Towards an On-line Handwriting Recognition Interface for Health

Service Providers using Electronic Medical Records

Viktor Mikhael M. Dela Cruz, Christian E. Pulmano and Ma. Regina Justina E. Estuar

Ateneo de Manila Univeristy, Katipunan Avenue, Quezon City, Metro Manila, Philippines

Keywords:

Electronic Health Records, Handwriting Recognition, Health Informatics, Usability.

Abstract:

The 2019 Universal Health Care Act in the Philippines has allowed healthcare service providers to have a

second look at using electronic medical records (EMRs) in their practice with tools that enable servicing the

poorest of the poor and coursing payments via EMR. A review of ﬁrst world country narratives, however, show

evidence of the substandard usability of EMRs. Physician work is impeded as almost two-thirds of consultation

time is spent documenting on an EMR instead conversing with patients face-to-face. This paper describes a

handwriting recognition interface for EMR data entry that is user-friendly and is unobstructive to the patient-

physician relationship. An initial prototype tested by medical students showed a handwriting recognition

accuracy of 34% while a second testing by health service providers showed a handwriting recognition accuracy

of 42%. Findings show that recognition is challenged by specialized words and accidental markings which

cause extra spaces and extra symbols. Additional features to the system as well as possible augmentations to

improve accuracy and efﬁciency through ontology, machine learning, and AI are also roadmapped.

1 INTRODUCTION

The promulgation of the 2019 Universal Health Care

Act (UHC) in the Philippines ushers a new era of

health care in the country (Congress of the Philip-

pines, 2018). In line with leveraging the latest in

health technology to facilitate efﬁcient health services

delivery, section 36 of the Universal Health Care Act

requires health service providers and insurers to uti-

lize the electronic health record (EHR), interchange-

ably referred to as electronic medical record (EMR)

(Congress of the Philippines, 2018).

An EMR is an electronic record of an individ-

ual’s health-related information that can be managed

by ofﬁcial clinicians and staff of a health care orga-

nization (Horowitz et al., 2008). It contains patient

information such as diagnoses, medicines, tests, al-

lergies, immunizations, and treatment plans (National

Cancer Institute, 2019). Physicians are able to access

core functions such as viewing, documentation and

care management, ordering, messaging, analysis and

reporting, patient-directed functionality, and billing

through EMRs (Smelcer et al., 2009).

Barriers such as high initial ﬁnancial costs, slow

and uncertain ﬁnancial payoffs, and high initial physi-

cian time costs deter medical practitioners from uti-

lizing EMRs (Miller and Sim, 2004) despite EMRs

providing many beneﬁts (Hillestad et al., 2005; Men-

achemi and Collum, 2011; Ben-Assuli et al., 2013;

Blumenthal and Glaser, 2007) and being implemented

on a global scale (McConnell, 2004). One underly-

ing cause of barriers to using EMRs is their poor us-

ability as most EMRs are not intuitive and have in-

terfaces that are not user-friendly (Miller and Sim,

2004; Smelcer et al., 2009; Belden et al., 2009;

Hill Jr et al., 2013). Almost 3 in every 4 physicians

agree that EMRs contribute great stress that leads to

burnout. However, studies show that this stress can

be greatly reduced by improving user interfaces and

including the users into the system design process

(Stanford Medicine, 2018; Gardner et al., 2018).

A solution that may improve the usability of

EMRs, decrease the time spent by physicians encod-

ing in EMRs, and ultimately increase adoption and

usage rate of EMRs is by giving physicians the ability

to manually scribe their patient encounter during con-

sultation. Most EMRs only allow free-text typing as

an off-the-shelf way of data entry but many physicians

still prefer to write (Smelcer et al., 2009; Tsoromokos

et al., 2017; Arvary, 2002). Additionally, electronic

charting can take up to 238.4% longer than manually

writing on paper (Hill Jr et al., 2013; Poissant et al.,

2005). It makes sense, therefore, to integrate hand-

writing recognition as an out-of-the-box functionality

Dela Cruz, V., Pulmano, C. and Estuar, M.

Towards an On-line Handwriting Recognition Interface for Health Service Providers using Electronic Medical Records.

