Browse Prior Art Database

Cognitive System and Method of Learning Medical Indicators Based on Physician Preferences in a Cognitive Medical Application

IP.com Disclosure Number: IPCOM000248422D
Publication Date: 2016-Nov-25
Document File: 5 page(s) / 115K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed are a system and mechanism to gather a practitioner’s preferences for deviations from the set of global rules in a health-based cognitive computing and analytics engine in order to provide a customized experience for the practitioner.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 30% of the total text.

Page 01 of 5

Cognitive System and Method of Learning Medical Indicators Based on Physician Preferences in a Cognitive Medical Application

Many medical applications try to provide a physician with only the clinical information relevant to a given patient. This is generally accomplished by looking for condition indicators that are globally defined within the application. For example, a "low" Spirometry result is an indicator of asthma or chronic obstructive pulmonary disease (COPD). A pulmonologist trying to diagnose a lung condition receives the results of this

Spirometry test.

This method, although effective, does not give the practitioner a way to determine what "low" means. Lab results usually come with a noted normal range, and an abnormal test result outside the normal range is considered high or low, etc. For many conditions that is acceptable; however in this example, when trying to determine if a patient has

asthma or COPD, a physician needs to know how much lower than normal the result is.

A low result could indicate asthma or emphysema, while a substantially low result

indicates COPD. Different practitioners look for different levels when trying to diagnose a given condition.

The objective is to provide a mechanism via a user interface for doctors and specialists to enter preferences for the display of condition indicators from a patient's electronic medical record (EMR).

The novel contribution is a system and mechanism to gather the practitioner's preferences for deviations from the set of global rules in order to provide a customized experience for the practitioner.

This disclosure relies on the existence of a set of global condition indicator rules as defined by subject matter experts (SMEs). A machine-learning model, trained using the Unified Medical Language System (UMLS), maps various medical concepts to various conditions and body systems. The system can use a another machine learning (ML) model to determine the values at which a medical concept becomes an indicator (existing art). The rules engines discussed herein are not necessarily a set of hard and fast rules, but a combination of a rules engine and trained models, etc.

This mechanism operates as follows:

1. Consultations between the practitioner and patients are recorded/transcribed 2. The resulting transcriptions are processed to map interactions between the doctor and patient to one or more of the global rules, providing a set of baseline preferences 3. The medical application interface displaying the condition indicators to the doctor is modified to conform to the users preferences 4. The interface provides a set of modifiers for the practitioner to alter the associated preferences, which feed back into the preference engine

5. The indicator preferences for practitioners are combined based on specialty/diagnosis to derive a baseline set of preferences for practitioners in a given field of medicine

6. This combined baseline is used to provide a more accurate list of co...