MOBILE TIMELINE EMR SYSTEM

Support System for Doctors’ Cognition/Analysis

Keisuke Ogawa, Kazunori Matsumoto, Masayuki Hashimoto, Tatsuaki Hamai

KDDI R&D Labs, Inc. 2-1-15 Ohara, Kamifukuoka, Saitama 356-8502, Japan

Yoshiaki Kondo

Tohoku University Graduate School of Medicine, Tōhoku, Japan

Wise Solutions, Inc., Plymouth, Michigan, U.S.A.

Keywords: Electronic medical records, Timeline, visualization of medical data, XEUS, Word completion, Adaptive

data mergence, Social patient list, Mobile, XML.

Abstract: In this paper, we propose a novel electronic medical record system (EMR) based on a brand new concept to

support doctors’ cognition and medical analysis wherever they are. Conventional EMR systems have the

advantage of helping doctors easily retrieve and manage mass medical records. On the other hand, medical

records have been expected to support doctors’ planning. Most conventional EMR systems don’t have an

appropriate function for such purpose, however. Because of its poor user interface, which is similar to

legacy medical records written on paper, they can’t help doctors analyze medical data that occurs

chronologically. To attain that purpose, the system has to have the ability to visualize medical data that

occurs over various time spans. This is because the relationship among the different medical data should be

observable when we look at it over various time spans. In addition, doctors aren’t always at their desks, so

they can’t always use EMR systems with a desktop PC. Therefore, in view of these problems, we propose a

system that has timeline interface which visualizes medical data that occurs over various time spans and its

client application works on a mobile device. In this manner, the system can support doctors’ cognition and

medical analysis wherever they are. In addition, we are verifying this system in the medical field.

1 INTRODUCTION

EMR systems are becoming popular in medicine

(A.L.Rector. 1996 , Anderson JD. 1999 , David W.

Bates et al. 2003 , Samuel J. Wang et al. 2003 , Jim

Johnson. 2010). This is because the system enables

doctors to manage mass medical data easily. As

represented by POMR (Weed LL. 1968), medical

record systems have been expected to support

doctors’ planning. But most of these EMR systems

are merely electronic data storage of legacy medical

records. For such purpose, the system has to have

the ability to visualize medical data that occurs over

various time spans. Because medical data often

occurs over various time spans, doctors have to

visualize it over various time spans for a medical

analysis. With these systems, doctors can look at

medical data for only a few days at most.

Accordingly, doctors can’t analyze medical data

effectively. In addition, there is another problem that

doctors don’t have much time to use the EMR at

their desks. As a result of these problems, doctors

desire a system that can support their cognition and

medical analysis anywhere. In view of these

problems, we propose a mobile timeline EMR

system that supports doctors’ cognition and medical

analysis. This system has the features listed below:

・ It has the ability to visualize medical data that

occurs chronologically over various time spans.

・ Its client application works on a mobile device

such as a mobile phone or a tablet PC.

・ In spite of the narrow bandwidth of wireless

mobile networks, the system responds quickly.

In this paper, we introduce the conventional EMR

systems in section 2 and reveal their problems. In

section 3, to solve these problems, we introduce the

mobile timeline EMR system and its technological

features.

405

Ogawa K., Matsumoto K., Hashimoto M., Hamai T. and Kondo Y..

MOBILE TIMELINE EMR SYSTEM - Support System for Doctors’ Cognition/Analysis.

DOI: 10.5220/0003287604050410

In Proceedings of the International Conference on Health Informatics (HEALTHINF-2011), pages 405-410

ISBN: 978-989-8425-34-8

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

2 CONVENTIONAL EMR

SYSTEMS

In this section, we introduce the conventional EMR

systems. They can be divided into two types used in

medicine as listed below:

・ A system that has a user interface similar to

legacy medical records written on paper (type-1).

・A system that has a user interface that summarizes

the data of medical examinations over the last few

days (type-2).

Figure1 shows the client application of a type-1

EMR system that displays SOAP information, the

patient’s information, and histories of medical

exams. With these systems, by treating medical data

as electronic data, doctors can search and manage

their patients’ medical data easily. This ability of

mass data management is a significant advantage

compared with legacy medical records.

Informationofthepatient

SOAP information

Historiesofmedicalexams

Figure 1: Conventional EMR system (type-1).

Figure2 shows the client application of a type-2

EMR system. It can display the data over the fixed

term. Due to the restriction of the size of the display,

in general, it displays the data over the short term.

