ATLAS Tutorial: Data Sources - Procedure

OHDSI
23 May 201905:49

Summary

TLDRThis video explores the Procedure Report feature within Atlas, a data analysis tool. Users can select data sources and reports to analyze, visualized through a tree map and table. The script demonstrates how to drill down into specific concepts, like 'Radiologic exam chest two views,' to reveal prevalence, patient demographics, and seasonal patterns. It also shows how to filter and search for concepts, such as 'coronary artery bypass,' providing a comprehensive overview of data trends and insights.

Takeaways

  • 🌐 The script is about exploring the 'Procedure Report' within the data sources capability of Atlas.
  • 🔍 Users can start from the Atlas homepage, navigate to 'Data Sources', and select a specific data source and report to explore.
  • 📊 The 'Procedure Report' initially presents a tree map representation showing the prevalence of all concepts in the procedure domain.
  • 🔢 The size of each box in the tree map indicates the concept prevalence, while the color represents the intensity of the concept.
  • 📝 A tabular display of the same information is available under the 'Table' tab, listing concept IDs, names, and prevalence data.
  • 🔍 The table can be sorted, searched, and filtered to find specific information about the concepts.
  • 🔎 By selecting a row in the table, users can drill down to get more detailed information about a particular concept.
  • 📈 The report includes various graphs, such as concept prevalence stratified by year, age, gender, and a seasonal pattern of the concept occurrence.
  • 🏥 The data source's provenance is identified, showing where the data for a given concept originates, such as inpatient or outpatient claims.
  • 📊 The report also includes box plots showing the age at the first occurrence of a concept, stratified by gender.
  • 📊 A cumulative frequency distribution graph is provided, showing the number of occurrences of a concept for given individuals.
  • 🔎 Users can filter the report by searching for specific concepts, such as 'coronary artery bypass', to explore their prevalence in detail.

Q & A

  • What is the main feature of Atlas explored in the video script?

    -The main feature explored in the video script is the 'procedure report' within the data sources capability of Atlas.

  • How does one access the procedure report in Atlas?

    -To access the procedure report in Atlas, one must go to the data sources from the Atlas homepage, then select the source of data and the report they wish to explore.

  • What is represented by the size of the boxes in the tree map representation of the procedure report?

    -The size of the boxes in the tree map representation corresponds to the concept prevalence within the data source.

  • How is the intensity of the concepts represented in the tree map?

    -The intensity of the concepts is represented by the color of the boxes in the tree map.

  • What additional information is available under the 'table' tab in the procedure report?

    -Under the 'table' tab, a line listing is available that represents each concept ID, its name, and the number of persons who have had that concept, along with the concept prevalence and records per person.

  • How many different concepts are shown in the table from the procedure report?

    -The table shows 27,000 different concepts within the particular data source.

  • What actions can be performed on the table in the procedure report?

    -The table in the procedure report can be sorted, searched, and filtered for any information.

  • What happens when a row in the table is selected?

    -When a row in the table is selected, it allows the user to drill down and find further information about that particular concept.

  • Can you provide an example of the information obtained after drilling down into a concept?

    -An example given is selecting the concept 'Radiologic exam chest two views frontal and lateral', which shows the number of patients, the prevalence of the concept within the data source, and the average number of times patients have that concept.

  • How is the concept prevalence information stratified in the first graph after drilling down into a concept?

    -The concept prevalence information is stratified by year, age, and gender, with each graph showing index year on the x-axis, age deciles, and gender represented by line series.

  • What does the second graph after drilling down into a concept represent?

    -The second graph represents the prevalence of the concept by month, showing a seasonal pattern of when chest radiologic exams occur.

  • What does the 'type concept ID' graph indicate?

    -The 'type concept ID' graph identifies the provenance of the data for the given source, showing where the data records for radiologic exams occur, such as inpatient and outpatient claims.

  • How is the age at the first occurrence of a concept displayed?

