eNS, EHR-based Noonan syndrome detection using deep learning.

Enter the diagnosis description

  1. Select "Query" from the top navigation bar
  2. Enter the diagnosis description text, one diagnosis text per line
  3. Click Submit

Statistics of the current model:

  • Number of patients with Noonan syndrome in training: 162
  • Number of presumable controls in training: 16200
  • Number of words in the description text vocabulary: 15725
  • Last update: 9/16/2021




Enter the diagnosis description texts


Method

Method: convnet model

convnet model

  • Diagnosis description texts were sequentially added together to form a single string of descriptions for each patient
  • A tokenizer was used to convert every word in the description text vocabulary to a unique integer representation
  • Each string was then converted into a sequence of integers and the first 1000 integers were used as the input features for the patient
  • All tokenized sequences were right-padded or truncated to be 1000 integers long