Healthcare Analytics Made Simple: Techniques in healthcare computing using machine learning and Python

Healthcare Analytics

Made Simple

Techniques in healthcare computing using machine learning and Python

Vikas (Vik) Kumar

BIRMINGHAM - MUMBAI

Healthcare Analytics Made Simple

Copyright © 2018 Packt Publishing

All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, without the prior written permission of the publisher, except in the case of brief quotations embedded in critical articles or reviews.

Every effort has been made in the preparation of this book to ensure the accuracy of the information presented. However, the information contained in this book is sold without warranty, either express or implied. Neither the author, nor Packt Publishing or its dealers and distributors, will be held liable for any damages caused or alleged to have been caused directly or indirectly by this book.

Packt Publishing has endeavored to provide trademark information about all of the companies and products mentioned in this book by the appropriate use of capitals. However, Packt Publishing cannot guarantee the accuracy of this information.

Commissioning Editor: Veena Pagare

Acquisition Editor: Divya Poojari

Content Development Editor: Eisha Dsouza

Technical Editor: Sneha Hanchate

Copy Editor: Safis

Project Coordinator: Namrata Swetta

Proofreader: Safis Editing

Indexer: Rekha Nair

Graphics: Jisha Chirayil

Production Coordinator: Shantanu Zagade

First published: July 2018

Production reference: 1280718

Published by Packt Publishing Ltd.

Livery Place

35 Livery Street

Birmingham

B3 2PB, UK.

ISBN 978-1-78728-670-2

www.packtpub.com

To my parents, Viren and Sarita; my sister, Monica; and Tuly, my 2018 Person of the Year.

mapt.io

Mapt is an online digital library that gives you full access to over 5,000 books and videos, as well as industry leading tools to help you plan your personal development and advance your career. For more information, please visit our website.

Why subscribe?

Spend less time learning and more time coding with practical eBooks and Videos from over 4,000 industry professionals

Improve your learning with Skill Plans built especially for you

Get a free eBook or video every month

Mapt is fully searchable

Copy and paste, print, and bookmark content

PacktPub.com

Did you know that Packt offers eBook versions of every book published, with PDF and ePub files available? You can upgrade to the eBook version at www.PacktPub.com and as a print book customer, you are entitled to a discount on the eBook copy. Get in touch with us at [email protected] for more details.

At www.PacktPub.com, you can also read a collection of free technical articles, sign up for a range of free newsletters, and receive exclusive discounts and offers on Packt books and eBooks. 

Foreword

Analytics is now an integral part of healthcare. It helps to optimize treatments, improve outcomes, and the reduce the overall cost of care. The availability of biomedical, healthcare, and operational big data enables hospitals and health systems to leverage past data to predict the future of patients and their clinical pathways. Predictive modeling and healthcare data science also help to design care pathways and operational strategies that could help in streamlining various aspects of healthcare delivery. However, healthcare analytics is an exciting field that requires skills in biomedicine, data science, and the technical stack, including databases, programming, data visualization, statistics, and machine learning. While there are several books with an in-depth account of the healthcare space and analytics tools and methods, there not many easy-to-read books that integrate these things together.

In his new and exciting book, Dr. Vikas Kumar (Vik) has now blended the critical learning points of healthcare and computer science with mathematics and machine learning. Being a physician and a data scientist, Vik has done a tremendous job in compiling complex datasets and explaining several use cases in healthcare analytics with comprehensive code in MySQL and Python.

I am sure that Healthcare Analytics Made Simple will be an important addition to the library of any data scientist who's interested in understanding the key concepts of biomedical and healthcare data. It will be an indispensable companion for readers from the domains of clinical informatics and health informatics to gain critical skills in the design, development, and validation of machine learning models. This book will also be useful for physicians and biomedical scientists who are interested in understanding the landscape of healthcare analytics. The book is a joy to read, and I enjoyed working through the examples. To conclude, Healthcare Analytics Made Simple is attempting to fill a gap in the field of healthcare analytics by providing a complete and comprehensive guide, resulting in an inter-disciplinary book that will be an easy read for computer scientists, software engineers, data scientists, and healthcare professionals alike.