However, there are various kinds of medical data

occurring over various time spans. So with these

systems, which can visualize medical data for only a

few days, doctors can’t always look at it and infer

the relationship among them. In other words, though

doctors can look at and understand the state of

patients who come two or three times with these

systems, they never can look at and understand the

state of patients who are suffering for years, such as

asthmatics, diabetics or patients suffering from

hypertension. Accordingly, these systems don’t

attain the purpose of supporting doctors’ analysis

and cognition.

InformationofthepatientMedicalEvents

Medicaldatainthefixedshortterm

Figure 2: Conventional EMR system (type-2).

3 MOBILE TIMELINE EMR

SYSTEM

As described above, in order to meet the demands of

supporting doctor’s cognition and medical analysis,

the system must have a function to visualize medical

data that occurs over various time spans and allow

doctors to look at medical data from any perspective.

So we introduced timeline interface to solve this

problem. Timeline interface visualizes the medical

data in chronological order. Timeline interface has a

multistage time scale such as years, months and days,

etc. With timeline interface, the system can visualize

chronological data over various time spans.

In addition, to meet the demands of mobility and

portability, we use a mobile device as a client of this

system. By adopting a mobile device, the system

gains a significant advantage in that doctors can

inspect and analyze the data anywhere, but mobile

devices have some problems as listed below:

(1) Difficulty with input and reading with a small

display

(2) Low data transmission rate through mobile

wireless network

To solve problem (1), we adopted the functions

listed below:

・ Advanced word completion using optimized

lexicon for each medical branch.

・ Social patient list that aims to reduce the input by

reusing the search histories in each medical branch.

HEALTHINF 2011 - International Conference on Health Informatics

406

For problem (2), we adopted the functions listed

below:

・ An Adaptive data mergence function that merges

neighboring data objects adaptively. This reduces

the amount of data transmitted.

・ An XML document encoding system that

reduces the amount of data transmitted by

transforming the XML plaintext to small binary data.

In this manner, we can realize the tool based on the

concept of supporting doctors’ cognition and

analysis wherever they are. These functions are

described below in detail.

3.1 Timeline Interface

Timeline interface is the most important part of this

system. Figure3 shows it. Doctors can change the

time scale to various time units by controlling the

bar on the upper part. For example, by clicking the

bar, doctors can change the day unit time scale to the

month unit scale or the hour unit scale. By changing

the time scale to a smaller unit scale, doctors can

observe medical data over a short time span in detail.

Conversely, by changing the time scale to a bigger

unit scale, doctors can look at medical data over a

long time span. Also, we can change the length of

the time unit by pinch-in/pinch-out.



Wecanchangethetimescaleeasily.

Zoomout

Zoomin

Namesofmedicaleventssuchasmedication,

disease

ins

ectionandsoon

Dataobjectswhichstandfor

medicalevents

Figure 3: Timeline interface.

The reasons why we use this interface are described

below:

・ There are various kinds of medical data. They

occur chronologically and there are often several

relationships among them.

・ When there is a relationship among the different

data, the appropriate time scale to observe exists. By

selecting the time scale properly, the system can

visualize the relationship among the data.

For example, take the case where we can discover

the relationships when we observe the data over a

long time span, even if we can’t discover it over a

short time span. Conversely, we can’t discover it

with an overly long time span. Accordingly, the

system must have a function that allows users to

select the appropriate time scale.

With this timeline interface, doctors can change

the time scale as they wish. Therefore, doctors can

observe the various relationships between various

data. In other words, doctors can inspect medical

data from various points of view. Figure4 shows an

example of observable relationships.

Sametimespan

ValueB

MedicationA

Wecan’tobserveanyrelationshipswiththe

shorttimescale.

We can discover the relationships with the long time

scale.Inthiscase,we canobservea possibilityofthe

fact that stopping the” medication A” causes a

decreaseo



“

valueB”.

MedicationA

ValueB

Figure 4: Visualization of the relationships among medical

data.

In this manner, this system can be not only a

management tool for medical data, but also a tool for

supporting doctors’ cognition and medical analysis.

