity of care in a world where different hospitals have different EHRs and thus use different terminologies will be resolved. 33/ 20
34
As IDs come to be associated with a plurality of codes from different coding systems We can run statistical tests to find outliers ◦ codes which were misapplied ◦ or ill-defined Or we can test and enhance rules for reasoning postulated by coding systems such as SNOMED-CT 34/ 20 Advantages of the solution
35
Over time, the ID repository will come to serve as a benchmark of correctness for coding systems, allowing automatic step-by- step improvements 35/ 20 Advantages of the solution
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.