Evaluating Health-care-related Active Learning Class Lectures using
Class Achievement and Text Mining of Free Descriptions
Kazuma Mihara, Takahito Tamai and Yukie Majima
Graduate School of Humanities and Sustainable System Sciences, Osaka Prefecture University, Japan
Keywords: Class Evaluation, Active Learning, Text Mining, Correspondence Analysis, Health Care Related Subjects.
Abstract: Active learning is defined as "a general term for professors and learning methods that incorporate
participation in active learning of the students, unlike teachers' unilateral lecture style education."
Universities that improve classes from the viewpoint of active learning are increasing in recent years. A
class evaluation questionnaire has been established to improve the understanding and satisfaction of
students' classes. In many cases, the Likert scale is used for the class evaluation questionnaire. There are
also aspects for which statistical processing is easy to do. However, it is difficult to ascertain the students '
specific opinions and ideas alone. Therefore, we attempted to evaluate health-care-related subjects from the
two viewpoints of ‘free description’ and ‘degree of accomplishment of class goal’ for active learning classes
aimed at students' subjective learning.
1 INTRODUCTION
In recent years, many universities in Japan have
adopted classes incorporating class evaluation
questionnaires and active learning. However, there
are indications that active learning classes are
attracting attention only to activities. They do not
engender learning improvement
(Matsushita, 2017).
Active learning in Japan is a generic name of a
professor or learning method that incorporates
participation in active learning by a student, unlike
the unidirectional lecture style of education by
teachers. It is defined as a means of training
universal capabilities including cognitive, ethical,
social abilities, cultural knowledge, knowledge, and
experience with active learning by students,
including discovery learning, problem solving
learning, experiential learning, survey learning, etc.
Group discussion, debate, group work, and other
methods within the classroom are also effective
active learning methods (Central Council, 2012).
Some reports describe effects on students' degree of
comprehension (Nekoda, 2012). Depending on the
class design, it is clear not only from Japan but also
from research conducted around the world that it
affects students' understanding and learning
motivation in various ways (Matthew-Maich et al.,
2016). Moreover, overseas universities are using
methods such as Problem Based Learning (PBL) as
one method of active learning, which is more
effective than lecture-based learning (Faisal et al.,
2016). Furthermore, in overseas medical education,
information and communications technologies (ICT)
and Technology Enabled Active Learning (TELE)
are used actively (McCoy et al., 2016). In overseas
research, active learning is said to help acquire
knowledge and establish knowledge in the field of
medical education (Graffam, 2007). Active learning
has effective aspects for the field of medical
education, but it is important to clarify whether
students are understanding contents well by
conducting proper evaluations when doing active
learning classes. In Japan, class evaluation
questionnaires have also been conducted from long
ago as part of class improvement: a five grade Likert
scale is often used for evaluation. Although this
scale is readily adaptable to statistical processing, it
is difficult to grasp concrete opinions and ideas from
such a scale. Therefore, we used free description. It
is extremely difficult to read enormous numbers of
entries carefully and assess them individually.
Furthermore, even assuming careful perusal in
understanding the text, the risk of subjective bias of
an analyst remains. Therefore, for this study, text
mining of free descriptions was conducted with
quantitative analysis of questionnaire results.
456
Mihara, K., Tamai, T. and Majima, Y.
Evaluating Health-care-related Active Learning Class Lectures using Class Achievement and Text Mining of Free Descriptions.
DOI: 10.5220/0007572704560461
In Proceedings of the 12th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2019), pages 456-461
ISBN: 978-989-758-353-7
Copyright
c
 2019 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved

2 RELATED RESEARCH
Analyses of free description parts of class evaluation
questionnaires have been done with text mining
many times in the past. Etchu (2015) attempted to
capture the overall trend of free description of
students while ensuring objectivity by summarizing
and presenting data, with multivariate analysis.
Matsukawa (2018) conducted a topic model analysis.
By linking extracted topics with free description data,
he presented information using a cross table for
features of topic distributions of subject groups
compared with the overall trend. In fact, although
many studies have been undertaken to visualize free
descriptions and elucidate trends, finding specific
factors to improve classes remains difficult.
Therefore, one must consider students' degree of
comprehension and satisfaction to the greatest
degree possible. Nevertheless, few analyses have
made such a class evaluation (Matsukawa and Saito,
2011). This study evaluated active learning classes
by assessment of items of class achievement goals
and free description data and by consideration of the
students' degree of comprehension, and evaluating
their mutual relation.
3 HOW TO PRACTICE ACTIVE
LEARNING CLASSES
3.1 Subject Course
For this study, we will examine three classroom
practices used by the same teacher. Two of these are
elective subjects for information system students:
‘health care system’ and ‘health care service’. These
are similar in class systems and have a small class
size of 20–30 people. ‘Epidemiology’ is a
compulsory subject for nursing care training courses
for nursing students. The students are mainly 100
people with simultaneous classes.
3.2 Active Learning Method
Active learning is defined as "a general term for
professors and learning methods that incorporate
participation in active learning of the students,
unlike teachers' unilateral lecture style education."
Various methods are used under the title of active
learning. Mizokami examined the active learning
method separately for lecture type classes and
practical type classes. (Mizokami, 2007) Therefore,
we classified the three classes to be analyzed this
time according to classification examples proposed
by Mizukami. Table 1 presents a summary of active
learning methods adopted for the respective classs.
‘Health care system’ and ‘health care service’ are
aimed at active learning, so they adopt group work.
‘Epidemiology’ emphasizes knowledge
conservation, and urges the creation of quizzes to
sustain knowledge necessary to pass a national
examination.
In particular, as group work, all three classes
have "Today's news presentation". This is that
designated students select topics of their own
interests about each class's them and share them
among students in presentation style.
Table 1: Active learning method used in class.
3.3 Achievement Level of the Class
Goal
In viewing the relevance of this analysis, we
specifically examined the ‘degree of
accomplishment’ self-evaluated by the students at
the end of the class for ‘class goals’. There are seven
items in ‘epidemiology’, five items in ‘health care
services’, and three items in the ‘health care system’.
The students evaluated self-assessed whether
each class goals were achieved in five-point Likert
scale.We used the average of them as personal class
goal's score.The items in each class achievement of
objectives are presented in Tables 2, 3, and 4.
Table 2: Epidemiology class objective.
Class objective
①
Can understand and explain concepts and basic terms of epidemiology.
②
Explain epidemiology frequency and indicators of risk or effect.
③
Can explain the epidemiological investigation method.
④
Explain the principle and method of mass screening.
⑤
Can explain the main demographics and health statistics.
⑥
Can explain the frequency and distribution of major diseases, risk factors and prevention.
⑦
Explain the importance of the epidemiological viewpoint in public health nurse activities.
Evaluating Health-care-related Active Learning Class Lectures using Class Achievement and Text Mining of Free Descriptions
457

Table 3: Health care service class objective.
Class objective
①
Through active learning, you can improve general -purpose abilities including cognitive,
logical, social skills, education, knowledge, and experience.
②
Think about the future modes of healthcare services and can practice one.
③
Understand and explain the fundamental human body structure, organ function s and
pathology necessary for basic knowledge of healthcare, international standard disease
classification, and clinical nursing.
④
Understand and explain the roles, duties and concrete measures of health care professionals
concerning the collection, storage, transmission, information disclosure and use of personal
information related to health care.
⑤
Can understand and explain general knowledge related to health care field (insurance /
medical / welfare) so that you can communicate smoothly with healthcare professionals.
Table 4: Health care system class objective.
Class objective
①
Understand the circumstances and the current state of information policies in our country,
about the necessity and application method of imaging communication technology to provid
e
advanced health care service (insurance / medical / welfare /) guaranteeing effective
picture and quality.
②
For the current needs of healthcare users and the problems that they hold (subject privacy
and protection of personal information, unification of health data and secondary usage
method.). Based on the development of information technology, cultivate the ability to create
concrete solutions and future developments.
③
Learn practical basic knowledge and basic technology skills related to health care systems to
plan, develop, operate, and maintain systems that operate in the healthcare field.
4 ANALYTICAL METHOD AND
RESULTS
4.1 Analysis Target
‘Health care system’, ‘health care service’, and
‘epidemiology’ were classified as subjects. The
target period of the healthcare system was the four
years of 2015–2018. The total questionnaire number
was 156, of which the free description response rate
was 42%. In healthcare service, the total number of
questionnaires was 104, with a free description
column response rate of 35% for the three years of
2015–2017. In epidemiology, for the seven years of
2011–2017, the questionnaires were 876, with a free
description response rate of 32%. For all 1136 cases
above, a 34% free description response rate was
obtained. Table 5 presents the duration of each class,
the number of aggregations, and the free listing
response rate.
4.1.1 Class Form
First, we analyze ‘health care system’, ‘health care
service’ and ‘epidemiology’ in two stages. ‘Health
care system’ and ‘health care service’ are elective
subjects with small classes. ‘Epidemiology’ is a
compulsory subject with students of different classes
participating. All are assessed together: the same
faculty adopted the active learning method.
Table 5: Class evaluation questionnaire number of total.
Class Period
Total
Questionnaire
Free description
response rate
Health care system 2015-2018 156 65(42%)
Health care
services
2015-2017 104 36(35%)
Epidemiology 2011-2017 876 282(32%)
Total 1136 383(34%)
4.1.2 Analytical Method
A summary of the analysis is presented in Figure 1.
(1) Sort three class evaluation questionnaires. As a
method of classification, averaging each item of the
achievement level of class goal and take an average
score for all the students for each class. Table 6
shows the average points of all participants in each
achievement level of class goal. For the average
points of each class goal, more than four points of
students are ‘high’ (upper group), more than 3.01
points are ‘intermediate’ (middle group), and 3
points or fewer are ‘low’ (lower group). The average
points of respective classes are presented in Table 7.
(2) Morphological analysis is applied to extract
feature data from sentences. Words are extracted.
Subsequently unnecessary parts of speech and
numbers are removed and the data are organized. (3)
Using data arranged for classification and analysis
by appropriate numerical processing, use
correspondence analysis to derive each feature
quantity. (4) To visualize the analysis results, we
plot the feature quantities obtained using
‘correspondence analysis’ on two dimensions and
consider their mutual relation.
① Class valuation questionnaire results classified
② Extract feature data (keywords etc.) from sentences
③ Classification and analysis
④ Visualization of analysis results
Figure 1: Analysis summary.
HEALTHINF 2019 - 12th International Conference on Health Informatics
458

Table 6: Average points of target achievement by
category.
Group
Health care
System
Health care
Services
Epidemiology All classes
Top group 4.31 4.24 4.27
Middle group 3.58 3.48 3.5
Low group 2.94 2.71 2.74
4.1.3 Analysis by Teaching Form
In the correspondence analysis described later, when
analyzing data by class form, no association was
found between the class achievement target degree
and free description data. The analysis combining
data of all three classes clarified that active learning
is linked to the degree of comprehension.
4.1.4 Overall Analysis
Using the morphological analysis tool RMeCab (12)
for the upper, middle and lower groups described in
4.1.2, divide the free description data into
morphemes (minimum units of language). From that,
only nouns were extracted. The occurrence
frequency of words was investigated. The numbers
on the graph represent the frequency of occurrence
of the nouns as a whole. The ratio by group is shown
in Figs. Table 7 shows the numbers of people in
respective groups and the numbers of nouns.
0.063
0.045
0.028
0.019
0.017
0.016 0.016 0.016
0.015
0.013
C
L
ASS
TES
T
EV E
R
Y
T
IM
E
P
RE
S
E
NT
A
T
I
ON
UND ER
S
TA N
DI
N
G
R
E
VI E W
E
P
ID EMI
OL O GY
MY
S
EL
F
TE
A
C
H
E
R
H
E
A
LT
H
C
AR E
Figure 2: Percentage of word appearance frequency of the
high group.
0.070
0.040
0.030
0.020 0.020 0.020 0.020
0.010 0.010 0.010
TE
S
T
C
L
ASS
EV E
R
Y T I
M
E
EP IDE M
I
OL
OG
Y
MY S EL
F
UND ER STA ND ING
RE
VI E W
ST U
D
Y
T
EA C
H
E
R
C
LI
C
KE
R
Figure 3: Percentage of word appearance frequency of
middle group.
2.850
1.600
1.000
0.800
0.610
0.600
0.550
0.500 0.500
0.450
T
E
ST
CLA SS
EV E RY
TI M E
EP
I
DE
MI
O
L
O
GY
PR
E
S
E
N
TA
T
ION
ST U DY
RE
VI E W
UN
DE
R
STA ND I NG
MY S EL F
T
E
AC H ER
Figure 4: Percentage of word appearance frequency of
lower group.
Table 7: Average point of target achievement by category.
Category
Number of
p
eo
p
le
Total noun
numbe
r
Top group
143
1341
Middle group
156
1525
Low group
76
620
4.2 Correspondence Analysis
Correspondence analysis is a technique for assessing
the relation between two discrete variables. Because
it can be replaced with a simple data matrix without
losing information of the original data, one can
clarify the structure of
a complicated data matrix.
When plotting the analysis results, a strong
correlation exists in the category level where the
coordinate points are close to one another: words
Evaluating Health-care-related Active Learning Class Lectures using Class Achievement and Text Mining of Free Descriptions
459

that are more relevant to categories are shown closer
together;
weaker words are shown farther away.
5 CONSIDERATION
We analyzed separately for ‘healthcare system and
healthcare service’ and ‘epidemiology’. Results
show no association between the class achievement
degree and free description data. In ‘health care
system, health care services’, the data are extremely
small. That sparseness of data can be regarded as a
factor because it is a class with few people.
Therefore, the active learning method adopted by the
same teacher analyzed all class data as the same
class. Results can be portrayed as shown in FIG. 5.
