Early cTAKES Work

C_ClinicalDocumentProcessing.class.php

<?php

class C_ClinicalDocumentProcessing extends Controller {
    var $test_data = array(
        // Patient ID 1
        "1" => array(
            // A new patient note ID 32
            "encounter_note_32" => array(
                "RXNORM" => array(
                    array(
                        "text" => "Clotrimazole",
                        "code" => "2623"
                    ),
                    array(
                        "text" => "Coumadin",
                        "code" => "202421"
                    )
                ),
                "SNOMEDCT_US" => array(
                    array(
                        "text" => "Calcium Channel Blockers",
                        "code" => "48698004"
                    )
                )
            ),
            // A new patient note ID 43
            "encounter_id_43" => array(
                "RXNORM" => array(
                    array(
                        "text" => "Coumadin",
                        "code" => "202421"
                    )
                ),
                "SNOMEDCT_US" => array()
            )
        ),
        // Patient ID 439
        "439" => array(
            // A new patient note ID 554
            "encounter_id_554" => array(
                "RXNORM" => array(
                    array(
                        "text" => "Coumadin",
                        "code" => "202421"
                    )
                ),
                "SNOMEDCT_US" => array(
                    array(
                        "text" => "Clotrimazole",
                        "code" => "387325003"
                    ),
                    array(
                        "text" => "Calcium Channel Blockers",
                        "code" => "48698004"
                    ),
                    array(
                        "text" => "Calcium",
                        "code" => "5540006"
                    )
                )
            )
        )
    );

    function __construct()
    {
        parent::__construct();
    }

    function default_action()
    {
        $this->check_action();
    }

    function check_action()
    {
        return \common\http\HttpResponseHelper::send(200, array("status" => "OK"), 'JSON');
    }

    function get_action($pid)
    {
        if (!in_array($pid, array("1", "439")) || empty($pid)) {
            return \common\http\HttpResponseHelper::send(404, array("status" => "Not Found"), 'JSON');
        }

        // TODO: Call to a service that does the MySQL query here

        $data = $this->test_data[$pid];
        return \common\http\HttpResponseHelper::send(200, array("status" => $data), 'JSON');
    }

    function post_action($pid, $encounter_id, $content="")
    {
        if (!in_array($pid, array("1", "439")) || empty($pid)) {
            return \common\http\HttpResponseHelper::send(404, array("status" => "Not Found"), 'JSON');
        }

        if (empty($encounter_id)) {
            // TODO: Use real UUID library
            $encounter_id = sprintf('%05d-%05d-%05d-%05d',
                mt_rand(0, 99999),
                mt_rand(0, 99999),
                mt_rand(0, 99999),
                mt_rand(0, 99999)
            );
        }

        // TODO: Send to cTAKES here

        return \common\http\HttpResponseHelper::send(201, array("status" => "Queued for Processing"), 'JSON');
    }

    function put_action($pid, $encounter_id)
    {
        // TODO: Logic here that marks a note as "read" so that it isn't returned in the get action...

        return \common\http\HttpResponseHelper::send(200, array("status" => "OK"), 'JSON');
    }
}

?>

parser.py

import json
import xmltodict
import sys

def convert_xml_to_dict(file):
  file = open(file, 'r')
  data = xmltodict.parse(file.read())
  file.close()
  return data

def extract_codes(data):
  raw_codes = data["xmi:XMI"]["refsem:UmlsConcept"]
  formatted_codes = dict()

  for code in raw_codes:
    if not formatted_codes.has_key(code['@codingScheme']):
      formatted_codes[code['@codingScheme']] = []

    nullable_code_name = 'unknown'
    if code.has_key('@code'):
      nullable_code_name = code['@code']

    formatted_codes[code['@codingScheme']].append({
      'text': code['@preferredText'],
      'code': nullable_code_name
    })

  return formatted_codes

def write_result_to_json_file(file, data):
  file = open(file, 'w')
  file.write(json.dumps(data))
  file.close()

file_without_extension = sys.argv[1].rsplit('.', 1)[0]
data = convert_xml_to_dict(file_without_extension + '.xml')
codes = extract_codes(data)
write_result_to_json_file(file_without_extension + '_grouped_codes.json', codes)

requirements.txt

xmltodict==0.11.0

http://localhost/openemr/controller.php?ClinicalDocumentProcessing&get&pid=1, for example :)