You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, I forked your code and wrapped it with a FHIR API. It seems to me that the predictions made by your model are not accurate. Please consider the attached screenshots. In one case of Breast Cancer it gets close, but recommends the code for right breast when unspecified. SO that code is wrong.
The second diabetes was a simple statement and although it produced a Diabetes code, it was for type II when the note said type I.
The text was updated successfully, but these errors were encountered:
Further analysis shows that the less surrounding text to the ICD-10 code, the less accurate the model is:
Example:
"Diabetes type II with complications", produces a correct result.
"Discharge Summary: Diabetes type II with complications", does not produce the correct code.
Hi, I forked your code and wrapped it with a FHIR API. It seems to me that the predictions made by your model are not accurate. Please consider the attached screenshots. In one case of Breast Cancer it gets close, but recommends the code for right breast when unspecified. SO that code is wrong.
The second diabetes was a simple statement and although it produced a Diabetes code, it was for type II when the note said type I.
The text was updated successfully, but these errors were encountered: