OCL Searches Gone Wrong

We’re going to use this thread as a place to document examples of where OCL’s search fell flat. If you search for a concept, a source, etc. and you didn’t see what you know is there in OCL, please put it here!

MSF’s Antenatal Convenience set did not appear in Global Search: https://app.openconceptlab.org/#/search/?q=antenatal%20convset

Expected concept: https://app.openconceptlab.org/#/orgs/MSFOCP/sources/MSFOCP/concepts/3080986/

Per Andy Kanter:

I just search for spo2 (SpO2) and it returned oxygen saturation for which it is a short name.

Per Laure: It’s like search is not considering short names. I was looking for FIM (functional independence measure) which doesn’t exist yet and the results were surprising. Same if you look for a simple short name like Hb (for haemoglobin) it won’t propose you Haemoglobin. Same with synonyms

Per Ellen: Looking for TSQ (Trauma screening questionnaire) which doesn’t exist (yet) in CIEL, OCL shows 8 search results but I have no idea why search matched with them. Try it.

I had an example where searching ‘HTS’ was returning results like “Thoughts” and “Nights”

Searching “vmmc_circ” should put concept “VMMC_CIRC” up at the top of the results, but "VMMC_CIRC_NAT is being prioritized.

Searching CIEL for weight (looking for “Weight (kgs)”) is disappointing. Finally found what I wanted on the 3rd page of results (almost at the very end of results).

In this case, it’s not an exact match of a name, but it exactly matches the beginning of the concept name, which I would expect to be ranked higher than matching a word in the middle of a concept name.

Exact match somehow was returning more search results than not using exact match? Search string was “lab test”.

Without exact match: 172 concepts returned

With exact match: 2463 concepts returned

The new OCL Search is live now, with many improvements and new features. A longer write-up is coming along soon.

1 Like

Nice job! Results are more accurate and visibly faster. Thanks!

Slight issue that has been discussed but not formally documented (at least not that I could find): Having visually apparent formatting for search results based on the method. For example, in the image below, the bolded results are the matching term. However, the fourth result is also a fuzzy match. It may confuse a user who doesn’t realize that the fourth term is a fuzzy match. Perhaps this could be shown by italicizing and bolding the word, for example.

It would be particularly helpful if we could list out all the potential visually distinct results:

  • Matched term = Bold
  • Fuzzy Match = Italicized?
  • Matched by a mapping - Has an icon (Currently)
    • Note: there is no visual distinction between Same-as and non-same-as mappings

FWIW, here are examples from ChatGPT-4 of similar terms:

  1. Artery vs. Artery: An “artery” is a blood vessel that carries oxygen-rich blood away from the heart. An “articular” relates to joints in the body.
  2. Dysphagia vs. Dysphasia: “Dysphagia” refers to difficulty swallowing, while “dysphasia” means difficulty speaking or understanding language due to a brain injury.
  3. Hypertension vs. Hyperextension: “Hypertension” is high blood pressure. “Hyperextension” is the excessive straightening of a body part, usually a joint.
  4. Myeloma vs. Melanoma: “Myeloma” is a cancer of plasma cells in the bone marrow. “Melanoma” is a serious skin cancer arising from pigment-producing cells.
  5. Atherosclerosis vs. Arteriosclerosis: “Atherosclerosis” is the buildup of fats and cholesterol in the arteries. “Arteriosclerosis” is a broader term that means hardening of the arteries, which can be due to atherosclerosis but can also have other causes.
  6. Prostate vs. Prostrate: The “prostate” is a gland in the male reproductive system. To “prostrate” oneself means to lie face down on the ground, often in reverence or submission.
  7. Gastritis vs. Gastric: “Gastritis” refers to inflammation of the stomach lining. “Gastric” simply refers to the stomach.
  8. Palpation vs. Palpitation: “Palpation” is a method used by medical professionals to feel the body with the hands to identify abnormalities. “Palpitation” refers to a noticeably rapid, strong, or irregular heartbeat.
  9. Lesion vs. Lession: A “lesion” refers to a broad term for any abnormal area in or on the body tissue, often caused by disease or trauma. “Lession” isn’t a standard medical term.
  10. Ileum vs. Ilium: “Ileum” is the third and final part of the small intestine. “Ilium” is the uppermost and largest bone of the pelvis.
  11. Celiac vs. Cilia: “Celiac” refers to a disease in which people can’t eat gluten because it harms their small intestine. “Cilia” are tiny, hair-like structures present on the surface of all mammalian cells.
  12. Complement vs. Compliment: In immunology, “complement” is a system of proteins that enhances the ability of antibodies and phagocytic cells to clear pathogens from an organism. “Compliment,” though not a medical term, means a polite expression of praise or admiration.
  13. Peroneal vs. Perineal: “Peroneal” relates to the fibula or the outer part of the lower leg. “Perineal” relates to the perineum, which is the area between the anus and the genitals.

Duration Units should be the first result because of how close it is, but it ends up in 6th place. Note that this is NOT a huge issue - just a suggestion to optimize if possible. People were generally happy to see it on the list at all. :slight_smile:

1 Like

@jamlung I can see on Staging and on QA “1732 - Duration Units” is the first result.
Will check whats wrong with Production.

Issue with “/” in search - it appears to not be bringing up an expected concept, even when the “/” is part of the concept’s synonym. Example search: https://app.openconceptlab.org/#/search/?q=tdf/ftc does not yield the expected CIEL concept 164854 (which has a synonym “TDF/FTC/NVP”). I instead had to search with a space character instead of a “/” i.e. https://app.openconceptlab.org/#/search/?q=tdf+ftc

Just searched for ALS and did not get the CIEL concept for ALS (amyotrophic lateral sclerosis) in the first page. I think if you have an exact match on the entire first word of a description/term it should come to the top, an exact match at the very top, of course.

Adding a screenshot of the results, and noting that this is the concept that we expected to see: https://app.openconceptlab.org/#/orgs/CIEL/sources/CIEL/concepts/148917/

Searching “Head CT” should return concept 846 at the top since its synonym is an exact match, but it doesn’t seem to be appearing in the search results. OCL appears to be prioritizing concepts whose display name matches one of the words, rather than boosting a concept with a synonym that matches both terms exactly.

cc: @askanter

Top match (846) because he is searching CIEL concepts only.