DOI: 10.5220/0008944403830390

In Proceedings of the 13th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2020) - Volume 5: HEALTHINF, pages 383-390

ISBN: 978-989-758-398-8; ISSN: 2184-4305

383

to modern EMRs as a way of unobtrusively replacing

old-fashioned paper charting without adding the extra

time taken to type nor the hassle of changing already

successful habituated workﬂow.

While handwriting recognition systems for

medicine have been implemented (Kumar et al.,

2016; Roy et al., 2017a; Roy et al., 2017b), only few

are concerned with on-line handwriting recognition

(Chen et al., 2010; Bandyopadhyay and Mukherjee,

2014; Holzinger et al., 2010). These systems, albeit

innovative and aim to contribute positively, may still

cause stress and prove to be counterproductive with-

out proper user feedback and assessment. This paper

describes a more user-friendly, less time-consuming,

and more usable means of data entry for EMRs

through a handwriting recognition interface that is

designed for the physician. The approach involves

utilizing open-source technologies and implementing

hardware-agnosticism to open possibilities for low

cost deployment of the solution. Furthermore,

iterative development phases are practiced to ensure

proper incorporation of the user in the design process.

The rest of the paper reports prototype initial re-

sults and ﬁndings as well as additional features and

possible augmentations that will be done in next it-

erations in order to improve handwriting recognition

accuracy and efﬁciency, and EMR usability.

2 HANDWRITING

RECOGNITION IN MEDICAL

CONTEXT

Innovative solutions to cumbersome data entry in the

ﬁeld of health are speech and handwriting recogni-

tion. In the case of the latter, there have been vari-

ous attempts to explore the subject matter and either

develop new software or optimize currently existing

solutions.

A survey was conducted to determine interest in

handwriting applications for EMRs (Arvary, 2002).

The survey was distributed over 411 primary care

physicians (PCPs) and garnered a total of 156 re-

sponses. The survey showed that 78% of the respon-

dents agreed that digital ink would be a useful sup-

plement to EMRs. Furthermore, no subgroup showed

less than 73% support for handwriting implementa-

tions on EMRs.

The DPP4BIT application is a web-based technol-

ogy that allows editing, management, import, and ex-

port of any document in digital form created for the

purpose of digital recording and handling of medical

equipment (Tsoromokos et al., 2017). Using Anoto’s

digital pen, DPP4BIT is able to perform on-line hand-

writing recognition on digital forms that are managed

by Health Care Units. A pilot test showed that 90% of

nursing staff found the setup absolutely user-friendly,

85% of users felt more conﬁdent using the digital pen,

and work efﬁciency and speed was highly improved.

One study proposed improving the recognition

component by incorporating medical knowledge into

the recognizer (Chen et al., 2010). This integration

provided a 5% increase in accuracy of recognition

from a ”BestConﬁdence” module that selects word

candidates with the highest conﬁdence (Chen et al.,

2010). This was achieved by augmenting the post-

processing phase of recognition on a semantic level

with a medical knowledge model.

Another study developed an on-line handwriting

recognition system that recognizes India’s second

most used language, Bengali, or Bangla, for the pur-

pose of web-based telemedicine in rural medical cen-

ters (Bandyopadhyay and Mukherjee, 2014). This

paved the way for low-cost, virtual medical consul-

tations between doctors and remote patients where

physical visits were either not needed or impossible

(Bandyopadhyay and Mukherjee, 2014). This was

achieved by utilizing shape-based features and em-

ploying pattern recognition techniques.

It is noteworthy that these studies have been im-

plemented in health systems in parts of the world such

as the United States, India, and Germany, and it can

be safely presumed that there are many more of the

same in other nations. This further strengthens the

need for an efﬁcient, effective, and satisfactory health

system within the Philippines.

3 PROTOTYPE

IMPLEMENTATION

On-line handwriting recognition involves accepting

handwriting input from a digital surface, analyzing

factors such as upward and downward strokes as well

as spaces and time from pen up and pen down, and

then processing the input in order to classify which

characters or words the input approximates to.

For this purpose, the graphical library MyScrip-

tJS was used (MyScript, 2019). MyScriptJS is a

JavaScript library that can be used in any web appli-

cation to bring handwriting recognition.

