Cohort Static Demo

Printer-friendly version

Do the ADA guidelines cover a diverse population, such that an African American diabetic patient on Metformin, can receive a recommendation?

Explanation: In this question we are trying to provide a physician with a comprehensive description of the characteristics, (in this case racial diversity) of patient groups utilized in research studies, exposing findings of effective treatment of diabetes with Metformin, or the long term effects of the same. A physician is probably interested to see if the study from our catalog, detailing the Long term effects of Metformin ( a treatment prescribed by the physician ) , is applicable to his/her patient solely on their African American descent. A query - Query1, is triggered to this particular study that matches the physician’s query criteria of “dbt:Intervention - dbt-i:Metformin” and “dbt:AfricanAmerican” to “dbt:ClinicalTrial” with a title of “Long-term Metformin Use and Vitamin B12 Deficiency in the Diabetes Prevention Program Outcomes Study”. However the query surfaces the results that the study doesn’t particularly have a clear stated African American population, and under the closed world assumption the physician is shown a AfricanAmerican racial percentage representation of 0.

However the physician is still curious to see the general applicability of his patient’s race to the populations of patients utilized in Metformin Clinical Trials, This time round just to ensure that the physician gets a clearer idea of the racial diversity, he/she makes a broader Cohort Q2 request to find all “dbt:Race” studied in “dbt:ClinicalTrial” with “dbt:PatientGroup” ( owl:Class of dbt:Patient in the alternate representation) administered a “dbt-i:Medicalntervention” of “dbt-i:Metformin”. The physician is shown race percentages of races - {White, AfricanAmerican, Asian, Other} represented in the only other study - “10-Year Follow-up of Intensive Glucose Control in Type 2 Diabetes“ ingested.

Note on Technical Representation: We have two methods of representing descriptive statistics in our project, one which doesn’t involve punning and reification and is realizable in the RDF world alone; and another which requires us to use N-Triples format and necessitates the need for defining descriptions and definitions on sets of patients represented as ‘ow:Class’es. Our alternate approach is inspired and borrowed from James McCusker et al’s paper on “A Provenance-Driven Semantics of Aggregation”. We have two query versions for our below two cohort queries, and each of them need to be run against the latest versions of our separate Individuals files - DiabetesTreatmentSupport_Individuals.rdf and DiabetesTreatmentSupport_Individuals_Alternate.ttl respectively. Please refer to queries from “Query for Alternate Representation” if utilizing the alternate N-Triples representation.

Walk through the workflow: The following steps returns the cohort information.

Legend

The above legend is used to describe the instances of the classes.

Classes

The above classes from our ontology are used to answer the competency question, which seeks to identify if an African American patient such as our patient Sam, is represented within the study detailing the physical adverse events of Metformin.

The treatment document provides us with a drug that is suggested to our patient, Sam, based on his profile (details on how this answer was found is in the Recommendation competency question). The suggested drug Metformin has a toxicity of Vitamin B12 Deficiency ( details of how this toxicity was arrived at are found in the Toxicity competency question) studied in a Clinical Trial cited within the Pharmacologic Interventions to Glycemic Treatment ( Chapter 8 ) in the ADA Standards of Medical Care guidelines.

Provenance

Above is a figure depicting the components of a Clinical Trial – “Long-term Metformin Use and Vitamin B12 Deficiency in the Diabetes Prevention Program Outcomes Study” attached as provenance to the Vitamin B12 Deficiency toxicity of Sam’s suggested drug Metformin. As seen in the figure, the VitaminB12DeficiencyStudy has a cohort, which is also associated to the treatment document. A physician might want to step into this cohort, to view if the patient populations of the study cohort are a good fit for his patient Sam.

Narrowed down

Now that we have narrowed down our provenance search on the long-term effects of Metformin usage, we can take a step towards finding the applicable patient group within the study administering the same medication of Metformin, suggested to our patient Sam. Seen above are instances and classes that are required to help us find patient group (s) within the study cohort, within which patients are given Metformin.

Patient Group

We have identified the patient group within the long-term effects of Metformin study administered a medication of Metformin, our physician now wants to find the percentage of patients like Sam’s racial descent within this group. The portion of the concept map seen in the above figure, aids towards finding racial percentages, for eg: White population within the MetforminPatientGroup.

Elaborate Description

Seen above is a more elaborate description of our previous figure wherein we delve into the percentage of MetforminGroupWhitePopulation via a hasMeasure property. A physician will be shown the hasValue of the MetforminGroupXRacePercentage if the race is present explicitly, else as in this case; the physician will be shown a value of 0 for African American patients within the MetforminPatientGroup.