3.2 Word Completion using Lexicon

for Medical Data

The input method is not only an important factor that

decides the usability of the system on mobile

devices, but also a difficult problem. This is because

mobile devices only have poor input accessories

MOBILE TIMELINE EMR SYSTEM - Support System for Doctors' Cognition/Analysis

407

such as small touch panels and keyboards. In

particular, in EMR systems, doctors have to input

special characters for medical treatment using these

poor input devices to write down the SOAP

information or to search patients. To solve this, it is

common knowledge that the word completion

method using a lexicon of medical words is effective

(Laird S. Cermak et al 1992, C. G. Chute et al 1999,

Hiroyuki Komatsu et al 2001). However, the words

used in medicine differ significantly among each

branch. In other words, there is a problem that using

the same lexicon among all the branches is

insufficient. For example, the phrase nephrotic

syndrome is often used by pediatricians, but is rarely

used by ophthalmologists. Accordingly, we optimize

the lexicon for each branch. Simply put, we changed

the bias of the TRIE (Donald R. Morrison 1968)

structure of the lexicon for each branch. Then for

each branch, by summarizing and analyzing the

doctors’ input history commonly, the system

succeeded in improving the accuracy of word

completion. Figure5 shows an example of the TRIE

structure in Japanese.

Oneletter

hangingtheweight

betweenletters

Onewor



Figure 5: TRIE structure for lexicon of medical words in

Japanese.

Here, we use φ for the head of the sentence and

$ for the tail of the sentence. By adding word

frequency of doctors’ own input history, word

frequency of other doctors’ input history in the same

branch and the cost of prediction to the tail of the

sentence, we can make the TRIE structures for each

branch.

3.3 Social Patient List

In this subsection, we introduce another function for

the purpose of reducing input, which is called the

social patient list in this system. By using the social

patient list, doctors can search for patients using

useful queries without using a keyboard that is

difficult to use. Social patient list is a function that

enables doctors to save search queries as a list of

patients and have them in common in each branch.

In other words, doctors in the same branch can

utilize the useful search histories of other doctors as

if they were their own. The figure presented below is

an example of a social patient list. By clicking the

display button, doctors can easily reuse the search

queries of other doctors.

Searchqueriesofotherdoctorsinthesamebranch

Figure 6: Social patient list.

3.4 Adaptive Data Mergence

The response speed of the system is a very important

factor for deciding the system’s usability. The

advantage of timeline interface is, as we described

above, the ability to visualize the relationships of

medical data with various time scales. If doctors

want to look at the data in the long time span, they

can enlarge the time scale as they wish. In this

manner, however, the system has to display a lot of

data objects at once. On the other hand, in this

system, since its client is a mobile device, the client

only has narrow wireless communication bandwidth.

In order to improve the response speed in this

system, we use adaptive data mergence. Adaptive

data mergence is a function that merges neighboring

data objects adaptively. We use the formula below

as a threshold to merge data objects.

___ ___

Start time of objectA end time of objectB

threshold x

timescale

−

As we described in Figure7, when doctors enlarge

the time scale, if the time gap of the data objects is

smaller than the threshold, the system merges the

objects to one object. In this manner, the system can

reduce the amount of data and improve the response

speed and usability.

HEALTHINF 2011 - International Conference on Health Informatics

408

hangingthetimescale Somedataobjects

Representingasintegratedone

objectautomatically

Figure 7: Adaptive data mergence.

In addition, there are various types of graphs for

the data objects and the best graphs for the data

objects are different from each other. If doctors want

to observe the flow of medication data, a line graph

is the best representation. In another example, if

doctors want to look at the frequency of the

examination data, a histogram is the best

representation. Adaptive data mergence has the

ability to select the best representation automatically

by learning the history in the same branch. As

represented below in Figure8, the system has the

ability to change the representation depending on the

data.



Representingthenumberofobjects

Somedataobjects

Representingthefigure

hangingthetimescale

Figure 8: Various graph representations for data objects.

3.5 XML Document Encoding System

Recently, MML (Kenji Araki et al. 2000 , Jinqiu

Guo et al 2003), which is the standard data format to

transmit medical data between two different EMR

systems, has been proposed. MML is a format based

on XML. In addition, in this mobile timeline EMR

system, we use a data format based on MML. The

common XML document is the plaintext data, so it

provides the advantage of high readability and high

extensibility. XML also results in a disadvantage in

respect to data size, safety and parse processing load

in terms of data reception, however. Therefore, in

this system, we solve these problems by adopting the

XEUS (Kobayashi Arei et al. 2001) XML document

encoding system. XEUS is the abbreviation of XML

document Encoding Universal Sheet and a system

for encoding XML documents. Now, we describe

the procedure of the XEUS system.

step: Define the XEUS sheet which is a document

that has the logical structure and a table of the

encoding. The XEUS sheet is described by XML.

step: By using this XEUS sheet in the encoder,

we can encode the large plaintext XML document

into a small binary document. This is because the

encoder can separate the plaintext XML document

into the logical structure and serialized data, and

transforms it into small binary data. Then in the

decoder, by using the same XEUS sheet, which

defines the logical structure, we can decode the

small binary document. Figure9 shows the concept

of the XEUS system.

