how do we encode glucose records?

glucose_codes.markdown

How do we encode glucose results?

We'd like to communicate using this type of thing:

• http://docs.indivohealth.org/en/v2.0.0/coding-systems.html
• http://wiki.chip.org/indivo/index.php/HOWTO:_install_Indivo_X
• http://dev.smartplatforms.org/reference/rest_api/
• http://dev.smartplatforms.org/reference/data_model/
• http://dev.smartplatforms.org/reference/app_manifest/

We have some example records here:

• https://github.com/medevice-users/diabetes/blob/master/trailing-500.txt
• https://github.com/jebeck/iPancreas
• https://github.com/bewest/decoding-carelink/tree/rewriting/analysis/bewest-pump The *.markdown files should line up with the *.csv in this directory.
• https://github.com/bewest/decoding-carelink/blob/rewriting/Carelink-208850-2012-08.csv
• https://github.com/bewest/decoding-carelink/blob/rewriting/Carelink-208850-2006-07-export.csv

EG:

2011-03-07T19:04:12	90
2011-03-07T17:21:34	143
2011-03-07T16:44:14	121
2011-03-07T14:45:46	144

So we have the time and the value, but what is the LOINC code?

LOINC

bewest@paragon:~/Downloads/loinc$ grep glucose ./* | wc -l
351
bewest@paragon:~/Downloads/loinc$ grep glucose ./* | grep -v -i -E "tolerance|enzyme|amniotic|peri" | wc -l
24
bewest@paragon:~/Downloads/loinc$ 
bewest@paragon:~/Downloads/loinc$ grep glucose ./* | grep -v -i -E "tolerance|enzyme|amniotic|peri" 
grep -i glucose ./* | grep -v -i -E "tolerance|enzyme|amniotic|peri"  | grep -i meter 
./LOINCDB.TXT:"14743-9"	"Glucose"	"SCnc"	"Pt"	"BldC"	"Qn"	"Glucometer"	"CHEM"	"OMH"	"19980116"	"ADD"		"ACTIVE"			1				"CORN SUGAR, D GLUCOPYRANOSE, D GLUCOSE, D GLUCOSE, DEXTROSE, GLU, GRAPE SUGAR,"						"Y"		"Glu; Gluc; Glucoseur; Substance concentration; Level; Point in time; Random; Cap blood; Cap bld; Capillary bld; Capillary blood; Finger stick; Quantitative; QNT; Quant; Quan; Glucomtr; Chemistry"	"Glucose BldC Glucomtr-sCnc"	"Both"				"mmol/L"	"Glucose [Moles/volume] in Capillary blood by Glucometer"	"NM/ST"	"PQ"			"mmol/L"					0	0	0	
./LOINCDB.TXT:"43151-0"	"Glucose meter device panel"	"-"	"Pt"	"^Patient"	"-"		"PANEL.DEVICES"	"SH"	"20051117"	"ADD"		"ACTIVE"			2										"N"		"; Glucose Meter Device Pnl; Glu; Gluc; Glucoseur; Point in time; Random; Pan; Panl"	"Glucose Meter Device Pnl"	"Order"					"Glucose meter device panel"										0	0	0	
./LOINCDB.TXT:"41653-7"	"Glucose"	"MCnc"	"Pt"	"BldC"	"Qn"	"Glucometer"	"CHEM"	"VA"	"20050705"	"ADD"		"ACTIVE"	"Glucose finger stick"		1										"N"	"mg/dL"	"Glu; Gluc; Glucoseur; Mass concentration; Level; Point in time; Random; Cap blood; Cap bld; Capillary bld; Capillary blood; Finger stick; Quantitative; QNT; Quant; Quan; Glucomtr; Chemistry"	"Glucose BldC Glucomtr-mCnc"	"Both"				"mg/dL"	"Glucose [Mass/volume] in Capillary blood by Glucometer"	"NM/ST"	"PQ"	"74-106;mg/dL"		"mg/dL"					0	41	0	
./LOINCDB.TXT:"41896-2"	"Class"	"Type"	"Pt"	"Glucose meter device"	"Nom"		"DEVICES"	"SH"	"20050727"	"ADD"		"ACTIVE"			2										"N"		"Typ; Point in time; Random; Glucose Mtr Dev; Nominal"	"Glucose Mtr Dev Class"	"Observation"					"Type of Glucose meter device"										0	0	0	
./LOINCDB.TXT:"41897-0"	"Vendor device name"	"ID"	"Pt"	"Glucose meter device"	"Nom"		"DEVICES"	"SH"	"20050727"	"ADD"		"ACTIVE"			2			"ONETOUCH® II, ONETOUCH® PROFILE®, ONETOUCH® SURESTEP®, ONETOUCH® FASTTAKE®, ONETOUCH® BASIC®, ONETOUCH® ULTRA®, ONETOUCH® ULTRASMART®, INDUO®, GLUCOMETER ELITE XL DIABETES CARE SYSTEM, ASCENSIA® CONTOURTM, ASCENSIA� BREEZE�, ACCU-CHEK® ACTIVE METER, ACCU-CHEK® ADVANTAGE SYSTEM, ACCU-CHEK® COMPACT METER, ACCU-CHEK® COMPLETE METER, ACCU-CHEK® VOICEMATE SYSTEM, PMP SELFCHECK� GLUCO"							"N"		"; Vendor name; Identifier; Ident; Point in time; Random; Glucose Mtr Dev; Nominal"	"Glucose Mtr Dev Vendor name"	"Observation"					"Glucose meter device Vendor name"										0	0	0	
./LOINCDB.TXT:"41898-8"	"Vendor device model code"	"ID"	"Pt"	"Glucose meter device"	"Nom"		"CHEM"	"SH"	"20050727"	"ADD"		"ACTIVE"			1										"N"		"; Model Cd; Identifier; Ident; Point in time; Random; Glucose Mtr Dev; Nominal; Chemistry"	"Model Cd Glucose Mtr Dev"	"Observation"					"Vendor device model code of Glucose meter"										0	0	0	
./LOINCDB.TXT:"41899-6"	"Vendor serial number"	"ID"	"Pt"	"Glucose meter device"	"Nom"		"DEVICES"	"SH"	"20050727"	"ADD"		"ACTIVE"			2										"N"		"; Serial #; Num; No; Identifier; Ident; Point in time; Random; Glucose Mtr Dev; Nominal"	"Glucose Mtr Dev Serial #"	"Observation"					"Glucose meter device Vendor serial number"										0	0	0	
./LOINCDB.TXT:"41900-2"	"Vendor software version"	"ID"	"Pt"	"Glucose meter device"	"Nom"		"DEVICES"	"SH"	"20050727"	"ADD"		"ACTIVE"			2										"N"		"; Software vers; Identifier; Ident; Point in time; Random; Glucose Mtr Dev; Nominal"	"Glucose Mtr Dev Software vers"	"Observation"					"Glucose meter device Vendor software version"										0	0	0	
./LOINCDB.TXT:"62852-9"	"PhenX - fasting plasma glucose for diabetes screening - glucometer protocol"	"-"	"Pt"	"^Patient"	"-"	"PhenX"	"PANEL.PHENX"	"PhenX"	"20110105"	"ADD"	"This protocol provides basic information for measuring blood glucose concentrations using a glucose meter or glucometer. Participants are asked to fast for at least 9 hours (generally overnight) prior to the test. If the 9-hour fast is not met, the Diabetes Working Group recommends that the test can still be completed as a ""casual"" glucose concentration (see protocol for casual glucose reference ranges for screening plus/minus Diabetes)."	"TRIAL"			2									"PhenX.140802"	"N"		"; Fast plasma gluc glucometer proto; Glu; Gluc; Glucoseur; Point in time; Random; SCN; Asymptomatic; Asympt"	"Fast plasma gluc glucometer proto"						"PhenX - fasting plasma glucose for diabetes screening - glucometer protocol"										0	0	0	

We have a few different types:

Fingerstick

Usually in mg/dL in US?

3. For finger sticks (measured by glucometer; units = mg/dL?), you should probably be looking at LOINC code 41653-7 http://s.details.loinc.org/LOINC/41653-7.html . https://groups.google.com/forum/?fromgroups=#!topic/smart-app-developers/0fXhUp7rHDw

Aha! Thanks, ~@joshmandell!

CGM Display, vs sensor

• https://github.com/jebeck/iPancreas/tree/master/public

how do we encode medical records?

• https://github.com/medevice-users/diabetes/wiki/Existing-Standards
• http://dev.smartplatforms.org/reference/app_manifest/
• https://github.com/chb/indivo_codingsystems/blob/master/data/loinc.py
• https://github.com/chb/json_ccda
• http://bioportal.bioontology.org/ontologies/47637?p=terms&conceptid=18041-4
• http://bioportal.bioontology.org/ontologies/1350?p=terms&conceptid=http%3A%2F%2Fpurl.bioontology.org%2Fontology%2FLNC%2FLP14635-4
• https://loinc.org/faq/getting-started/structure-of-loinc-codes-and-names/
• http://dev.smartplatforms.org/reference/data_model/#VitalSign

How do I say:

A "glucose value" is a CodedResult? LabResult of LOINC-41653-7.
There is a medical record for patient @bewest.
The medical record for patient @bewest has a list of these *glucose values*:
  * ts, value
  * ts, value
  * ...

Eg, just a minimal graph to say here is a list of sugar values for this patient.
It turns out there are some interesting semantics to play with here... I think if I could get started with a few demo RDF sentences, everything would start to unravel... it would be nice for example to retain the provenance of how the data was collected. Eg:

Here is an auditing session of meter #xyz on date:
Which are labresults of LOINC... for patient *link to foaf*

Then I can upload all raw sessions, and if my API is SMART/triple aware, I can ask for a distinct set of results over time, while just uploading the full results of each auditing session.

To say "here are two glucose values for bewest," you could do something like:

@prefix foaf: <http://xmlns.com/foaf/0.1/> .
@prefix sp: <http://smartplatforms.org/terms#> .
@prefix dcterms: <http://purl.org/dc/terms/> .

_:bewest a sp:MedicalRecord.

[] a sp:LabResult;
   sp:belongsTo _:bewest;
   dcterms:date "2013-04-01T12:12:00";
   sp:labName [
     a sp:CodedValue;
     sp:code <http://purl.bioontology.org/ontology/LNC/41653-7>
   ];
  sp:quantitativeResult [
    a sp:QuantitativeResult;
    sp:valueAndUnit [
    a sp:ValueAndUnit;
    sp:value "82";
    sp:unit "mg/dL"
    ]
  ].

[] a sp:LabResult;
   sp:belongsTo _:bewest;
   dcterms:date "2013-04-01T13:30:00";
   sp:labName [
     a sp:CodedValue;
     sp:code <http://purl.bioontology.org/ontology/LNC/41653-7>
   ];
  sp:quantitativeResult [
    a sp:QuantitativeResult;
    sp:valueAndUnit [
    a sp:ValueAndUnit;
    sp:value "95";
    sp:unit "mg/dL"
    ]
  ].

As for representing the auditing sessions, it's probably easiest to pick a representation that closely matches the actual device data (whatever that may be), and then later convert from there to whatever you need downstream. So up-front you might just capture something like:

[] a dm:AuditingSession;
   dcterms:date "2013-04-02T15:00:00";
   hasReading [
     a dm:FingerStickReading;
     dcterms:date "2013-04-01T13:30:00";
     dm:rawFingerStickValue "92" -- may even be expressed in some device-specific units?
   ], [
     a dm:FingerStickReading;
     dcterms:date "2013-04-01T18:30:00";
     dm:rawFingerStickValue "110"
  ], [
     a dm:FingerStickReading;
     dcterms:date "2013-04-02T09:00:00";
     dm:rawFingerStickValue "132"
  ]
].

(Just added missing sp:belongsTo predicates to the 1st examples.)