Dr. Shameer Khader, PhD

Director of Healthcare Data Science and Bioinformatics

Northwell Health, New York

Contributors

About the author

Dr. Vikas (Vik) Kumar grew up in the United States in Niskayuna, New York. He earned his MD from the University of Pittsburgh, but shortly afterwards he discovered his true calling of computers and data science. He then earned his MS in the College of Computing at Georgia Institute of Technology and has subsequently worked as a data scientist for both healthcare and non-healthcare companies. He currently lives in Atlanta, Georgia.

Thank you to Mark Braunstein, James Cheng, Shameer Khader, Bryant Menn, Srijita Mukherjee, and Bob Savage for their helpful comments on the book drafts.

About the reviewer

Seungjin Kim is currently a software engineer at Arcules, transforming video data into intelligence and providing a product based on distributed machine learning architecture. Previously, he was a software engineer at a genetic startup, providing a quality frontend user experience for patients accessing genetic products. He received his M.D. from the Medical School for International Health at the Ben-Gurion University of the Negev in Israel in 2015, and he received his B.S. in computer science and Engineering from the University of California in 2008.

Packt is searching for authors like you

If you're interested in becoming an author for Packt, please visit authors.packtpub.com and apply today. We have worked with thousands of developers and tech professionals, just like you, to help them share their insight with the global tech community. You can make a general application, apply for a specific hot topic that we are recruiting an author for, or submit your own idea.

Table of Contents

Title Page

Copyright and Credits

Healthcare Analytics Made Simple

Dedication

Packt Upsell

Why subscribe?

PacktPub.com

Foreword

Contributors

About the author

About the reviewer

Packt is searching for authors like you

Preface

Who this book is for

What this book covers

To get the most out of this book

Download the example code files

Download the color images

Conventions used

Get in touch

Reviews

Introduction to Healthcare Analytics

What is healthcare analytics?

Healthcare analytics uses advanced computing technology

Healthcare analytics acts on the healthcare industry (DUH!)

Healthcare analytics improves medical care

Better outcomes

Lower costs

Ensure quality

Foundations of healthcare analytics

Healthcare

Mathematics

Computer science

History of healthcare analytics

Examples of healthcare analytics

Using visualizations to elucidate patient care

Predicting future diagnostic and treatment events

Measuring provider quality and performance

Patient-facing treatments for disease

Exploring the software

Anaconda

Anaconda navigator

Jupyter notebook

Spyder IDE

SQLite

Command-line tools

Installing a text editor

Summary

References

Healthcare Foundations

Healthcare delivery in the US

Healthcare industry basics

Healthcare financing

Fee-for-service reimbursement

Value-based care

Healthcare policy

Protecting patient privacy and patient rights

Advancing the adoption of electronic medical records

Promoting value-based care

Advancing analytics in healthcare

Patient data – the journey from patient to computer

The history and physical (H&P)

Metadata and chief complaint

History of the present illness (HPI)

Past medical history

Medications

Family history

Social history

Allergies

Review of systems

Physical examination

Additional objective data (lab tests, imaging, and other diagnostic tests)

Assessment and plan

The progress (SOAP) clinical note

Standardized clinical codesets

International Classification of Disease (ICD)

Current Procedural Terminology (CPT)

Logical Observation Identifiers Names and Codes (LOINC)

National Drug Code (NDC)

Systematized Nomenclature of Medicine Clinical Terms (SNOMED-CT)

Breaking down healthcare analytics

Population

Medical task

Screening

Diagnosis

Outcome/Prognosis

Response to treatment

Data format

Structured

Unstructured

Imaging

Other data format

Disease

Acute versus chronic diseases

Cancer

Other diseases

Putting it all together – specifying a use case

Summary

References and further reading

Machine Learning Foundations

Model frameworks for medical decision making

Tree-like reasoning

Categorical reasoning with algorithms and trees

Corresponding machine learning algorithms – decision tree and random forest

Probabilistic reasoning and Bayes theorem

Using Bayes theorem for calculating clinical probabilities

Calculating the baseline MI probability

2 x 2 contingency table for chest pain and myocardial infarction

Interpreting the contingency table and calculating sensitivity and specificity

Calculating likelihood ratios for chest pain (+ and -)