Handwriting recognition is requested by supply-

ing user-correspondent application keys and a pre-

ferred language to the appropriate API URL. This

request is sent to the MyScriptJS server for process-

ing by their recognizer. MyScript performs three key

processes simultaneously in order to ﬁnd the closest

HEALTHINF 2020 - 13th International Conference on Health Informatics

384

recognition result. These processes are symbol clas-

siﬁcation, segmentation, and linguistic-based analy-

sis. The ﬁnished result is sent back to the client

in MyScript’s JSON Interactive Ink eXchange (JIIX)

format.

Figure 1 illustrates a simple implementation of

the handwriting interface using MyScriptJS written

in JavaScript as accessed on a desktop PC. Users

are greeted with a generous writing area and a text

ﬁeld that shows the corresponding word recognition

of what they write after scribing.

Figure 1: MyScript implementation prototype accessed on

a desktop PC.

The interface was deployed on a private server us-

ing Google services as a means of testing and pro-

totyping. Current features of the prototype include

a full-ﬂedged handwriting recognition module, input

ﬁelds for correct text translations and optional identi-

ﬁers, and the option to either export data locally or to

a private cloud database. The basic ﬂow of a test is

depicted in Figure 2.

Figure 2: Flow of tasks during a prototype test.

4 RESULTS, DISCUSSION,

AND RESEARCH ROADMAP

4.1 Local Experiment

As an initial evaluation of the prototype’s capabili-

ties, students in their 3rd year of studying medicine in

the Philippines were asked to write down 5 common

words used or written by physicians during consulta-

tions. The students were given the choice of using the

prototype on a PC or on a mobile device. After their

testing, the students were asked to take note of the

words they wrote and what were actually recognized.

Figure 3 illustrates a sample use case of the proto-

type. The 10 students who participated in the testing

were able to produce a total of 50 samples which were

manually categorized into four classiﬁcations: accu-

rate, wrong, ”with extra characters”, and ”with extra

spaces”. From the 50 samples, the prototype was able

to achieve an accuracy of 34%. Breaking the num-

bers down, the prototype was able to get 17 out of 50

recognitions correct while it got 16 out of 50 recog-

nitions wrong. Furthermore, 10 out of 50 recogni-

tions, albeit correct, were padded with extra charac-

ters and 7 out 50 recognitions, even though correct,

were padded with extra spaces.

Three students used the prototype on a PC while

the other seven tested it on a mobile device. However,

this was not enough to produce signiﬁcantly different

results. Those who used PCs with the mouse to write

were able to get similar results as those who used mo-

bile devices with ﬁngers to write.

Figure 3: Sample handwriting test case.

After further observation of the local experiment

handwriting samples, some conjectures made to ex-

plain the errors include accidental markings such as

can be seen in the bottom left of Figure 4, slant hand-

writing direction such as in the case of Figure 5,

and poor recognition handling of the letter ”i” such

as in the case of Figure 6 which shows overempha-

sized superscript dots on the letter i’s. With these in

mind, punctuation marks with dots such as the ques-

tion mark (?) and the exclamation point (!) may cause

problems with recognition. Additionally, words writ-

ten on an unequal number of lines may also challenge

the handwriting recognition.

4.2 Real-world Scenario Experiment

In an attempt to gather data from real-world scenar-

ios, a second testing of the prototype was conducted

by health professionals during a series of clinic visits.

Towards an On-line Handwriting Recognition Interface for Health Service Providers using Electronic Medical Records

385

Figure 4: Handwriting test case with wrong recognition due

to accidental marking.

Figure 5: Handwriting test case with wrong recognition due

to slant handwriting.

Figure 6: Handwriting test case with wrong recognition due

to overemphasized superscript dots.

Participants were composed of allied health profes-

sionals working in health establishments based in the

Western Visayas region of Philippines. A total of 15

clinics were visited and 9 respondents agreed to test,

2 of which were nurses, 3 were midwives, and 4 were

doctors.

Participants of the test were asked to write a ran-

dom ﬁrst name, a random diagnosis, and a random

prescription. The ﬁrst case was to assess the proto-

type’s name recognition, while the second case was to

assess the extent of medical-related terms the proto-

type can recognize, and the last case was to assess the

prototype’s number and symbol recognition. More-

over, the participants were asked to write with their

pointer ﬁnger on a mobile phone sporting a 6.2 inch

display which effectively provided a writing surface

that was 4.5 inches in width and 2 inches in height at

any given time.

Data gathered included the correct words, the ac-

tual recognized words, and the respective stroke ob-

jects that composed each test. Upon test completion,

data collected was sent to a private database for stor-

age and further analysis.

Script and cursive handwriting styles were the two

handwriting styles observed where script was used

generally while cursive was used only by doctors.

Figures 7 and 8 illustrate good representatives of the

general look of the received script and cursive hand-

writing samples.

Figure 7: Sample test case written in script.

Figure 8: Sample test case written in cursive.

The 9 respondents were able to produce a total

of 27 phrases (9 names, 9 diagnoses, 9 prescriptions)

that comprised of 100 words. For the second testing,

the prototype was able to achieve an overall accuracy

of 42% where accurate means perfect recognition in-

cluding capitalization. Further analysis showed that

the prototype was able to achieve an accuracy of 55%

with names, 38% with diagnoses, and 0% with pre-

scriptions. Additionally, an accuracy of 48% and 29%

HEALTHINF 2020 - 13th International Conference on Health Informatics

386

on script and cursive handwriting styles respectively

were also achieved. However, extra characters and

extra spaces occurred more frequently in this round

of testing compared to the ﬁrst testing.

In the case of names, it was observed that com-

mon names were easily recognized but more unique

ones challenged the prototype. Moreover, in the case

of diagnoses, disregarding less detailed handwriting,

the prototype was able to recognize those composed

of common words such as ”Upper Respiratory Tract

Infection” and ”Dengue Fever” but had difﬁculties

with specialized ones such as ”Tonsillopharyngitis”.

Furthermore, in the case of prescriptions, the proto-

type performed poorly in recognizing symbols such

as the number sign (#), abbreviations such as ”BID”

or ”sig”, and numbers that were less deﬁned such as

the ”500” depicted in Figure 8.

4.3 Research Roadmap

The next steps involves exploring ways to improve

the current handwriting recognition implementation.

Two possible augmentations may be integrating med-

ical ontologies for word suggestions and comparing

recognitions to a local best approximate for word cor-

rection. These, as well as other additions, will be fur-

ther explicated in the succeeding sections.

At the end state of research, EMR data entry will

then only follow a simple 3-step process: (1) a physi-

cian writes down notes on the interface during consul-

tation, (2) the interface processes handwriting input

and trains itself, and (3) the interface extracts and au-

tomatically maps key clinical text to their respective

EMR ﬁelds. Figure 9 outlines the process ﬂow once

all features have been studied and implemented.

4.3.1 Medical Ontologies for Word Suggestion

A novel attempt in aiding handwriting recognition is

by introducing an artiﬁcial intelligence that searches

a medical ontology for related concepts that may aug-

ment the current text input written by the user. A med-

ical ontology is a model of the knowledge from a clin-

ical domain (Jovic et al., 2007). It contains all of the

relevant concepts related to the diagnostics, treatment,

clinical procedures and patient data. Essentially, on-

tologies are designed in a way that allows ﬂuid knowl-

edge inference and reasoning. This may allow the

handwriting recognition to provide quick text correc-

tion and beneﬁcial word suggestions as users scribe.

While medical ontologies have been extensively

studied and have been proven to be beneﬁcial (Za-

man et al., 2017; Mate et al., 2015; Cases et al.,

2014; Washington et al., 2009; Sarntivijai et al., 2016;

Maarouf et al., 2017; Scheuermann et al., 2009), the

application of these has yet been explored in the con-

text of text correction and word suggestion for med-

ical handwriting recognition systems. Medical on-

tologies used in the previous studies include the Hu-

man Phenotype Ontology (HPO) (Robinson et al.,

2008) and the Ontology for General Medical Science

(OGMS) (Scheuermann et al., 2009). The OGMS will

serve as the pilot medical ontology utilized in the con-

tinuation of this study.

The OGMS is a framework of terms that encom-

passes the ﬁeld of diseases, from causes and mani-

festations to diagnostic acts, as recognized and inter-

preted in the clinic (Scheuermann et al., 2009). It in-

cludes very general terms used in the clinical settings

such as ”patient”, ”diagnosis”, and ”disease”. This

will provide the handwriting recognition interface a

good backbone for ontological text input augmenta-

tion.

4.3.2 Local Best Approximate for Correction

MyScriptJS uses Interactive Ink in order to processes

user input in real time (MyScript, 2019). Through

Interactive Ink, given some handwriting, MyScriptJS

is able to transform each character into a stroke object

which includes an array of x and y’s which are the cur-

rent coordinates of the pointer on the surface, an array

of t’s which is the current timestamp of the pointer

event, an array of p’s which is the current pressure

information associated to the event, and a pointerId

which is an identiﬁer for the current pointer. Each of

the properties are recorded per unit of time until all

strokes are ﬁnished.

An attempt in improving the accuracy of the hand-

writing recognition can be made by adding a func-

tion that compares the MyScriptJS text translation to

a best approximate recognition from a local knowl-

edge base. The local knowledge base contains previ-

ous strokes and their corresponding text translations.

This will be built using data structures such as ar-

rays and objects using JavaScript but is still subject to

change once efﬁciency comparisons have been made

with data structure implementations from other pro-

gram languages.

The proposed process is depicted in Figure 10.

Strokes will simultaneously be processed by the

MyScriptJS server and by the local knowledge base.

The local knowledge base will check for an entry

that closely matches the current strokes. If there is a

match, the match will be offered to the user as a sug-

gestion. The user can then decide if the best approx-

imate from the local knowledge base is better than

the MyScriptJS text translation by either keeping the

current translation or selecting the suggestion. The

record is updated in the knowledge base or is added

Towards an On-line Handwriting Recognition Interface for Health Service Providers using Electronic Medical Records

387

Figure 9: Goal process ﬂow following simple 3-step EMR data entry.

to the knowledge base if it does not currently exist.

Through this, the knowledge base can learn over

time as more strokes are recorded along with their cor-

responding user-selected correct text. Additionally,

the handwriting recognition interface will be able to

give out more accurate suggestions. It is possible to

look at superseding the MyScriptJS text translations

with the local best approximates once a high-enough

conﬁdence level is achieved.

4.3.3 Clinical Text Extraction

Once the handwriting input has been converted, the

resulting plain text is fed into a clinical text analyzer

to be able to ﬁlter only the words, phrases, and sen-

tences that are important for the EMR to store. For

this purpose, the natural language processing system

cTAKES will be used (Savova et al., 2010). Devel-

oped by the Mayo Clinic, cTAKES is an open-source

NLP system that allows the extraction of information

from clinical free-text. cTAKES prides itself in be-

ing powerful, fast, scalable, modular, portable, and

free. Some of what it can do with clinical texts are

event discovery, negation and uncertainty detection,

time expression discovery, as well as detect certain

keywords that fall under a certain Uniﬁed Medical

Language System (UMLS) classiﬁcation.

4.3.4 Integration within an EMR

SHINEOS+ is a web and mobile-based system that

aims to address the data management needs of doc-

tors, nurses, midwives, and other allied health profes-

sionals in the Philippines (The Secured Health Infor-

mation Network Exchange, 2019). The SHINEOS+

EMR service sports a number of features such as

patient proﬁling and consultation recording, inter-

network referral, automated reminders, and ePre-

scriptions.

Within SHINEOS+, physicians are able to cre-

ate proﬁles for their patients as well as add consul-

tation entries for them. A patient record contains ba-

sic information such as name, gender, age, birthday,

and record location. Consultation data includes com-

plaints and vitals and physicals at time of consulta-

tion.

After ﬁltering the text from the handwriting inter-

face and extracting key clinical text, the resulting set

HEALTHINF 2020 - 13th International Conference on Health Informatics

388

Figure 10: Methodology for accuracy improvement.

of strings will be automatically mapped to their corre-

sponding ﬁelds in the SHINEOS+ EMR service, cre-

ating a seamless transition from handwritten script to

digitized text without breaking natural workﬂow.

5 CONCLUSIONS

This paper looked into the poor usability of modern-

day EMRs and described a handwriting recognition

interface that captures physician handwriting in real-

time and converts the digital input into plain text.

The interface was tested by various health practition-

ers and the results were reported. Future additions to

greatly extend the interface’s functionality, practical-

ity, and usability were also discussed.

The Universal Healthcare Act in the Philippines is

changing the nation’s healthcare landscape. With the

UHC in play, EMRs are now at the forefront of health

services. However, as beneﬁcial as EMRs are, the

rapid advancement of technology has changed their

requirement from being functional to being usable.

