zimolzak / spep-nlp

Natural language classification of serum protein electrophoresis reports

SPEP classifier

Natural language processing of serum protein electrophoresis reports, using support vector machine and rules-based approaches.

PDF: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7477876/pdf/CCI.19.00167.pdf

PMC: https://www.ncbi.nlm.nih.gov/pmc/articles/pmid/32813561/

PubMed: https://pubmed.ncbi.nlm.nih.gov/32813561/

Journal: https://www.doi.org/10.1200/CCI.19.00167

Ryu JH, Zimolzak AJ. Natural Language Processing of Serum Protein Electrophoresis Reports in the Veterans Affairs Health Care System. JCO Clin Cancer Inform. 2020 Aug;4:749-756. doi: 10.1200/CCI.19.00167.

Code is public domain, because created as work for hire by an employee of US Federal Government.

Details about iterations of rules-based

First attempt

[1] 0.82
   yhat
y     0   1
  0 126  41
  1  13 120

Second

   yhat
y     0   1
  0 166   1
  1  14 119
[1] 0.95

Third

   yhat
y     0   1
  0 166   1
  1   9 124
[1] 0.9666667

Fourth

   yhat
y     0   1
  0 166   1
  1   5 128
[1] 0.98

After fixing 3 annotation errors

   yhat
y     0   1
  0 169   1
  1   2 128

[1] 0.99