Calculating the post-test probability of MI given the presence of chest pain

Corresponding machine learning algorithm – the Naive Bayes Classifier

Criterion tables and the weighted sum approach

Criterion tables

Corresponding machine learning algorithms – linear and logistic regression

Pattern association and neural networks

Complex clinical reasoning

Corresponding machine learning algorithm – neural networks and deep learning

Machine learning pipeline

Loading the data

Cleaning and preprocessing the data

Aggregating data

Parsing data

Converting types

Dealing with missing data

Exploring and visualizing the data

Selecting features

Training the model parameters

Evaluating model performance

Sensitivity (Sn)

Specificity (Sp)

Positive predictive value (PPV)

Negative predictive value (NPV)

False-positive rate (FPR)

Accuracy (Acc)

Receiver operating characteristic (ROC) curves

Precision-recall curves

Continuously valued target variables

Summary

References and further reading

Computing Foundations – Databases

Introduction to databases

Data engineering with SQL – an example case

Case details – predicting mortality for a cardiology practice

The clinical database

The PATIENT table

The VISIT table

The MEDICATIONS table

The LABS table

The VITALS table

The MORT table

Starting an SQLite session

Data engineering, one table at a time with SQL

Query Set #0 – creating the six tables

Query Set #0a – creating the PATIENT table

Query Set #0b – creating the VISIT table

Query Set #0c – creating the MEDICATIONS table

Query Set #0d – creating the LABS table

Query Set #0e – creating the VITALS table

Query Set #0f – creating the MORT table

Query Set #0g – displaying our tables

Query Set #1 – creating the MORT_FINAL table

Query Set #2 – adding columns to MORT_FINAL

Query Set #2a – adding columns using ALTER TABLE

Query Set #2b – adding columns using JOIN

Query Set #3 – date manipulation – calculating age

Query Set #4 – binning and aggregating diagnoses

Query Set #4a – binning diagnoses for CHF

Query Set #4b – binning diagnoses for other diseases

Query Set #4c – aggregating cardiac diagnoses using SUM

Query Set #4d – aggregating cardiac diagnoses using COUNT

Query Set #5 – counting medications

Query Set #6 – binning abnormal lab results

Query Set #7 – imputing missing variables

Query Set #7a – imputing missing temperature values using normal-range imputation

Query Set #7b – imputing missing temperature values using mean imputation

Query Set #7c – imputing missing BNP values using a uniform distribution

Query Set #8 – adding the target variable

Query Set #9 – visualizing the MORT_FINAL_2 table

Summary

References and further reading

Computing Foundations – Introduction to Python

Variables and types

Strings

Numeric types

Data structures and containers

Lists

Tuples

Dictionaries

Sets

Programming in Python – an illustrative example

Introduction to pandas

What is a pandas DataFrame?

Importing data

Importing data into pandas from Python data structures

Importing data into pandas from a flat file

Importing data into pandas from a database

Common operations on DataFrames

Adding columns

Adding blank or user-initialized columns

Adding new columns by transforming existing columns

Dropping columns

Applying functions to multiple columns

Combining DataFrames

Converting DataFrame columns to lists

Getting and setting DataFrame values

Getting/setting values using label-based indexing with loc

Getting/setting values using integer-based labeling with iloc

Getting/setting multiple contiguous values using slicing

Fast getting/setting of scalar values using at and iat

Other operations

Filtering rows using Boolean indexing

Sorting rows

SQL-like operations

Getting aggregate row COUNTs

Joining DataFrames

Introduction to scikit-learn

Sample data

Data preprocessing

One-hot encoding of categorical variables

Scaling and centering

Binarization

Imputation

Feature-selection

Machine learning algorithms

Generalized linear models

Ensemble methods

Additional machine learning algorithms

Performance assessment

Additional analytics libraries

NumPy and SciPy

matplotlib

Summary

Measuring Healthcare Quality

Introduction to healthcare measures

US Medicare value-based programs

The Hospital Value-Based Purchasing (HVBP) program

Domains and measures

The clinical care domain

The patient- and caregiver-centered experience of care domain

Safety domain

Efficiency and cost reduction domain

The Hospital Readmission Reduction (HRR) program

The Hospital-Acquired Conditions (HAC) program

The healthcare-acquired infections domain

The patient safety domain

The End-Stage Renal Disease (ESRD) quality incentive program

The Skilled Nursing Facility Value-Based Program (SNFVBP)