Query 1: What is the percentage of African American patients represented in the study detailing the long-term effects of Metformin?
Representation 1

  1. prefix dbt: <https://tw.rpi.edu/Courses/Ontologies/2018/DiabetesTreatmentSupport/DiabetesTreatmentSupport/>
  2. prefix resource: <http://semanticscience.org/resource/>
  3. prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
  4. prefix sio: <http://semanticscience.org/resource/>
  5. prefix dct: <http://purl.org/dc/terms/>
  6.  
  7. select DISTINCT ?studyTitle ?medicationDrug ?AfricanRacePercentageVal WHERE {
  8.  
  9. ?study a dbt:ClinicalTrial .
  10. ?study dct:title ?studyTitle .
  11. ?study dbt:hasCohort ?studyCohort .
  12. ?studyCohort dbt:hasPatientGroup ?patientGroup .
  13. ?patientGroup dbt:hasIntervention ?studyIntervention .
  14. ?studyIntervention a dbt:Medication ;
  15. dbt:hasDrug ?medicationDrug .
  16. ?medicationDrug a dbt:AMPKDrug .
  17. ?patientGroup dbt:hasCharacteristic ?race .
  18. ?race a ?raceType .
  19. ?raceType rdfs:subClassOf* dbt:Race .
  20. ?race dbt:hasMeasure ?racePercentage .
  21. ?racePercentage sio:SIO_000300 ?racePercentageVal .
  22.  
  23. BIND (
  24. IF(?raceType = dbt:BlackOrAfricanAmerican,
  25. ?racePercentageVal, 0)
  26. AS ?AfricanRacePercentageVal
  27. )
  28.  
  29.  
  30. FILTER (regex(str(?studyTitle), "Long-term Metformin") ) .
  31.  
  32. }

Representation 2

  1. prefix dbt: <https://tw.rpi.edu/Courses/Ontologies/2018/DiabetesTreatmentSupport/DiabetesTreatmentSupport/>
  2. prefix resource: <http://semanticscience.org/resource/>
  3. prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
  4. prefix sio: <http://semanticscience.org/resource/>
  5. prefix dct: <http://purl.org/dc/terms/>
  6.  
  7. select DISTINCT ?studyTitle ?medicationDrug ?AfricanRacePercentageVal WHERE {
  8.  
  9. ?study a dbt:ClinicalTrial .
  10. ?study dct:title ?studyTitle .
  11. ?study dbt:hasCohort ?studyCohort .
  12. ?studyCohort dbt:hasPatientGroup ?patientGroup .
  13. ?patientGroup dbt:hasIntervention ?studyIntervention .
  14. ?studyIntervention a dbt:Medication ;
  15. dbt:hasDrug ?medicationDrug .
  16. ?medicationDrug a dbt:Biguanide .
  17. ?subPatient rdfs:subClassOf ?patientGroup .
  18. ?subPatient sio:SIO_000008 ?attr .
  19. ?attr sio:SIO_000300 ?racePercentageVal .
  20. ?subPatient rdfs:subClassOf ?restriction .
  21. ?restriction a owl:Restriction .
  22. ?restriction owl:someValuesFrom ?raceType .
  23.  
  24. ?raceType rdfs:subClassOf* dbt:Race .
  25.  
  26. BIND (
  27. IF(?raceType = dbt:BlackOrAfricanAmerican,
  28. ?racePercentageVal, 0)
  29. AS ?AfricanRacePercentageVal
  30. )
  31.  
  32. FILTER (regex(str(?studyTitle), "Long-term Metformin") ) .
  33.  
  34. }

Result

Study Title Medication Drug African American Percentage
Long-term Metformin Use and Vitamin B12 Deficiency in the Diabetes Prevention Program Outcomes Study Metformin 0

Query 2 is structurally the same as query 1, and hence we will only be depicting the portions of the concept map below that are modified from the previous walkthrough. As seen above, the physician received a non-affirmative answer for the representation ( or lack thereof) of African American patients in the long-term Metformin study. Hence, the physician will trigger another query to find patients like Sam in another study (s). Our query 2 provides an overall view of racial representations in the only other study – “10-Year Follow-up of Intensive Glucose Control in Type 2 Diabetes”. We provide a brief walkthrough for our second query below.

Query 2

We are still interested in patients like Sam within the 10-Year follow up study administered medication of Metformin. However in this study, the Metformin patient group is also put on additional ConventionalTherapy (dietary restrictions).