Inefﬁcient and unwieldy health systems cannot be tol-

erated in today’s fast moving world where patients

come by the dozens. It is in the best interest of this re-

search that physicians are given the proper tools to use

so that they can worry less about compliance and fo-

cus more on tending to others. A handwriting recog-

nition interface that can serve as a plug-in to existing

EMRs hopes to be beneﬁcial in providing a seamless

interface in doctor-patient interaction.

ACKNOWLEDGEMENTS

The researchers would like to thank the Ateneo

School of Science and Engineering as well as the De-

partment of Science and Technology for subsidizing

this study through the SOSE Industry Grant and the

DOST-ERDT grant respectively. Additionally, the re-

searchers would also like to thank the Ateneo Center

for Computing Competency Research (ACCCRE) for

their support.

REFERENCES

Arvary, G. (2002). A primary care physician perspective

survey on the limited use of handwriting and pen com-

puting in the electronic medical record. Journal of In-

novation in Health Informatics, 10(3):161–172.

Bandyopadhyay, A. and Mukherjee, B. (2014). Web based

bangla online handwritten prescription system for

telemedicine through multimedia conferencing. Inter-

national Journal of Advanced Research in Computer

Engineering & Technology (IJARCET), 3(12).

Belden, J. L., Grayson, R., and Barnes, J. (2009). Deﬁn-

ing and testing emr usability: Principles and proposed

methods of emr usability evaluation and rating. Tech-

nical report, Healthcare Information and Management

Systems Society (HIMSS).

Ben-Assuli, O., Shabtai, I., and Leshno, M. (2013). The im-

pact of ehr and hie on reducing avoidable admissions:

controlling main differential diagnoses. BMC Medical

Informatics and Decision Making, 13(1):49.

Blumenthal, D. and Glaser, J. P. (2007). Information tech-

nology comes to medicine.

Cases, M., Briggs, K., Steger-Hartmann, T., Pognan, F.,

Marc, P., Klein

oder, T., Schwab, C., Pastor, M.,

Wichard, J., and Sanz, F. (2014). The etox data-

sharing project to advance in silico drug-induced tox-

icity prediction. International journal of molecular

sciences, 15(11):21136–21154.

Chen, Q., Gong, T., Li, L., Tan, C. L., and Pang, B. C.

(2010). A medical knowledge based postprocessing

approach for doctor’s handwriting recognition. In

2010 12th International Conference on Frontiers in

Handwriting Recognition, pages 45–50. IEEE.

Congress of the Philippines (2018). An act instituting uni-

versal health care for all ﬁlipinos, prescribing reforms

in the health care system, and appropriating funds

therefor.

Gardner, R. L., Cooper, E., Haskell, J., Harris, D. A.,

Poplau, S., Kroth, P. J., and Linzer, M. (2018). Physi-

cian stress and burnout: the impact of health infor-

mation technology. Journal of the American Medical

Informatics Association, 26(2):106–114.

Towards an On-line Handwriting Recognition Interface for Health Service Providers using Electronic Medical Records

389

Hill Jr, R. G., Sears, L. M., and Melanson, S. W. (2013).

4000 clicks: a productivity analysis of electronic med-

ical records in a community hospital ed. The American

journal of emergency medicine, 31(11):1591–1594.

Hillestad, R., Bigelow, J., Bower, A., Girosi, F., Meili, R.,

Scoville, R., and Taylor, R. (2005). Can electronic

medical record systems transform health care? poten-

tial health beneﬁts, savings, and costs. Health affairs,

24(5):1103–1117.

Holzinger, A., Schl

”ogl, M., Peischl, B., and Debevc, M.

(2010). Optimization of a handwriting recognition

algorithm for a mobile enterprise health information

system on the basis of real-life usability research. In

International Conference on E-Business and Telecom-

munications, pages 97–111. Springer.

Horowitz, J., Mon, D., Bernstein, B., and Bell, K. (2008).

Deﬁning key health information technology terms.

Department of Health and Human Services, Ofﬁce

of the National Coordinator for Health Information

Technology.

Jovic, A., Prcela, M., and Gamberger, D. (2007). Ontolo-

gies in medical knowledge representation. In 2007

29th International Conference on Information Tech-

nology Interfaces, pages 535–540. IEEE.

Kumar, S., Sahu, N., Deep, A., Gavel, K., and Ghost, R.

(2016). Ofﬂine handwriting character recognition (for

use of medical purpose) using neural network.