    -The age at the first occurrence of a concept is displayed in a box plot, stratified by gender, with age on the y-axis.

  • What does the cumulative frequency distribution graph show?

    -The cumulative frequency distribution graph shows the number of occurrences of a particular concept for given individuals, indicating how many people had multiple occurrences of a chest x-ray.

  • How can a user search for a specific concept within the procedure report?

    -A user can search for a specific concept, such as 'coronary artery bypass', by typing the term in the search bar, which filters the set of concepts to those that match the search string.

  • What additional information is provided when a specific concept is selected?

    -When a specific concept is selected, the user can further explore the prevalence of that concept by year, age, gender, by month, by provenance of data, and by age.

  • Where can one find more information about Atlas and the Odyssey community?

    -More information about Atlas and the Odyssey community can be found at Odyssey.org.

Outlines

00:00

📊 Exploring Procedure Reports in Atlas

The video introduces the procedure report feature within the data sources capability of Atlas. From the Atlas homepage, users can navigate to data sources, select the desired data source and report, and view a tree map representation of concept prevalence within the procedure domain of the OMOP Common Data Model. The tree map displays concept prevalence by box size and intensity by color. Additionally, a tabular view is available, showing a detailed listing of concept IDs, names, prevalence, and records per person. Users can sort, search, and filter this table, and drill down into specific concepts for further details. For example, selecting 'Radiologic exam chest two views frontal and lateral' reveals data on patient numbers, average occurrences, and prevalence over time. Various graphs illustrate concept prevalence by year, age, gender, month, data provenance, age at first occurrence, and cumulative frequency distribution, providing comprehensive insights into the data.

05:01

🔍 Filtering and Drilling Down Concepts in Atlas

The video continues by demonstrating the filtering and drilling down capabilities within the procedure report in Atlas. Users can search for specific concepts, such as 'coronary artery bypass,' and filter the results accordingly. Selecting a concept allows users to explore its prevalence by various dimensions such as year, age, gender, month, and data provenance. This section concludes with an invitation to visit Odyssey.org for more information about Atlas's data sources capability and the broader Odyssey community.

Mindmap

Keywords

💡Procedure Report

A 'Procedure Report' in the context of the video refers to a type of report that details the prevalence and distribution of medical procedures within a dataset. It is a key concept as it is the central focus of the video's demonstration, showing how to explore and analyze data related to medical procedures using Atlas. For example, the script describes how to select a procedure report to view a tree map representation of concept prevalence.

💡Atlas

Atlas is the platform being demonstrated in the video, which offers capabilities for exploring and analyzing healthcare data. It is integral to the video's theme as it provides the interface through which users can access data sources and generate reports. The script mentions navigating the Atlas homepage to access data sources and reports.

💡Data Sources

In the script, 'Data Sources' refers to the various datasets available within the Atlas platform. These sources are crucial for generating reports and are the foundation of the analysis presented in the video. The video script illustrates how to select a data source to explore specific reports, such as the 'Procedure Report'.

💡Tree Map

A 'Tree Map' is a visualization technique used in the video to represent the prevalence of concepts within the procedure domain. The size of each box in the tree map indicates the prevalence of a concept, and the color represents the intensity. This concept is central to understanding how data is visually presented in the video.

💡Concept Prevalence

'Concept Prevalence' is a measure of how common a particular concept is within the dataset. It is a fundamental aspect of the video's analysis, as it quantifies the occurrence of medical procedures. The script uses the example of a 'Radiologic exam chest two views' to illustrate how concept prevalence is displayed and analyzed.

💡Tabular Display

A 'Tabular Display' is another method of presenting data, as mentioned in the script when switching to the 'table tab'. It lists concepts in a table format, showing concept IDs, names, and prevalence data. This is an alternative way to view information that complements the tree map and is part of the video's exploration of data representation.

💡Drill Down

'Drill Down' refers to the process of selecting a specific piece of data to view more detailed information. In the video, this term is used to describe how users can click on a row in the table to explore further details about a concept, such as the prevalence of 'Radiologic exam chest two views' among patients.

