Finding precise causal multi-drug-drug interactions for adverse drug reaction analysis

Patent metadata

Field	Value
Publication number	US-11211169-B2
Application number	US-201816176022-A
Country	US
Kind code	B2
Filing date	Oct 31, 2018
Priority date	Mar 6, 2018
Publication date	Dec 28, 2021
Grant date	Dec 28, 2021

Abstract

Mechanisms are provided for implementing a framework to learn multiple drug-adverse drug reaction associations. The mechanisms receive and analyze patient electronic medical record data and adverse drug reaction data to identify co-occurrences of references to drugs with references to adverse drug reactions (ADRs) to thereby generate candidate rules specifying multiple drug-ADR relationships. The mechanisms filter the candidate rules to remove a subset of one or more rules having confounder drugs specified in the subset of one or more candidate rules, and thereby generate a filtered set of candidate rules. The mechanisms further generate a causal model based on the filtered set of candidate rules. The causal model comprises, for each ADR in a set of ADRs, a corresponding set of one or more rules, each rule specifying a combination of drugs having a causal relationship with the ADR.

First claim

What is claimed is:

1. A method, in a data processing system comprising at least one processor and at least one memory, the at least one memory comprising instructions executed by the at least one processor to cause the at least one processor to implement a framework for learning multiple drug-adverse drug reaction associations and predicting patient adverse drug reactions (ADRs), the method comprising:
executing, by a co-occurrence logic module of the framework, computer natural language processing logic on real world evidence comprising patient electronic medical record (EMR) data and ADR data, to extract features, from natural language content of the real world evidence, specifying co-occurrences of first features referencing drugs with second features referencing adverse drug reactions;
generating, by the co-occurrence logic module, one or more drug-ADR rules, based on the extracted first features and second features, specifying multiple drug-ADR relationships, wherein each of the one or more drug-ADR rules specifies a corresponding drug pattern comprising a plurality of drugs and corresponding adverse drug reaction;
filtering, by a confounder filter logic module of the framework, the one or more drug-ADR rules to remove a subset of one or more drug-ADR rules having confounder drugs specified in the subset of drug-ADR rules, and thereby generate a filtered set of drug-ADR rules wherein filtering the one or more drug-ADR rules further comprises:
calculating, for a first drug-ADR rule of the one or more drug-ADR rules, an improvement metric specifying an amount of improvement of a corresponding association score for the first drug-ADR rule over an association score for another second drug-ADR rule specifying a sub-pattern of the corresponding drug pattern of the first drug-ADR rule and the corresponding adverse drug reaction of the first drug-ADR rule; and

determining, for the first drug-ADR rule, whether to maintain the first drug-ADR rule or remove the first drug-ADR rule based on a value of the improvement metric;
generating, by a causal association logic module of the framework, a computer executed causal model based on the filtered set of drug-ADR rules, wherein the computer executed causal model comprises, for each ADR in a set of ADRs, a corresponding set of one or more computer executed rules, each computer executed rule specifying one or more drugs having a causal relationship with the ADR which is executed on other patient EMR data, wherein generating the causal model based on the filtered set of drug-ADR rules further comprises, for each ADR:
generating a directed acyclic graph data structure in a Bayesian learning framework, wherein the directed acyclic graph data structure represents each drug pattern of each of the drug-ADR rules, in the filtered set of drug-ADR rules associated with the ADR, as a variable of the directed acyclic graph data structure;
determining transition probabilities for transitions between variables specified in the directed acyclic graph data structure;
determining causal relationships among one or more drug patterns, represented by the variables, and the ADR based on the directed acyclic graph data structure and the transition probabilities; and
generating, for the ADR, the corresponding set of one or more rules based on the determined causal relationships;

evaluating other patient EMR data by applying the computer executed causal model to drug history data present in the other patient EMR data to identify probabilities of a patient encountering one or more ADRs in a set of ADRs; and
generating, for the patient, a patient model based on the identified probabilities of the patient encountering one or more ADRs in the set of ADRs.

2. The method of claim 1 , wherein executing computer natural language processing logic on the real world evidence comprises:
generating the one or more drug-ADR rules based on the identified co-occurrences, wherein each drug-ADR rule specifies a drug pattern identifying a plurality of drugs and a corresponding ADR; and
selecting a sub-set of the one or more drug-ADR rules as a basis for generating the computer executed causal model based on at least one of a support metric or a confidence metric, wherein the support metric measures, for an associated drug-ADR rule, a number of instances of the co-occurrence in the real world evidence, and wherein the confidence metric measures, for an associated drug-ADR rule, a probability of the associated ADR given the drugs in the drug pattern of the associated drug-ADR rule.

3. The method of claim 2 , wherein identifying all co-occurrences of references to drugs with reference to ADRs in the real world evidence comprises:
performing natural language processing of the real world evidence to identify at least one of terms, phrases, or medical codes identifying references to drugs and references to ADRs;
evaluating relative distances within the real world evidence, between each identified term, phrase or medical code identifying references to drugs and references to ADRs; and
identifying co-occurrences based on the relative distances.

4. The method of claim 2 , wherein selecting a sub-set of the one or more drug-ADR rules comprises, for each drug-ADR rule:
generating a contingency table data structure, where each entry in the contingency table data structure comprises a number of patient electronic medical records that satisfy a condition of the row and column of the contingency table data structure corresponding to the entry, wherein:
a first row of the contingency table data structure corresponds to patient electronic medical records that contain all drugs in the drug pattern of the drug-ADR rule,
a second row of the contingency table data structure corresponds to patient electronic medical records that contain none of the drugs in the drug pattern of the drug-ADR rule,
a first column of the contingency table data structure corresponds to the patient electronic medical records that contain the ADR in the drug-ADR rule, and
a second column of the contingency table data structure corresponds to the patient electronic medical records that do not contain the ADR in the drug-ADR rule.

5. The method of claim 1 , further comprising:
identifying one or more confounder drugs present in the subset of one or more drug-ADR rules having confounder drugs; and
updating a confounder drug resource information data source based on the identified one or more confounder drugs.

6. The method of claim 1 , further comprising:
configuring a cognitive computing system to modify logic of the cognitive computing system to implement the computer executed causal model; and
executing, by the cognitive computing system, an automated treatment recommendation operation that provides a treatment recommendation output to a computing device associated with a medical practitioner based on the patient model, wherein determining, for the first drug-ADR rule, whether to maintain the first drug-ADR rule or remove the first drug-ADR rule based on a value of the improvement metric further comprises:
comparing the improvement metric corresponding to the first drug-ADR rule to an improvement metric threshold value;
in response to the improvement metric corresponding to the first drug-ADR rule not being equal to or greater than the improvement metric threshold value, determining that a confounder drug is present in the corresponding drug pattern of the first drug-ADR rule; and
identifying the confounder drug in the corresponding drug pattern based on a difference between the corresponding drug pattern and the sub-pattern.

7. A computer program product comprising a computer readable storage medium having a computer readable program stored therein, wherein the computer readable program, when

Assignees

• IBM

Inventors

• Dey Sanjoy
• Ghalwash Mohamed
• Zhang Ping

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G16H50/70 Primary

  for mining of medical data, e.g. analysing previous cases of other patients · CPC title

• G06N5/025

  Extracting rules from data · CPC title

• G16H70/40 Primary

  relating to drugs, e.g. their side effects or intended usage · CPC title