MVMLima / Classification-algorithm-of-Congenital-Zika-Syndrome-characterizations-diagnosis-and-validation

Classification algorithm of Congenital Zika Syndrome: characterizations, diagnosis and validation

Classification algorithm for Congenital Zika Syndrome: characterizations, diagnosis and validation

Table of contents

• General info
• Compilation
• Input data
• Files
• How to build the classifier
• How to run the classifier
• References

General info

In this directory, we present the code to construct a Classification algorithm for Congenital Zika Syndrome (CZS)

The code originated from the project described in the paper [1]. The purpose of this algorithm is to be able to identify cases of CZS automatically based on electronic records, serogy and imaging reports.

Compilation

We performed our statistical analysis using Python version 3.6.5. We computer classification evaluation using Jupyter Notebook. Necessary packages:

• scikit-learn
• matplotlib
• spacy

Input data

we use the Brazilian electronic records RESP (Public Health Event Record) linked with SINASC (Live Birth Information System). We use results obtained by the classification performed by a group of specialists described in the article [2] for training.

Data from cranial circonferencing standards according to gender and week of development was obtained from InterGrowth standards

Files

• circumference.csv : Data from cranial circonferencing standards.
• gerateData_RuleClass.py : this script organize database and applied rule classification.
• NLP.py : code for performing natural language processing
• classifier_1.py: this script create a classify for group 1
• classifier_2.py: this script create a classify for group 2
• runClass1.py: this script performs group 1 classification
• runClass2.py: this script performs group 2 classification
• words.dat: this file contains words used to classify texts
• classific1.dat : this file is the classify algorithm for Group 1
• classific2.dat : this file is the classify algorithm for Group 2
• evaluation.ipynb : this script contains the analyses performed for the article

How to build the classifier

The data for construction of the classifiers could not be made available for ethical reasons. It is ultilized data of examinations of real people. Must run the scripts in the following order: 1 gerateData_RuleClass.py 2 classifier_1.py 3 classifier_2.py

How to run the classifier

Must run the scripts in the following order: 1 gerateData_RuleClass.py 2 runClass1.py.py 3 runClass2.py.py

References

[1] Classification algorithm for Congenital Zika Syndrome: characterizations, diagnosis and validation. (submited to Scientific Reports)

[2] França, Giovanny VA, et al. "Congenital Zika virus syndrome in Brazil: a case series of the first 1501 livebirths with complete investigation." The lancet 388.10047 (2016): 891-897.