💡Stratification

Stratification in the context of the video is the process of organizing data into groups based on certain characteristics, such as year, age, and gender. It is used to analyze concept prevalence across different demographics, as shown in the graphs that display data trellised by age deciles and separated by gender.

💡Seasonal Pattern

A 'Seasonal Pattern' is identified in the video's analysis of medical procedure data, indicating that certain procedures, like chest radiologic exams, occur more frequently during specific times of the year. The script mentions a pattern of increased exams in the winter months, providing insight into seasonal trends within the data.

💡Provenance of Data

'Provenance of Data' pertains to the origin or source of the data within the dataset. The video discusses how to identify where data comes from, such as inpatient and outpatient claims records, using a donut plot to show the distribution of a concept's occurrence in different record types.

💡Cumulative Frequency Distribution

A 'Cumulative Frequency Distribution' is a statistical representation shown in the video that illustrates the number of occurrences of a concept for given individuals. It helps to understand the frequency at which individuals experience a particular medical procedure, such as the varying numbers of chest x-rays received by patients.

💡Search Filter

'Search Filter' is a feature within Atlas that allows users to search for specific concepts within the dataset. The video demonstrates how to use the search filter to find concepts like 'coronary artery bypass', highlighting the platform's functionality for targeted data exploration.

Highlights

Exploration of the procedure report within the data sources capability of Atlas.

Access to the procedure report through the Atlas homepage and data sources list.

Tree map representation of the prevalence of all concepts in the procedure domain.

Concept prevalence indicated by the size of boxes and intensity by color in the tree map.

Availability of a tabular display of concept prevalence and details under the 'table' tab.

Line listing in the table includes concept ID, name, and number of persons with the concept.

Capability to sort, search, and filter the table for detailed information.

Drill-down feature to find further information about a selected concept.

Example given: Radiologic exam chest two views frontal and lateral with 13 million patients.

Statistical data on the average occurrence of the chest radiologic exam concept.

Graphs showing concept prevalence stratified by year, age, gender, and index year.

Identification of a seasonal pattern in chest radiologic exams occurring in winter.

Visualization of the provenance of data for radiologic exams in inpatient and outpatient claims.

Donut plot showing the proportion of record appearances in outpatient records.

Age at first occurrence of a concept stratified by gender and represented in box plots.

Cumulative frequency distribution graph representing the number of occurrences of a concept.

Drill-down information available for all concepts within the procedure domain.

Filter feature for searching specific concepts such as 'coronary artery bypass'.

Summary of the procedure report and information about Atlas and the Odyssey community.

Transcripts

play00:01

[Music]