10 year follow up

The 10-Year follow up study is more granular in the representation of races within the Metformin + Conventional Therapy Patient group. We illustrate this granular racial breakdown in the above figure.

Percentage of different races in the study

A physician can see a racial percentage breakdown of races within the Metformin + Conventional Therapy patient group within the 10 Year follow up study as seen above. All percentage breakdown are detailed in results of query 2.

Query 2: Since this study doesn’t call out the population, give me the African American representation from another Metformin study?
Representation 1

  1. prefix dbt: <https://tw.rpi.edu/Courses/Ontologies/2018/DiabetesTreatmentSupport/DiabetesTreatmentSupport/>
  2. prefix resource: <http://semanticscience.org/resource/>
  3. prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
  4. prefix sio: <http://semanticscience.org/resource/>
  5. prefix dct: <http://purl.org/dc/terms/>
  6.  
  7. select ?studyTitle ?medicationDrug ?raceType ?racePercentageVal WHERE {
  8.  
  9. ?study a dbt:ClinicalTrial .
  10. ?study dct:title ?studyTitle .
  11. ?study dbt:hasCohort ?studyCohort .
  12. ?studyCohort dbt:hasPatientGroup ?patientGroup .
  13. ?patientGroup dbt:hasIntervention ?studyIntervention .
  14. ?studyIntervention a dbt:Medication ;
  15. dbt:hasDrug ?medicationDrug .
  16. ?medicationDrug a dbt:AMPKDrug .
  17.  
  18. ?patientGroup dbt:hasCharacteristic ?race .
  19. ?race a ?raceType .
  20. ?raceType rdfs:subClassOf* dbt:Race .
  21. ?race dbt:hasMeasure ?racePercentage .
  22. ?racePercentage sio:SIO_000300 ?racePercentageVal .
  23.  
  24. FILTER (regex(str(?studyTitle), "10-Year Follow-up") ) .
  25.  
  26.  
  27. }

Representation 2
  1. prefix dbt: <https://tw.rpi.edu/Courses/Ontologies/2018/DiabetesTreatmentSupport/DiabetesTreatmentSupport/>
  2. prefix resource: <http://semanticscience.org/resource/>
  3. prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
  4. prefix sio: <http://semanticscience.org/resource/>
  5. prefix dct: <http://purl.org/dc/terms/>
  6.  
  7. select DISTINCT ?studyTitle ?medicationDrug ?raceType ?racePercentageVal WHERE {
  8.  
  9. ?study a dbt:ClinicalTrial .
  10. ?study dct:title ?studyTitle .
  11. ?study dbt:hasCohort ?studyCohort .
  12. ?studyCohort dbt:hasPatientGroup ?patientGroup .
  13. ?patientGroup dbt:hasIntervention ?studyIntervention .
  14. ?studyIntervention a dbt:Medication ;
  15. dbt:hasDrug ?medicationDrug .
  16. ?medicationDrug a dbt:Biguanide .
  17. ?subPatient rdfs:subClassOf ?patientGroup .
  18. ?subPatient sio:SIO_000008 ?attr .
  19. ?attr a dbt:Percentage .
  20. ?attr sio:SIO_000300 ?racePercentageVal .
  21. ?subPatient rdfs:subClassOf ?restriction .
  22. ?restriction a owl:Restriction .
  23. ?restriction owl:someValuesFrom ?raceType .
  24.  
  25. ?raceType rdfs:subClassOf* dbt:Race .
  26.  
  27. FILTER (regex(str(?studyTitle), "10-Year Follow-up") ) .
  28.  
  29.  
  30. }

Result

Study Title Medication Race Type Race Percentage
10-Year Follow-up of Intensive Glucose Control in Type 2 Diabetes Metformin African American 7.8
10-Year Follow-up of Intensive Glucose Control in Type 2 Diabetes Metformin Asian 6.4
10-Year Follow-up of Intensive Glucose Control in Type 2 Diabetes Metformin White 84.8
10-Year Follow-up of Intensive Glucose Control in Type 2 Diabetes Metformin Other Race 1

Overview

We have presented the racial characteristics portion of our patient group studied in a research study, however the characteristics are of different types - {Anthropometric, OrderedLaboratoryTestResult, Demographic}. We have presented a sample competency question for querying attribute values for cohorts, and we present the entire cohort concept map below which makes it possible to run query lookups on other characteristics.

Click here to go back to Demonstration and Queries page

This was developed as a part of the Ontology Engineering course supervised by Prof. Deborah McGuinness and Ms. Elisa Kendall at RPI in Fall'18