Browse Prior Art Database

A Method to do Multi Model Analysis of a Patient Context for Efficient Medical Diagnosis

IP.com Disclosure Number: IPCOM000246149D
Publication Date: 2016-May-12

Publishing Venue

The IP.com Prior Art Database

Abstract

The process of diagnosis of a patient involves analysis of several factors such as symptoms and external attributes known to the doctor at the point of diagnosis. These are • Symptoms • Medical History of the patient • Allergy and Immunization details of the patient • Patient geo location and Epidemic around that location • Current seasonal diseases • Social/Family background of patient • Patient life style, diet and habitual details. Problem Statement Mostly during diagnosis, some of these factors are ignored or not even collected during diagnosis because at present, we do not have any existing system which can process/cleanse these attributes, extract some relevant information using NLP techniques, and finally help the doctor to diagnose the disease by trying to map the available attributes to possible diseases.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 32% of the total text.

Page 01 of 11

A Method to do Multi Model Analysis of a Patient Context for Efficient Medical Diagnosis

The process of diagnosis of a patient involves analysis of several factors such as symptoms and external attributes known to the doctor at the point of diagnosis. These are

• Symptoms • Medical History of the patient • Allergy and Immunization details of the patient • Patient geo location and Epidemic around that location • Current seasonal diseases • Social/Family background of patient
• Patient life style, diet and habitual details.

Problem Statement


Mostly during diagnosis, some of these factors are ignored or not even collected during diagnosis because at present, we do not have any existing system which can process/cleanse these attributes, extract some relevant information using NLP techniques, and finally help the doctor to diagnose the disease by trying to map the available attributes to possible diseases.

Summary

I. Collect symptoms and external attributes.

Gather medical history of the patient along with allergy/immunization details. Process involves collecting the symptoms from the patient.

Collect geo location of the patient along with other available personal details. Collect other patient information of the patient such as Social/Family background, Life style, Diet, Habitual details.

II. Data Cleansing / Improvement


Collected symptoms are cleansed with a dictionary of similar symptoms using relationship extraction techniques.

Find the current seasonal diseases/epidemic around the geo location of the patient and the corresponding possible symptoms and cleanse patient provided symptoms with that.

Cleansing dictionary is applied with knowledge on the Geo/Social background of

1


Page 02 of 11

the patient.

Ex: If the patient comes from a location where 'dengue fever' is common, for a symptom of 'body pain', suggestion can be 'muscle and joint pains'. For a symptom of 'measles', suggestion can be as simple as 'rashes'.

III. Diagnosis


The diagnosis must include diet, life style changes and the diagnosis must be made more towards getting the root cause for such patients and this may differ between every patient coming from different background, no two patients suffering from same symptoms necessarily have to have same disease - this is in a way helping us to think different perspective of diagnosis for similar symptoms and to avoid errors in diagnosis.

Cleansed Symptoms, Geo/Social Background, Diet, Medical History etc. can be considered as features for diagnosis of the disease using classification algorithms giving appropriate weight to individual features. This will result in a set of possible diseases which can be diagnosed from the given details.

Difference

The present article is building a diagnosis decision support system based on analysis of data captured from a variety of sources representing the context that is internal to the patient, viz. specific disease, patient history, life style diseases, and external to the...