play00:09

today we're going to explore the

play00:11

procedure report within the data sources

play00:14

capability of Atlas from the Atlas

play00:17

homepage we can go to data sources and

play00:20

from the data sources list we can select

play00:24

the source of data that we want to

play00:26

explore as well as the report that we

play00:28

would like to explore here we're

play00:31

selecting the procedure report once we

play00:34

select the procedure report we are

play00:36

brought to a tree map representation of

play00:38

the prevalence of all concepts in the

play00:41

procedure domain with inside the OU mop

play00:43

common data model each of these boxes

play00:46

represents the concept prevalence the

play00:48

size of the box is the concept

play00:50

prevalence and the color of the box

play00:51

represents the intensity of those

play00:54

particular concepts a tabular display of

play00:57

this information is also available under

play00:59

the table tab if I select that tab we

play01:02

see a line listing that represents each

play01:04

concept ID its name and then the number

play01:08

of persons who have had that concept

play01:09

that concept prevalence and the records

play01:12

per person this shows 27,000 different

play01:17

concepts within this particular source

play01:20

this table can be sorted or searched and

play01:24

filtered for any information if I select

play01:27

any row in this table we can drill down

play01:30

to find further information about that

play01:32

concept here for example if I select the

play01:36

concept of Radiologic exam chest two

play01:41

views frontal and lateral we can see

play01:44

that there are 13 over 13 million

play01:45

patients and 16 percent of this data

play01:47

source that have that particular concept

play01:49

and on average those that have a

play01:51

radiologic exam of the chest have that

play01:55

concept 2.3 times if I scroll down we

play02:00

can gain additional information about

play02:02

how this concept plays out in this data

play02:04

source the first graph shows the concept

play02:07

prevalence this information is

play02:09

stratified by year age NJ

play02:13

so each graph shows in index year on the

play02:17

x axis it is trellis by age deciles and

play02:22

the gender is represented by the line

play02:24

series with blue indicating males and

play02:26

pink indicating females the y axis of

play02:30

each of these plots shows the concept

play02:31

prevalence per thousand persons so here

play02:34

we can see for the concept of Radiologic

play02:36

exam of the chest we can see that this

play02:39

concept appears in the data from the

play02:41

years 2000 through 2017

play02:44

it appears that chest radiologic exams

play02:47

occur more in older individuals than

play02:50

younger individuals occurs slightly more

play02:52

in women were relative to men and seemed

play02:56

to have had a slight increase in the

play02:59

year 2009 the next graph down shows the

play03:03

prevalence of this concept by month here

play03:06

the x-axis represents the calendar month

play03:11

and the y-axis represents the concept

play03:13

prevalence per thousand persons here we

play03:16

can see a seasonal pattern in when chest

play03:18

radiologic exams occurring in the winter

play03:21

of each year the next graph shown is the

play03:26

type concept ID this identifies where

play03:29

the provenance of the data is for this

play03:31

given source and here we can see that

play03:34

radiologic exams are occurring in both

play03:36

inpatient and outpatient claims records

play03:39

for this particular data source and if

play03:42

we hover over any element within this

play03:45

donut plot we can see the proportion of

play03:47

where this record appears in this

play03:50

particular source we can see that the

play03:52

prevalence of this concept appears most

play03:54

commonly in the outpatient record in the

play03:56

fourth position representing 89% of the

play03:59

data records the next graph shows you

play04:02

the age at the first occurrence of one

play04:04

of these concepts it is stratified by

play04:06

gender and we show age as on the y axis

play04:09

so here if I hover over any of these box

play04:13

plots we can see a distribution of the

play04:16

values here indicating that the median

play04:18

age of people at the first time that

play04:20

they've had

play04:21

chest x-ray is 50 and then finally the

play04:25

graph below shows us a cumulative

play04:26

frequency distribution representing the

play04:29

number of occurrences of a particular

play04:31

concept for given individuals so here

play04:34

whereas 16 percent of the people had at

play04:37

least one occurrence of a chest x-ray we

play04:40

can see that 8 percent of these people

play04:42

had 2 or more all the way down to some

play04:44

individuals having 7 or 8 chest x-rays

play04:49

within the data this drill down

play04:52

information is available for all

play04:53

particular concepts within the procedure

play04:56

domain in the top right filter if I'm

play04:59

interested in searching for any

play05:01

particular concept such as coronary

play05:04

artery bypass I can type in that word

play05:07

and the set of concepts will be filtered

play05:10

to those that meet that string search

play05:13

here I can see coronary artery bypass

play05:15

has occurred in 140,000 individuals in

play05:18

this database selecting that concept

play05:21

allows me to drill down to information

play05:24

to further explore the prevalence of

play05:27

that concept by year age gender by month

play05:31

by provenance of data in by age this

play05:35

includes our summary of the procedure

play05:36

report for more information about the

play05:40

data sources capability atlas as a whole

play05:43

or anything about the entire odyssey

play05:45

community check us out at Odyssey org

Rate This

5.0 / 5 (0 votes)

Related Tags
Health AnalyticsData SourcesAtlas PlatformProcedure ReportConcept PrevalenceData VisualizationDemographic InsightsMedical TrendsData FilteringHealthcare Data