library/dupscore.inc.php

   1 <?php
   2
   3 /**
   4  * Note this may be included by CLI scripts, so don't do anything web-specific here!
   5  *
   6  * @package   OpenEMR
   7  * @link      http://www.open-emr.org
   8  * @author    Rod Roark <rod@sunsetsystems.com>
   9  * @copyright Copyright (c) 2021 Rod Roark <rod@sunsetsystems.com>
  10  * @license   https://github.com/openemr/openemr/blob/master/LICENSE GNU General Public License 3
  11  */
  12
  13 // The SQL returned by this function is an expression that computes the duplication
  14 // score between two patient_data table rows p1 and p2.
  15 //
  16 function getDupScoreSQL()
  17 {
  18     return
  19     //  5 First name
  20     "5 * (SOUNDEX(p1.fname) = SOUNDEX(p2.fname)) + " .
  21     //  3 Last name
  22     "3 * (SOUNDEX(p1.lname) = SOUNDEX(p2.lname)) + " .
  23     //  4 Any phone number
  24     "4 * (" .
  25       "(TRIM(p1.phone_home) != '' AND ( " .
  26       "REPLACE(REPLACE(p1.phone_home, '-', ''), ' ', '') IN ( " .
  27       "REPLACE(REPLACE(p2.phone_home, '-', ''), ' ', ''), " .
  28       "REPLACE(REPLACE(p2.phone_biz , '-', ''), ' ', ''), " .
  29       "REPLACE(REPLACE(p2.phone_cell, '-', ''), ' ', '')))) " .
  30     "OR (TRIM(p1.phone_biz) != '' AND ( " .
  31       "REPLACE(REPLACE(p1.phone_biz , '-', ''), ' ', '') IN ( " .
  32       "REPLACE(REPLACE(p2.phone_biz , '-', ''), ' ', ''), " .
  33       "REPLACE(REPLACE(p2.phone_cell, '-', ''), ' ', '')))) " .
  34     "OR (TRIM(p1.phone_cell) != '' AND ( " .
  35       "REPLACE(REPLACE(p1.phone_cell, '-', ''), ' ', '') = " .
  36       "REPLACE(REPLACE(p2.phone_cell, '-', ''), ' ', ''))) " .
  37     ") + " .
  38     //  6 Birth date
  39     "6 * (p1.DOB IS NOT NULL AND p2.DOB IS NOT NULL AND p1.DOB = p2.DOB) + " .
  40     //  7 Email
  41     "7 * (TRIM(p1.email) != '' AND TRIM(p1.email) = TRIM(p2.email)) + " .
  42     // 15 Government ID
  43     "15 * (TRIM(p1.ss) != '' AND TRIM(p1.ss) = TRIM(p2.ss))";
  44 }