Maarouf, H., Taboada, M., Rodriguez, H., Arias, M., Sesar,

A., and Sobrido, M. J. (2017). An ontology-aware in-

tegration of clinical models, terminologies and guide-

lines: an exploratory study of the scale for the assess-

ment and rating of ataxia (sara). BMC medical infor-

matics and decision making, 17(1):159.

Mate, S., K

opcke, F., Toddenroth, D., Martin, M.,

Prokosch, H.-U., B

urkle, T., and Ganslandt, T. (2015).

Ontology-based data integration between clinical and

research systems. PloS one, 10(1):e0116656.

McConnell, H. (2004). International efforts in imple-

menting national health information infrastructure and

electronic health records. World hospitals and health

services: the ofﬁcial journal of the International Hos-

pital Federation, 40(1):33–7.

Menachemi, N. and Collum, T. H. (2011). Beneﬁts and

drawbacks of electronic health record systems. Risk

management and healthcare policy, 4:47.

Miller, R. H. and Sim, I. (2004). Physicians’ use of elec-

tronic medical records: barriers and solutions. Health

affairs, 23(2):116–126.

MyScript (2019). About myscriptjs.

National Cancer Institute (2019). Deﬁnition of electronic

medical record.

Poissant, L., Pereira, J., Tamblyn, R., and Kawasumi, Y.

(2005). The Impact of Electronic Health Records on

Time Efﬁciency of Physicians and Nurses: A System-

atic Review. Journal of the American Medical Infor-

matics Association, 12(5):505–516.

Robinson, P. N., K

ohler, S., Bauer, S., Seelow, D., Horn,

D., and Mundlos, S. (2008). The human phenotype

ontology: a tool for annotating and analyzing human

hereditary disease. The American Journal of Human

Genetics, 83(5):610–615.

Roy, P. P., Bhunia, A. K., Das, A., Dhar, P., and Pal, U.

(2017a). Keyword spotting in doctor’s handwriting on

medical prescriptions. Expert Systems with Applica-

tions, 76:113–128.

Roy, P. P., Bhunia, A. K., Das, A., Dhar, P., and Pal, U.

(2017b). Keyword spotting in doctor’s handwriting

on medical prescriptions. Expert Systems with Appli-

cations, 76:113–128.

Sarntivijai, S., Vasant, D., Jupp, S., Saunders, G., Bento,

A. P., Gonzalez, D., Betts, J., Hasan, S., Koscielny,

G., Dunham, I., et al. (2016). Linking rare and com-

mon disease: mapping clinical disease-phenotypes to

ontologies in therapeutic target validation. Journal of

biomedical semantics, 7(1):8.

Savova, G. K., Masanz, J. J., Ogren, P. V., Zheng, J., Sohn,

S., Kipper-Schuler, K. C., and Chute, C. G. (2010).

Mayo clinical text analysis and knowledge extraction

system (ctakes): architecture, component evaluation

and applications. Journal of the American Medical

Informatics Association, 17(5):507–513.

Scheuermann, R. H., Ceusters, W., and Smith, B. (2009).

Toward an ontological treatment of disease and di-

agnosis. Summit on translational bioinformatics,

2009:116.

Smelcer, J. B., Miller-Jacobs, H., and Kantrovich, L.

(2009). Usability of electronic medical records. Jour-

nal of usability studies, 4(2):70–84.

Stanford Medicine, T. H. P. (2018). How doctors feel about

electronic health records.

The Secured Health Information Network Exchange (2019).

About shineos+.

Tsoromokos, D., Tsaloukidis, N., Zarakovitis, D., and

Lazakidou, A. (2017). Use of the dpp4bit system for

the management of hospital medical equipment. In

ICIMTH, pages 161–164.

Washington, N. L., Haendel, M. A., Mungall, C. J.,

Ashburner, M., Westerﬁeld, M., and Lewis, S. E.

(2009). Linking human diseases to animal models us-

ing ontology-based phenotype annotation. PLoS biol-

ogy, 7(11):e1000247.

Zaman, S., Sarntivijai, S., and Abernethy, D. R. (2017). Use

of biomedical ontologies for integration of biologi-

cal knowledge for learning and prediction of adverse

drug reactions. Gene regulation and systems biology,

11:1177625017696075.

HEALTHINF 2020 - 13th International Conference on Health Informatics

390