Index

As this ebook edition doesn't have fixed pagination, the page numbers below are hyperlinked for reference only, based on the printed edition of this book.

A

absorption, distribution, metabolism, and extraction (ADME) 114

academic medical centers (AMCs) 123

active learning pipelines 181

adverse drug reaction (ADR) 128

adverse event clustering model pipeline

building, on SageMaker 135, 136

deploying 140, 141

inference, running 140, 141

Jupyter notebooks, running 136

reviewing 136-140

adverse events 125

AI for healthcare and life sciences, industry trends 183

clinical conditions, treating 184

digital twins, using 185

Internet of Things (IoT), using 185

robotics, using 185

telehealth and remote care 184

AI in healthcare and life sciences, future 186

blockchain 188

federated learning 187

quantum computing 188, 189

reinforcement learning 186

virtual reality (VR) 187

AI in healthcare and life sciences, key factors 180

active learning pipelines 181

democratization with no-code AI tools 181, 182

infrastructure and models 182

multimodal data 180

responsible AI 183

AI services

for healthcare 30, 31

for life sciences 30, 31

algorithm 7

decision tree 7

linear regression 7

non-linear regression 7

Alphafold 114, 116

Amazon Comprehend Medical 21-23, 68, 108

Amazon Forecast 150

dataset, importing 152

forecast generation 153

forecasting models, training 152

Amazon Forecast algorithms 150

Autoregressive Integrated Moving Average (ARIMA) 151

CNN-QR 151

DeepAR+ 152

Exponential Smoothing (ETS) 151

Non-Parametric Time Series (NPTS) 151

Prophet 151

Amazon HealthLake 27

data, creating into data store 27, 29

data, exporting from 29, 30

data, importing into data store 27, 29

data, querying from 29, 30

Amazon Kendra 130

Amazon machine images (AMIs) 15

Amazon Personalize 148

reference link 148

Amazon QuickSight 150

Amazon Resource Name (ARN) 153

Amazon SageMaker 15, 31, 107

Amazon SageMaker Studio 68

Amazon SageMaker training 87

architecture 87, 88

options 88, 89

Amazon Textract 31, 61, 107

Amazon Transcribe 23

Amazon Transcribe Medical 23, 24

batch transcription job 25

custom vocabulary 26, 27

streaming 24, 25

Amazon Web Services (AWS) 148

analytics services, on AWS

reference link 102

assays 114

asynchronous inference on SageMaker 106

at-risk patients

identifying 36, 37

auditability 166

Augmented AI (A2I)

reference link 181

AutoGluon 91

reference link 91

automatic speech recognition (ASR) 23

AutoML 182

AutoML framework 91

Autoregressive Integrated Moving Average (ARIMA) algorithm 151

reference link 151

autoscaling of models on SageMaker

reference link 104

AWS AI services

for healthcare 21

for life sciences 21

used, for building smart medical transcription application 57

AWS Command Line Interface (CLI) 22

AWS console 22

AWS Data Exchange (ADX)