The Home Health Value-Based Program (HHVBP)

The Merit-Based Incentive Payment System (MIPS)

Quality

Advancing care information

Improvement activities

Cost

Other value-based programs

The Healthcare Effectiveness Data and Information Set (HEDIS)

State measures

Comparing dialysis facilities using Python

Downloading the data

Importing the data into your Jupyter Notebook session

Exploring the data rows and columns

Exploring the data geographically

Displaying dialysis centers based on total performance

Alternative analyses of dialysis centers

Comparing hospitals

Downloading the data

Importing the data into your Jupyter Notebook session

Exploring the tables

Merging the HVBP tables

Summary

References

Making Predictive Models in Healthcare

Introduction to predictive analytics in healthcare

Our modeling task – predicting discharge statuses for ED patients

Obtaining the dataset

The NHAMCS dataset at a glance

Downloading the NHAMCS data

Downloading the ED2013 file

Downloading the list of survey items – body_namcsopd.pdf

Downloading the documentation file – doc13_ed.pdf

Starting a Jupyter session

Importing the dataset

Loading the metadata

Loading the ED dataset

Making the response variable

Splitting the data into train and test sets

Preprocessing the predictor variables

Visit information

Month

Day of the week

Arrival time

Wait time

Other visit information

Demographic variables

Age

Sex

Ethnicity and race

Other demographic information

Triage variables

Financial variables

Vital signs

Temperature

Pulse

Respiratory rate

Blood pressure

Oxygen saturation

Pain level

Reason-for-visit codes

Injury codes

Diagnostic codes

Medical history

Tests

Procedures

Medication codes

Provider information

Disposition information

Imputed columns

Identifying variables

Electronic medical record status columns

Detailed medication information

Miscellaneous information

Final preprocessing steps

One-hot encoding

Numeric conversion

NumPy array conversion

Building the models

Logistic regression

Random forest

Neural network

Using the models to make predictions

Improving our models

Summary

References and further reading

Healthcare Predictive Models – A Review

Predictive healthcare analytics – state of the art

Overall cardiovascular risk

The Framingham Risk Score

Cardiovascular risk and machine learning

Congestive heart failure

Diagnosing CHF

CHF detection with machine learning

Other applications of machine learning in CHF

Cancer

What is cancer?

ML applications for cancer

Important features of cancer

Routine clinical data

Cancer-specific clinical data

Imaging data

Genomic data

Proteomic data

An example – breast cancer prediction

Traditional screening of breast cancer

Breast cancer screening and machine learning

Readmission prediction

LACE and HOSPITAL scores

Readmission modeling

Other conditions and events

Summary

References and further reading

The Future – Healthcare and Emerging Technologies

Healthcare analytics and the internet

Healthcare and the Internet of Things

Healthcare analytics and social media

Influenza surveillance and forecasting

Predicting suicidality with machine learning

Healthcare and deep learning

What is deep learning, briefly?

Deep learning in healthcare

Deep feed-forward networks

Convolutional neural networks for images

Recurrent neural networks for sequences

Obstacles, ethical issues, and limitations

Obstacles

Ethical issues

Limitations

Conclusion of this book

References and further reading

Other Books You May Enjoy

Leave a review - let other readers know what you think

Preface

The functional aim of this book is to demonstrate how Python packages are used for data analysis; how to import, collect, clean, and refine data from Electronic Health Record (EHR) surveys; and how to make predictive models with this data, with the help of real-world examples.

Who this book is for

Healthcare Analytics Made Simple is for you if you are a developer who has a working knowledge of Python or a related programming language, even if you are new to healthcare or predictive modeling with healthcare data. Clinicians interested in analytics and healthcare computing will also benefit from this book. This book can also serve as a textbook for students enrolled on an introductory course on machine learning for healthcare.