The high-ranking group was placed on the left side,
the intermediate group was placed on the upper right,
and the lower group was on the lower right. As
closer to the position where it is arranged, it
becomes a word that characterizes the group. Words
gathered in the center that are not particularly
different between groups. Table 8 shows examples
of words that characterize each group. Keywords
surrounded by the large red circle on the upper left
are middle frequency terms with the upper group. As
inferred from the arrangement of the keywords, the
middle and upper groups include ‘active learning’
and ‘group’. Many words related to the active
learning method are presented in Table 1, such as
‘work’ and ‘clicker’. These groups described the
active learning method to a great degree. When these
words were searched from the original free
description data of the middle and upper group,
almost all of the descriptions were positive
descriptions related to active learning. From the
above, one can infer that active learning functions
effectively for the middle and upper groups for the
degree of class achievement. These analyses
demonstrate that the group that appreciated the
degree of achievement of the class goal more
frequently expressed ‘words about the active
learning’ method compared with the group with a
low evaluation. By contrast, one method is not
described at all. It is possible to give more effective
feedback such as examining teaching materials.
The average value of the achievement level of
the class target used this time is a subjective
evaluation of each student, which is not necessarily
the same as the actual degree of understanding or
knowledge acquisition.
Therefore, future studies are needed in order to
examine the introduction of objective understanding
level indicators.
Figure 5: Corresponding analysis result.
Table 8: Words characterizing each group.
Top &
Middle
Low
Center
System The Study News
Healthcare Research Description
Group Usually Understanding
Lecture Slide Class
Active Learning Score Meaning
Life Weekly Textbook
Field Nurse Epidemiology
Clicker Homework Interest
6 SUMMARY
This study used text mining from free description
data and class achievement to evaluate health-care
related classes using active learning. Based on
comparison, we quantitatively ascertained
characteristics of student groups and score
differences.
Analysis using free description and class
achievement evaluation will help to ascertain
whether active learning is effective for learning in
health care related classes, or not. The analysis
included classes with few data. Beneficial results
were not obtained from analyses for the respective
classes. For this reason, the same learning method
was adopted by the same teacher for the same class.
Then we assessed data of the three classes together.
Future studies will apply free description data
collection methods to assess each class.
HEALTHINF 2019 - 12th International Conference on Health Informatics
460

ACKNOWLEDGEMENTS
This work was supported by JSPS KAKENHI Grant
Numbers JP 17H04433.
REFERENCES
Etchu, K. 2015. Analysis of Questionnaire Evaluation
Questionnaire by Text Mining: Attempt to
Visualize Free Description by Co-occurrence
Network, Bulletin of Miyagi University of
Education Center for Information Processing
Center, 22, pp.67-74
Central Council: qualitative change report (2012).
Faisal, R., Khalil-ur-Rehman, Bahadur S, et al. 2016.
Problem-based learning in comparison with
lecture- based learning from medical students, J
Pak Med Assoc. 66 (6): 650-3.
Graffam, B. 2007. Active learning in medical education:
Strategies for beginning implementation, Med
Teach. , 29 (1): 38-42.
Matthew-Maich N, Hammond C, Sheremet D, Roche
C.2016.Nursing students' perceptions of
effective problem-based learning tutors, Nurs
Stand. Nov 16;31(12):48-59.
McCoy L, Pettit RK, Lewis JH, Bennett T, Carrasco N, et
al.2015. Developing Technology-Enhanced
Active Learning for Medical Education:
Challenges, Solutions, and Future Directions,
The Journal of the American Osteopathic
Association,Vol.115,202-
211.doi:10.7556/jaoa.2015.042
Matsushita K., 2017.Deep Active Learning ins Science
Education: Focusing on the Practice and
Research of Conceptual Change, Science
education research, 41, (2),77-84
Nekoda Y., 2012. The use of clicker in epidemiology
lecture and student's reaction, Journal of
Japanese Society for Nursing Research, 35, 137-
143
Matsukawa H, Saito T., 2011. Development and
Evaluation of a Reporting System of the Student
Course Evaluation Utilizing Data and Text
Mining Technology, Journal of the Japan Society
for Educational Technology, 35 (3), 217-226
Matsukawa H., 2018, Analysis of Free Descriptions of
Course Evaluation Questionnaires Using Topic
Mode, Journal of the Japan Society for
Educational Technology Engineering, 41 (3),
233-244
Mizokami S., 2007, Problems associated with the
introduction of active learning, Nagoya Higher
Education Research, 7, 269 - 287
RMeCa: (http://rmecab.jp/wiki/index.php?RMeCab）
(Last viewed date，2018/9/27)
Evaluating Health-care-related Active Learning Class Lectures using Class Achievement and Text Mining of Free Descriptions
461