Narrowbandwidth

Broadbandwidth

Figure 9: XEUS system.

In this mobile timeline EMR system, the EMR

server that has medical data has the encoder and the

mobile device that displays the EMR data has the

decoder. In general, XML documents result in the

data becoming one-fifth the original size. This is a

significant help for the client of this system, which

only has narrow wireless communication bandwidth.

MOBILE TIMELINE EMR SYSTEM - Support System for Doctors' Cognition/Analysis

409

4 CONCLUSIONS

In this paper, we proposed an EMR system based on

the brand new concept. This system has a significant

advantage to support doctors’ cognition and medical

analysis wherever they are. The system has the three

features described below.

・ Since this system has timeline interface, doctors

can look at and analyze medical data by using

various time scales. This is a significant help for

doctors’ cognition and analysis.

・ Since its client is a mobile device, the system can

support doctors’ cognition and medical analysis

wherever they are.

・ In spite of using a mobile device as a client, this

system guarantees ease of use as much as possible.

In addition to the features described above, making

use of the advantage of easy use, there is a

possibility that the doctors can use this system as an

educational tool.

REFERENCES

A. L. Rector., 1996. Computer-based Patient Records,

Yearbook of Medical Informatics, Section 2, 195-198.

Anderson J. D., 1999. Increasing the acceptance of clinical

information systems. MD Computing.16(1)62-5.

David W. Bates, Mark Ebell, Edward Gotlieb, John Zapp,

and H. C. Mullins., 2003. A Proposal for Electronic

Medical Records in U.S. Primary Care, J Am Med

Inform Assoc.

Samuel J. Wang, Blackford Middleton, Lisa A. Prosser,

Christiana G. Bardon, Cynthia D. Spurr, Patricia J.

Carchidi, Anne F. Kittlera, Robert C. Goldszer, David

G. Fairchild, Andrew J. Sussman, Gilad J. Kuperman,

David W. Bates,2003. A cost-benefit analysis of

electronic medical records in primary care. The

American Journal of Medicine.

Jim Johnson., 2010. Making A Successful Transition To

Electronic Medical Records. Healthcare Technology

Online.

Weed LL., 1968. Medical records that guide and teach.

New England Journal of Medicine 278:593-600.

Laird S. Cermak, Mieke Verfaellie, Marie Sweeney and

Larry L. Jacoby, 1992. Fluency versus conscious

recollection in the word completion performance of

amnesic patients, Brain and Cognition Volume 20,

Issue 2, Pages 367-377.

C. G. Chute, P. L. Elkin, D. D. Sherertz, and M. S.

Tuttle,1999. Desiderata for a clinical terminology

server, Proc AMIA Symp.

Hiroyuki Komatsu, Akira Takabayashi, Toshiyuki Masui,

2001. Predictive Text Input with Japanese Dynamic

Abbreviation Expansion Method “Nanashiki”, WISS.

Donald R. Morrison,1968. PATRICIA-Practical

Algorithm To Retrieve Information Coded in

Alphanumeric, jacm.

Kenji Araki, Katsuhiro Ohashi, Shunji Yamazaki,

Yasuyuki Hirose, Yoshinori Yamashita, Ryuichi

Yamamoto, Kazushi Minagawa, Norihiro Sakamoto

and Hiroyuki Yoshihara ,2000. Medical Markup

Language (MML) for XML-based Hospital

Information Interchange, Journal of Medical Systems.

Jinqiu Guo, Kenji Araki, Koji Tanaka, Junzo Sato,

Muneou Suzuki, Akira Takada, Toshiaki Suzuki,

Yusei Nakashima and Hiroyuki Yoshihara, 2003. The

Latest MML (Medical Markup Language) Version 2.3

XML-Based Standard for Medical Data

Exchange/Storage, Journal of Medical Systems.

Kobayashi Arei, Takagi Satoru, Muramatsu Shigeki, Baba

Akira, Matsumoto Kazunori, Inoue Naomi, 2001. The

general XML document encoding system "XEUS",

IEIC Technical Report. (Institute of Electronics,

Information and Communication Engineers).

HEALTHINF 2011 - International Conference on Health Informatics

410