What this book covers

Chapter 1, Introduction to Healthcare Analytics, provides a definition of healthcare analytics, lists some foundational topics, provides a history of the subject, gives some examples of healthcare analytics in action, and includes download, installation, and basic usage instructions for the software in this book.

Chapter 2, Healthcare Foundations, consists of an overview of how healthcare is structured and delivered in the US, provides a background on legislation that's relevant to healthcare analytics, describes clinical patient data and clinical coding systems, and provides a breakdown of healthcare analytics.

Chapter 3, Machine Learning Foundations, describes some of the model frameworks used for medical decision making and describes the machine learning pipeline, from data import to model evaluation.

Chapter 4, Computing Foundations – Databases, provides an introduction to the SQL language and demonstrates the use of SQL in healthcare with a healthcare predictive analytics example.

Chapter 5, Computing Foundations – Introduction to Python, gives a basic overview of Python and the libraries that are important for performing analytics. We discuss variable types, data structures, functions, and modules in Python. We also give an introduction to the pandas and scikit-learn libraries.

Chapter 6, Measuring Healthcare Quality, describes the measures used in healthcare performance, gives an overview of value-based programs in the US, and demonstrates how to download and analyze provider-based data in Python.

Chapter 7, Making Predictive Models in Healthcare, describes the information contained in a publicly available clinical dataset, including downloading instructions. We then demonstrate how to make predictive models with this data, using Python, pandas, and scikit-learn.

Chapter 8, Healthcare Predictive Models – A Review, reviews some of the current progress being made in healthcare predictive analytics for select diseases and application areas by comparing machine learning results to those obtained by using traditional methods.

Chapter 9, The Future – Healthcare and Emerging Technologies, discusses some of the advances being made in healthcare analytics through using the internet, introduces the reader to deep learning techniques in healthcare, and states some of the challenges and limitations facing healthcare analytics.

To get the most out of this book

Helpful things to know include the following:

High school math, such as basic probability, statistics, and algebra

Basic familiarity with a programming language and/or basic programming concepts

Basic familiarity with healthcare and a working knowledge of some clinical terminology

Please follow the instructions in Chapter 1, Introduction to Healthcare Analytics for setting up Anaconda and SQLite.

Download the example code files

You can download the example code files for this book from your account at www.packtpub.com. If you purchased this book elsewhere, you can visit www.packtpub.com/support and register to have the files emailed directly to you.

You can download the code files by following these steps:

Log in or register at www.packtpub.com.

Select the SUPPORT tab.

Click on Code Downloads & Errata.

Enter the name of the book in the Search box and follow the onscreen instructions.

Once the file is downloaded, please make sure that you unzip or extract the folder using the latest version of:

WinRAR/7-Zip for Windows

Zipeg/iZip/UnRarX for Mac

7-Zip/PeaZip for Linux

The code bundle for the book is also hosted on GitHub at https://github.com/PacktPublishing/Healthcare-Analytics-Made-Simple. In case there's an update to the code, it will be updated on the existing GitHub repository.

We also have other code bundles from our rich catalog of books and videos available at https://github.com/PacktPublishing/. Check them out!

Download the color images

We also provide a PDF file that has color images of the screenshots/diagrams used in this book. You can download it here: http://www.packtpub.com/sites/default/files/downloads/HealthcareAnalyticsMadeSimple_ColorImages.pdf.

Conventions used

There are a number of text conventions used throughout this book.

CodeInText: Indicates code words in text, database table names, folder names, filenames, file extensions, pathnames, dummy URLs, user input, and Twitter handles. Here is an example: Mount the downloaded WebStorm-10*.dmg disk image file as another disk in your system.

A block of code is set as follows:

string_1 = '1'

string_2 = '2'

string_sum = string_1 + string_2

print(string_sum)

When we wish to draw your attention to a particular part of a code block, the relevant lines or items are set in bold:

test_split_string = 'Jones,Bill,49,Atlanta,GA,12345'

output = test_split_string.split(',')

print(output)

Bold: Indicates a new term, an important word, or words that you see onscreen. For example, words in menus or dialog boxes appear in the text