Wikidata:Property proposal/medical finding model

medical finding models edit

possible medical findings edit

Originally proposed at Wikidata:Property proposal/Natural science

Descriptionpossible medical findings of a medical conditions
Representsmedical finding (Q639907)
Data typeItem
Domainphysiological condition (Q7189713)
Allowed valuesmedical finding (Q639907)
Example
See alsosymptoms and signs (P780), medical examination (P923)
  •   Done Created as possible medical findings (P5131). ChristianKl17:23, 6 May 2018 (UTC)[reply]
  • suggests the existence of edit

    Originally proposed at Wikidata:Property proposal/Natural science

  •   Done Created as suggests the existence of (P5132). ChristianKl17:24, 6 May 2018 (UTC)[reply]
  • has evaluation edit

    Originally proposed at Wikidata:Property proposal/Natural science

  •   Done Created as has evaluation (P5133). ChristianKl17:24, 6 May 2018 (UTC)[reply]
  • evaluation of edit

    Originally proposed at Wikidata:Property proposal/Natural science


    Motivation

     
    Figure 1
     
    Figure 2
     
    Figure 3

    In diagnostics, w:en:Medical findings are important concepts, and are dealt with in medical records by physicians or other medical staffs. I'm planning to model the relations among disease (Q12136), medical finding (Q639907) and medical attribute (Q44476427) on Wikidata, as shown in figure 1 to 3.

    Currently we have no way of linking between disease (Q12136) and medical finding (Q639907). We can't use symptoms and signs (P780) to link to medical finding (Q639907), because the scope of medical finding (Q639907) is broader than that of symptom (Q169872) (for example, ST elevation (Q7394928) is not a symptom, because it cannot be noticed by a patient). And we can't use medical examination (P923) because the value of medical examination (P923) is not a medical finding (Q639907), but a method to get the medical finding (Q639907). We can't use has cause (P828) nor has effect (P1542) because the causal relationship between disease (Q12136) and medical finding (Q639907) are unspecified.

    It would be preferable if the links between disease (Q12136) and medical finding (Q639907) are bidirectional, because some medical findings are specific to diseases, but others are non-specific; and some diseases present only a couple of characteristic findings, but others present a great variety of findings with a low probability.

    Also, there is no way of linking between medical finding (Q639907) and medical attribute (Q44476427). To avoid confusions, corresponding medical finding (Q639907) and medical attribute (Q44476427) should be linked each other. By doing so, it increases the robustness of the data structure.

    --Okkn (talk) 20:42, 21 April 2018 (UTC)[reply]

      Notified participants of WikiProject Medicine

    Discussion

      Support --Csisc (talk) 21:06, 21 April 2018 (UTC)[reply]

      Support --Geoide (talk) 09:27, 22 April 2018 (UTC)[reply]

    -- What is the source of this data? Gstupp (talk) 14:15, 22 April 2018 (UTC)[reply]

    @Gstupp: Thank you for your comment. Medical findings themselves are annotated in SNOMED CT, so they have UMLS CUI (P2892) (ex. finding of heart rate (Q44484223) → C0577802). The relations between diseases and findings are described in many medical textbooks, although as far as I know there is no opensource database about this. --Okkn (talk) 00:43, 23 April 2018 (UTC) --Okkn (talk) 00:51, 23 April 2018 (UTC)[reply]
    •   Support All David (talk) 15:23, 22 April 2018 (UTC)[reply]
    • Not sure There are multiple models here. I recognize the first one here as a property proposal for "medical finding". The other three look like plans to create items, not properties, for medical finding, and to apply these items as the values of existing properties. Even if we do create a property then it also seems correct to me to create items for individual medical findings because they meet notability criteria and would be useful. If items existed, and also if we established this property, then for heart rate (Q1073121) the property "medical finding" would be → finding of heart rate (Q44484223), which seems redundant. I could be misunderstanding here. Is it correct that there are multiple non-compatible proposals being made here? Blue Rasberry (talk) 20:48, 24 April 2018 (UTC)[reply]
      • @Bluerasberry: Not only the first one, but also other three properties proposed here are relations that can't be express by using the existing properties, and all of them are consistently needed to describe the specific medical findings semantically. I didn't use "medical finding" itself for the property name, because we need to distinguish the two relations "disease (hyperthyroidism) → medical finding (tachycardia)" and "medical attribute (heart rate) → medical finding (finding of heart rate)". Items like finding of heart rate (Q44484223) may seem redundant, but there are structural needs to have superclasses of individual medical findings, such as normal heart rate (Q44483931), tachycardia (Q209583) and bradycardia (Q217111). --Okkn (talk) 00:27, 25 April 2018 (UTC)[reply]

      Support Gtsulab (talk) 17:04, 30 April 2018 (UTC)[reply]