reference link 181

AWS HIPAA Eligible Services 22

AWS Key Management Service (KMS) 28

AWS machine learning stack 14, 15

AI services layer 15

ML frameworks and infrastructure layer 15

ML services layer 15

AWS ML services 14

for healthcare 21

for life sciences 21

AWS SDK 22

AWS security options

URL 101

AWS storage services

URL 100

B

base pairs 95

Basic Stand Alone (BSA) 72

batch transform on SageMaker 106

Bayesian 151

BERT 118

bias

detecting 171

viewing 173

Big Data Analytics 127

binary classification model 38

binning 12

BioBERT 109

biochemical assays 115

Bio Creative II Gene Mention Recognition 109

black box 159

blockchain 188

C

CAR T-cell therapy 115

Center for Medical Devices and Radiological Health (CDRH) 83

Centers for Disease Control and Prevention (CDC) 23

Centers for Medicare and Medicaid Services (CMS) 36, 180

ChEMBL 118

Cheminformatics 114

chromatographic approach 114

chromosomes 95

classification models 130

clinical document processing in healthcare

automating 53, 54

claim forms 54

discharge summary 54

medical test reports 54

mental health reports 54

patient history forms 54

clinical NER application

building 107, 108

Clinical Research Organizations (CROs) 20

clinical trial

workflow 125-127

clinical trial workflow

RWE studies 127

clustering 38, 102

cold start 106

compounds

language models 118

computer-aided drug design (CADD) 115

computer vision (CV) model 67, 127, 159

Compute Services

URL 101

concept identifiers 23

confidence score 9, 23

congestive heart failure (CHF) 37

container

adapting, for SageMaker inference 119

adapting, for SageMaker training 119

Convolutional Neural Network - Quantile Regression (CNN-QR) 151

convolutional NNs (CNNs) 151

cryo-electron microscopy (cryo-EM) 113, 116, 117

crystallography 115

custom containers

in SageMaker 118

custom libraries 120

cytosine (C) 95

D

data anonymization 164

data encryption 164

data masking 164

data parallel libraries 183

data processing 12

data quality monitoring

setting up 171

dataset

acquiring 172

dataset group 152

item metadata 152

related time series 152

target time series 152

decentralized trials 184

decryption key 164

DeepAR+ algorithm 152

deep learning (DL) 149

Deep Purpose 120

Defense Innovation Unit (DIU) 183

deoxyribonucleic acid (DNA) 95

Department of Defense (DoD) 183

Department of Health and Human Services (HHS) 161

depth of reads 98

determined 18

diagnosis-related group (DRG) 72

Difference in Conditional Outcome (DCO) 168

Difference in Label Rates (DLR) 169

Difference in Positive Proportions in Predicted Labels (DPPL) 168

Digital Imaging and Communications in Medicine (DICOM) 85

reference link 85

digital twins 185

directed acyclic graph (DAG) 131

distributed training libraries 183

reference link 183

DrugBank 118

drug discovery and design process 114, 115

structure-based drug design 115, 116

druggable sites 116

drug protein interaction (DPI) 116

drug target interaction (DTI) 117

E

edge deployment 185

edge inference 185

efficacy 115, 123

Elastic Map Reduce (EMR) 102

electronic health record (EHR) 55, 161

electronic medical record (EMR) 127

encryption key 164

Ethical Principles for Artificial Intelligence 183

evaluation metric 8

exclusion criteria 125

exome sequencing 97

explainability reports

in SageMaker Studio 173-177

explanation of benefits 67

Exponential Smoothing (ETS) 151

reference link 151

extract, transform, load (ETL) 181

F

Fast Healthcare Interoperability Resources (FHIR) 27

FASTQ 98

feature engineering 12

feature engineering, techniques

binning 12

label encoding 12

one hot encoding 12, 13

scaling 12

standardizing 12

federated learning 187

fee-for-service 36

field-programable gate array (FPGA) 99

first-generation sequencing 98

Food and Drug Administration (FDA) 19, 83, 123

Framingham risk score 38

G

gene-based therapy 115

General Data Protection Regulation (GDPR) 162

generative models 117

genome 95

genome-wide association studies (GWAS) 99

genomic code

acquiring 108, 109

genomic data, processing challenges 99

access control 100, 101

analysis 102

compute 101

interpretation 102

privacy 100, 101

sharing 100, 101

storage volume 100

genomic entities

building 107, 108

genomics 20, 21

Genomics CLI 101

URL 101

genomic sequencing 95, 97

evolution 98, 99

genomic sequencing stages

primary analysis 98

secondary analysis 98

tertiary analysis 98

genomic test report

acquiring 108, 109

genomic workflows

machine learning (ML), applying to 102, 103

global explanations 165

Glue 102

Good Clinical Practices (GCPs) 162

Good Laboratory Practices (GLPs) 162

Good Manufacturing Practices (GMPs) 162

good practices (GxP) 162

GPT model 118

graph databases 113

graph neural networks 129

Gremlin 113

guanine (G) 95

H

HbA1C values 38

healthcare claims processing 65, 67

healthcare claims processing workflows

ML, implementing 67, 68

healthcare continuum 56

healthcare interoperability 27

healthcare payers 19

healthcare providers 18

Health Information Technology for Economic and Clinical Health (HITECH) 161

Health Insurance Portability and Accountability Act (HIPAA) 161

Health Language International (HL7) 27

high-throughput screening (HTS) 115

homology modeling 116

Hugging Face Hub 109

reference link 109

Hugging Face transformers

reference link 109

human genome 95

Human Genome Project (HGP) 95

Human Intestinal Absorption (HIA) 121

human-in-the-loop method 9

human-in-the-loop pipeline 166

hyperparameters 7

Hypertext Transfer Protocol (HTTP) 28

hypothesis testing phase 129

I

image recognition 54

inclusion criteria 125

individual case safety report (ICSR) 128

inference 8, 9

asynchronous batch process 9

synchronous real-time process 9

inference container, adapting

reference link 119

informed consent 126

informed consent document 126

Inpatient Public Use Files (PUF) 72

institutional review board (IRB) 126

Integrated Development Environment (IDE) 31

Investigational New Drug (IND) 127

J

JSON format 131

Jupyter lab 68

Jupyter notebook

running 155, 156

Jupyter Notebook

running 120, 121, 136, 172, 173

K

Kendra 61

L

label encoding 12

language models

for compounds 118

for proteins 118

large molecules 113, 114

lead optimization 115

local explanations 165

Local Interpretable Model-Agnostic Explanations (LIME) 165

reference link 165

local mode 121

M

machine learning (ML) 4, 5, 51

applying, in healthcare 18

applying, in life sciences 18

applying, to clinical trials 129

applying, to medical devices 85, 87

applying, to molecular data 116

applying, to PV 129

applying, to radiology imaging system 85, 87

implementing, in healthcare claims processing workflows 67, 68

on AWS 14

supervised learning model 5

unsupervised learning model 6

machine learning (ML) application to genomics 102

disease detection and diagnosis 103

drugs and therapies, discovery 103

knowledge mining and searches 103

labs and sequencers 102, 103

machine learning (ML), for pharmaceutical supply chain

drug demand, forecasting 148, 149

implementing 148

market sentiment and completive analysis 149

patients, targeting 148

predictive maintenance, implementing 149

providers, targeting 148

sales outlook, forecasting 148, 149

macromolecules 113

Magnetic Resonance Imaging (MRI) 19

Mean Squared Error 8

medical device 19, 20, 83

classes 83

ML applying to 85, 87

SaMD 84

medical image classification model

building 91, 92

building, with SageMaker 89

code, acquiring 89, 90

dataset, acquiring 89

evaluating 91, 92

training 91, 92

medical imaging 19, 20

Medicare claim amounts

ML model, building to predict 71

MIMIC 3

reference link 180

MIMIC CxR

reference link 180

ML algorithms

implementing, for patient risk stratification 37-39

ML, clinical trials and PV

adverse event detection 130

literature search 129

protocol design 129

real-world data analysis 131

reporting 130

trial participant recruitment 130

ML, for breast cancer risk prediction

dataset, importing 41-43

implementing 40

model, analyzing 45-48

model, building 43-45

predicting, from model 48

ML frameworks

PyTorch 8

scikit-learn 8

TensorFlow 8

ML, impact over healthcare and life sciences industry

genomics 20, 21

healthcare payers 19

healthcare providers 18

medical device 19, 20

medical imaging 19, 20

pharmaceutical organizations 20

ML in healthcare and life sciences, challenges 161

bias and transparency 163

generalizability 163

regulations 161

reproducibility 163

security and privacy 162

ML in healthcare and life sciences, challenges option 163

auditability and review 166, 167

bias detection 164, 165

data anonymization 164

data encryption 164

explainability 164, 165

reproducible ML pipeline, building 165

ML lifecycle

data processing 12

exploring 9, 10

feature engineering 12

model deployment 13

model training 13

problem definition 10

stages 9

ML, medical devices and radiology imaging system

examples 86

ML model 8

building, steps 39, 40

building, to predict Medicare claim amounts 71

portability 8

ML models on the edge 20

ML model, to predict Medicare claim amounts

building 78

data, acquiring 72

evaluating 78

feature engineering 72-77

training 78

ML Operations (MLOps) 13

MLOps pipeline 13, 69

ML terminologies

algorithms 7

inference 8, 9

model 8

training 7

model drift 165

model input parameters 7

model package 134

model parallel libraries 183

model predictions

for healthcare coverage 171

with SageMaker Clarify 169

model quality monitoring

setting up 171

molecular data 111, 113

large molecules 113, 114

machine learning (ML), applying to 116

small molecules 113

molecular modeling 115

molecular property prediction 117

model, building on SageMaker 120

molecular reaction prediction 116

molecular structure prediction 117

molecule 114

Moores law

reference link 179

multiplexing 99

MXNet 88

N

Named Entity Extraction (NEE) 22

named entity recognition (NER) 68, 130, 150

national institute of health (NIH) 123

Natural Language Processing (NLP) 21, 55, 68, 118, 127, 150, 170

NEGATION 22

Neo4j 113

neural network (NN) 149, 159

newline delimited JSON (ndjson) format 30

Next-Generation Sequencing (NGS) 21, 97

n-grams 118

Non-Parametric Time Series (NPTS) algorithm

climatological forecaster 151

NPTS 151

seasonal climatological forecaster 151

seasonal NPTS 151

nucleotides 95

O

one hot encoding 12

applying 13

OpenFDA

reference link 181

operational efficiency in healthcare 52, 53

optical character recognition (OCR) 54, 67

Oxford Nanopore Technologies (ONT) 99

P

Pacific Biosciences (PacBio) 99

Pandas Flavor 120

pandas-flavor 0.3.0

reference link 120

partial dependence plot (PDP) 169

patient risk stratification

ML algorithms, implementing 37-39

permutation importance 165

personally identifiable information (PII) 162

pharmaceutical organizations 20

pharmaceutical sales 147

machine learning, applying 148

pharmaceutical sales forecasting model

building, with Amazon Forecast 153

dataset, acquiring 154, 155

Jupyter notebook, executing 155, 156

pharmaceutical supply chain industry

challenges 146, 147

consumers 146

distributors 146

drug manufacturers 145

landscape 145

machine learning, applying 148

pharmaceutical sales 147

wholesalers 146

pharmacology 113

Pharmacovigilance (PV) 128

hypothesis testing 129

signal generation 128

pharmacy benefit manager (PBM) 146

pharma sales rep 147

phenotype 95, 114

phenotypic screening 114

picture archiving and communications system (PACS) 85

placebo 125

post-market surveillance (PMS) 128

post-training bias metrics 167, 168

Difference in Conditional Outcome (DCO) 168

Difference in Label Rates (DLR) 169

Difference in Positive Proportions in Predicted Labels (DPPL) 168

Recall Difference (RD) 169

precision medicine 21, 116

predicting 8

prediction 4

pre-trained genetic entity detection model 109, 110

reference link 109

pre-training bias metrics 167

Class Imbalance (CI) 167

Conditional Demographic Disparity in Labels (CDDL) 168

Difference in Proportions in Labels (DPl) 168

Kullback-Leibler (KL) divergence 168

reference link 168

principal component analysis (PCA) 102

problem definition, ML 10

appetite, for experimentation 11

dataset availability 11

predictive element 10, 11

Prophet 151

reference link 151

protected health information (PHI) 22, 57, 161, 180

protein databank (PDB) 114

protein-protein interactions (PPIs) 117

proteins

language models 118

protein structures 114

PyTDC 0.3.7

reference link 120

PyTorch 88

Q

quantitative structure- activity relationship (QSAR) 117

quantum bits 188

quantum computers 188

quantum computing 188

quantum mechanics 188

quantum qubits 188

R

radiology imaging system

components 84, 85

ML applying to 85, 87

raw genomic data 98

RDKit 120

real-time endpoint options, on SageMaker 104

host models with preprocessing logic 105

multiple models in different containers 105

multiple models in one container 104

single model 104

real-world data (RWD) 127

real-world evidence (RWE) 127

Recall Difference (RD) 169

recommendation engine 148

recurrent neural networks (RNNs) 118

recurrent NN (RNN)-based architecture 152

Registry of Open Data

reference link 181

Registry of Open Data on AWS

URL 101

regression model 38, 130

regulatory development kit (RDK) 84

reinforcement learning 186

responsible AI 183

reference link 183

ribonucleic acid (RNA) 95

risk stratification 36

at-risk patients, identifying 36, 37

RNA sequencing 97

Root Mean Square Error (RMSE) 40, 153

RWE studies 127

S

S3 bucket

creating 57, 58

SageMaker

adverse event clustering model pipeline, building 135, 136

custom containers 118

molecular property prediction model, building 120

used, for building medical image classification model 89

SageMaker Canvas 31, 39, 40

reference link 182

SageMaker Clarify 160, 167

post-training bias metrics 168

pre-training bias metrics 167

used, for detecting bias 167

used, for model predictions 169

SageMaker Clarify post-training metrics

reference link 169

SageMaker Data Wrangler 69

data, analyzing 70

data and workflows, exporting 70

data, importing 69

data, transforming 70

SageMaker Ground Truth

reference link 181

SageMaker Inference 104

asynchronous inference 106

batch transform 106, 107

container, adapting 119

real-time endpoint options 104, 105

reference link 107

serverless inference 105, 106

SageMaker Model Monitor 160, 167

data quality monitoring, setting up 171

model quality monitoring, setting up 171

used, for monitoring models 170

SageMaker model registry 131-135

SageMaker Pipelines 69, 131

caching 133, 134

steps, defining 132, 133

SageMaker Processing jobs 69

SageMaker Python SDK 134

reference link 104

SageMaker resources

reference link 121

SageMaker Studio 31

explainability reports 173-177

SageMaker Studio notebooks 71

SageMaker training

container, adapting 119

SageMaker Training job 69

SageMaker training mode 121

SARS-CoV-2 virus 97

scaling 12

scikit-learn 88

second-generation sequencing 99

secret key 164

SHapley Additive exPlanations (SHAP) 165

reference link 165

short-read sequences 99

signal generation 128

signal versus noise 128

simple notification service (SNS) 106

Simple Storage Service (S3) 171

simplified molecular-input line-entry system (SMILES) 113

single-molecule real-time (SMRT) 99

single nucleotide polymorphisms (SNPs) 98

small molecule pathway database (SMDB) 113

small molecules 113

smart factories 149

smart medical transcription application

audio file, downloading 58, 59

building, with AWS AI services 57

Python script, downloading 58, 59

running 59-61

S3 bucket, creating 57, 58

software as a medical device (SaMD) 84, 183

reference link 84

Solexa 99

speaker diarization 56

spectroscopy 113

speech-to-text 55, 127

sponsors 123

spontaneous reporting 128

standardizing 12

state of the art (SOTA) 159, 179

structure-based drug design 114-116

subject matter experts (SMEs) 11

supervised learning (SL) algorithm 152

supervised ML model 5

Synthea 28, 172

reference link 172

T

TAR file

URL 108

target-based drug discovery (TBDD) 114

target discovery 111

telehealth 184

TensorFlow 88

term frequency-inverse document frequency (tf-idf) 135

Textract 67

The Cancer Genome Atlas (TCGA) 108

reference link 180

Therapeutics Data Commons (TDC) 120

third-generation sequencing (TGS) 99

thymine (T) 95

time series analysis 38

training 4, 7

transcription 95

transformer architecture 118

reference link 118

U

unique identifier (UID) 171

unsupervised ML model 6

update model feature 40

V

value-based care model 36

variant call files (VCFs) 98

variant calling 98

versioning 166

virtual reality (VR) 187

voice-based applications in healthcare

medical transcription software 55

remote patient monitoring and patient adherence solutions 56

telemedicine and telehealth services 56

virtual assistants and chatbots 55, 56

working with 55

W

Weighted Quantile Loss (wQL) 153

whole genome sequencing (WGS) 97

wholesale acquisition cost (WAC) 145

Workflow Description Language (WDL) 101

X

XGBoost 88

X-ray crystallography 113, 116

Z

ZINC 118