Algorithm, data pipeline, and method to detect inaccuracies in comorbidity documentation

Abstract

Systems and methods for detecting inaccuracies in comorbidity documentation are provided. An example method can include retrieving patient data from a hospital records database, generating a first data structure representing a list of patient identifiers, retrieving patient symptom data for each patient identifier listed in the first data structure, generating, a second data structure by assigning the retrieved patient symptom data to the eligible patients represented in the first data structure, retrieving, a plurality of lookup tables, generating a third data structure by applying the lookup tables to the second data structure to assign a CC or a MCC value to at least one unique patient identifier of the second data structure, and storing the third data structure to the memory. In some implementations, the method can include updating the hospital records database with the CC or the MCC value assigned to the at least one unique patient.

1 . A system for detecting inaccuracies in comorbidity documentation, comprising a memory and a processor configured to:
 retrieve patient data from a hospital records database;   generate a first data structure representing a list of patient identifiers, each corresponding to an eligible patient having a hospital admission during a specified time period;   retrieve patient symptom data for each patient identifier listed in the first data structure from one or more of a professional records database, a hospital records database, or a clinical records database;   generate a second data structure by assigning the retrieved patient symptom data to the eligible patients represented in the first data structure;   retrieve, from the memory, a plurality of lookup tables, the plurality of lookup tables comprising:
 a first table, comprising diagnostic codes assigned to chronic condition categories, 
 a second table, comprising a list of clinical discrete data corresponding to unambiguous indicators of each chronic condition, and 
 a third table, comprising diagnostic codes assigned to complication or comorbidity (CC) or major complication or comorbidity (MCC) values for each chronic condition category; 
   generate a third data structure by applying the lookup tables to the second data structure to assign a CC or a MCC value to at least one unique patient identifier of the second data structure; and   store the third data structure to the memory.   
     
     
         2 . The system of  claim 1 , wherein the data processing system generates the first data structure by:
 retrieving a set of criteria for identifying hospitalized patients in the patient data;   applying the set of criteria to the patient data to identify hospitalized patients; and   performing, using the retrieved set of criteria, an extract, transform, load process on the patient data to populate the first data structure with the identified hospitalized patients.   
     
     
         3 . The system of  claim 2 , wherein the set of criteria comprises a lookup table specifying encounter types and encounter locations indicative of a patient hospitalization. 
     
     
         4 . The system of  claim 1 , wherein the second data structure includes all patient symptom data for a predetermined period of time up to and including the current admission. 
     
     
         5 . The system of  claim 4 , wherein the second data structure includes an attribute column indicating whether each patient symptom in the patient symptom data was retrieved from the professional records database, the hospital records database, or the clinical records database. 
     
     
         6 . The system of  claim 1 , wherein the diagnostic codes comprise ICD-9 and ICD-10 codes. 
     
     
         7 . The system of  claim 1 , wherein the chronic condition categories comprise cancer, major depression, epilepsy, ischemic heart disease, osteoporosis, chronic obstructive pulmonary disease, osteoarthritis, dementia, cerebrovascular disease, asthma, bipolar disorder, high cholesterol, obesity, malnutrition, hypertension, chronic kidney disease, congestive heart failure, diabetes mellitus, hyponatremia, hypernatremia, and acute renal failure. 
     
     
         8 . The system of  claim 1 , wherein the list of clinical discrete data comprises one or more of lab results, vital signs, or medication data that unambiguously indicate the presence of a medical diagnosis. 
     
     
         9 . The system of  claim 1 , wherein the third data structure is an entity attribute table comprising columns for unique patient identifiers, conditions, CC or an MCC values, and classifications of whether each condition was present in one or more of the professional records database, the hospital records database, or the clinical records database. 
     
     
         10 . The system of  claim 1 , wherein the data processing system is configured to: update the hospital records database with an indication of the CC or the MCC value assigned to the at least one unique patient. 
     
     
         11 . A method, executed by a data processing system having a memory and a processor, of detecting inaccuracies in comorbidity documentation, comprising:
 retrieving, by the data processing system, patient data from a hospital records database;   generating, by the processor, a first data structure representing a list of patient identifiers, each corresponding to an eligible patient having a hospital admission during a specified time period;   retrieving patient symptom data for each patient identifier listed in the first data structure from one or more of a professional records database, a hospital records database, or a clinical records database;   generating, by the processor, a second data structure by assigning the retrieved patient symptom data to the eligible patients represented in the first data structure;   retrieving, from the memory, a plurality of lookup tables, the plurality of lookup tables comprising:
 a first table, comprising diagnostic codes assigned to chronic condition categories, 
 a second table, comprising a list of clinical discrete data corresponding to unambiguous indicators of each chronic condition, and 
 a third table, comprising diagnostic codes assigned to complication or comorbidity (CC) or major complication or comorbidity (MCC) values for each chronic condition category; 
   generating, by the processor, a third data structure by applying the lookup tables to the second data structure to assign a CC or a MCC value to at least one unique patient identifier of the second data structure; and   storing the third data structure to the memory.   
     
     
         12 . The method of  claim 11 , comprising:
 generating the first data structure by:
 retrieving a set of criteria for identifying hospitalized patients in the patient data; 
 applying the set of criteria to the patient data to identify hospitalized patients; and 
 performing, using the retrieved set of criteria, an extract, transform, load process on the patient data to populate the first data structure with the identified hospitalized patients. 
   
     
     
         13 . The method of  claim 12 , wherein the set of criteria comprises a lookup table specifying encounter types and encounter locations indicative of a patient hospitalization. 
     
     
         14 . The method of  claim 11 , wherein the second data structure includes all patient symptom data for a predetermined period of time up to and including the current admission. 
     
     
         15 . The method of  claim 14 , wherein the second data structure includes an attribute column indicating whether each patient symptom in the patient symptom data was retrieved from the professional records database, the hospital records database, or the clinical records database. 
     
     
         16 . The method of  claim 11 , wherein the diagnostic codes comprise ICD-9 and ICD-10 codes. 
     
     
         17 . The method of  claim 11 , wherein the chronic condition categories comprise cancer, major depression, epilepsy, ischemic heart disease, osteoporosis, chronic obstructive pulmonary disease, osteoarthritis, dementia, cerebrovascular disease, asthma, bipolar disorder, high cholesterol, obesity, malnutrition, hypertension, chronic kidney disease, congestive heart failure, diabetes mellitus, hyponatremia, hypernatremia, and acute renal failure. 
     
     
         18 . The method of  claim 11 , wherein the list of clinical discrete data comprises one or more of lab results, vital signs, or medication data that unambiguously indicate the presence of a medical diagnosis. 
     
     
         19 . The method of  claim 11 , wherein the third data structure is an entity attribute table comprising columns for unique patient identifiers, conditions, CC or an MCC values, and classifications of whether each condition was present in one or more of the professional records database, the hospital records database, or the clinical records database. 
     
     
         20 . The method of  claim 11 , comprising:
 updating the hospital records database with an indication of the CC or the MCC value assigned to the at least one unique patient.