Where do we go from here? Evaluating the effectiveness of a nursing MOOC over six years

Abstract

Background

Massive open online courses (MOOCs) have transformed global education, yet their long-term effectiveness and evolving learner engagement remain underexplored. This study aims to comprehensively evaluate a nursing MOOC over six years, examining learner engagement, identifying distinct learner profiles, and assessing changes across different developmental stages to inform future MOOC design.

Methods

A retrospective study was conducted on 4171 completers of the Medical Nursing MOOC on a Chinese MOOC platform, covering eleven semesters from 2018 to 2023. Latent profile analysis (LPA) categorized learners based on unit test scores, and profile distributions were compared across the MOOC’s developmental stages.

Results

The Medical Nursing MOOC attracted 69,642 registrants with a 5.99% completion rate. Among the 4171 individuals who completed the course, latent profile analysis identified six distinct learner types, demonstrating significant differences in overall learning effect (H = 2823.604, P < 0.001). The chi-squared analysis revealed significant differences between the proportions of the six profiles regarding MOOC developmental stages (χ² = 695.464, P < 0.001).

Conclusion

Findings highlight the evolving role of MOOCs in nursing education. Despite challenges in long-term engagement, the increasing proportion of highly engaged learners and declining dropout rates indicate growing effectiveness and sustainability. These insights provide evidence-based guidance for optimizing MOOC design and implementation.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12909-026-08679-w.

Introduction

A growing interest has been observed in the use of massive open online courses (MOOCs), as witnessed in global Google trends over the last 15 years (see Fig. 1). It also reveals that the development of MOOCs can be divided into several stages. The number of MOOC-led online learning courses and the user base they attract have grown substantially [1]. Furthermore, around 2019, the public policies of social distancing due to the COVID-19 pandemic made online learning a necessity, which has indelibly marked the evolution of MOOCs. China was among the first countries to bear the brunt of the pandemic, and with over forty-seven million students engaging in higher education, the crisis became a massive natural experiment for utilizing MOOC-led online learning courses to facilitate remote education.

Fig. 1.

Google Trends for MOOC related keywords

After years of practice during this global pandemic, MOOCs have become a significant and standard component of credentialed University education in China, especially for high-level universities such as those designated as “985/211/double first-class universities”. The formats of MOOCs have matured and diversified, including online and offline hybrid education models [2]. With the advancement in computer science, sophisticated technologies such as artificial intelligence [3] and visual analytics approaches [4] have been used to improve the quality and human-computer interaction of MOOCs.

MOOCs have a considerable impact on the efficiency of higher education [5], which is due to their ability to break down barriers to education [6]. They can provide learners with high-quality and flexible online courses that are accessible regardless of time or location and without any financial cost. This equalizes learning opportunities for students from diverse educational backgrounds, mitigating the negative impact of the Matthew effect in traditional education where students with higher achievement in the past have greater access to learning from better teachers [7]. Additionally, MOOCs are recognized as an effective method for delivering continuing education to enhance the skills of healthcare professionals throughout life-long learning [8, 9].

However, there have long been numerous concerns regarding the effectiveness of MOOCs. In contrast to traditional classes, MOOCs impose few restrictions on admission, withdrawal, submission of assignments and assessments [10]. As a result, the completion rates among MOOC learners are generally low, ranging from 0.7% to 52.1%, with a median rate of 12.6% [11]. However, completion rates alone do not fully reflect the effectiveness of MOOCs [12], factors such as reach, learning outcomes, and learners’ performance are equally important [13]. Researchers are increasingly directing their attention toward understanding the learning behavior characteristics of MOOC participants [10, 14–16], and it has become a mainstream approach for evaluating learning effectiveness, which is essential for determining the value of MOOCs [17]. At the same time, traditional variable-centered approaches often overlook heterogeneity among learners and may lead to biased parameter estimates. In contrast, person-centered methods such as Latent Profile Analysis (LPA) enable researchers to capture this heterogeneity by identifying distinct subgroups of individuals with similar response patterns. The classification schemes derived from LPA are conceptually meaningful and methodologically valuable for data-driven development and integration of learner typologies. Given the diversity of learner engagement and performance in MOOCs, LPA offers a theoretically grounded and empirically supported approach to uncovering meaningful learner profiles that cannot be observed using aggregate scores alone.

Currently, a significant challenge confronting MOOCs is determining their future trajectory, especially given the limitations in existing literature regarding their effectiveness across the process of adoption. Examining how MOOCs are adopted and adapted over time not only deepens our understanding of their current outcomes and limitations but also provides insights for optimizing course design and management. Accordingly, this study aims to comprehensively evaluate the performance of a nursing MOOC in enhancing learning outcomes and practical applications, analyzing data from 2018 to 2023 while identifying patterns and pathways for future development. Therefore, the research questions of this study are:

1. What distinct learner profiles can be identified in the nursing MOOC based on learners’ performance and engagement patterns?

2. How do the distributions of learner profiles and engagement patterns change across different developmental stages of the MOOC?

The medical nursing MOOC

Medical Nursing is a foundational professional course for undergraduate nursing students and a core component of clinical nursing education. It emphasizes the holistic assessment and management of patients with internal medical conditions. As a discipline essential for both undergraduate and continuing nursing education, it plays a critical role in the long-term professional development of nurses. The Medical Nursing MOOC was conceptualized before the COVID-19 pandemic and officially launched in November 2018. The course is delivered twice annually, with each term spanning 15 weeks. This study analyzes all eleven semesters of the MOOC offered from its inception to December 2023.

The curriculum framework was led by the chief editor of the Medical Nursing national textbook, who served as the primary course designer. The teaching team included nursing educators from four leading universities in Guangdong Province and clinical nursing specialists from affiliated hospitals. The course covers seven major textbook chapters, encompassing 24 common internal medical diseases and specialized diagnostic and therapeutic nursing procedures. A structured set of learning resources was provided for each chapter, including instructional videos, formative assessments, and written materials. The MOOC contains 51 teaching videos with a total duration of 618 min. To evaluate learners’ performance, the course employs an extensive assessment system comprising a multi-item test bank, seven unit tests (7,178 items), and a final examination (225 items). The design and reporting of this MOOC followed the TIDieR framework to ensure a comprehensive and transparent description of the educational intervention. Detailed information on course content and structure is available in Supplementary File 1.

Methods

Design

This study was a retrospective observational study using anonymized data, and the data were collected from user-generated records of the Medical Nursing MOOC platform (https://www.icourse163.org/learn/SYSU-1003467010).

Participants

The Medical Nursing MOOC was conducted over eleven semesters, including the spring and autumn semesters from 2018 to 2023. All learners who registered for the course, regardless of whether they received a comprehensive evaluation, were included in the descriptive baseline count. Among them, only those who received a comprehensive evaluation were viewed as having completed the MOOC and were included in the LPA.

Outcome indicators

Overall course completion rate

Learners with a total evaluation were viewed as having completed the Medical Nursing MOOC. The overall course completion rate was defined as the number of learners who completed the course divided by the total number of registered learners.

Learning effect for unit tests

Learning effect was defined as the learners’ scores in each unit test. Due to variations in question numbers and full scores among unit tests, this study aimed to calculate standardized scores to reflect general performance. To construct a standardized score that ranged from 0 to 100, the original scores of each unit were divided by the possible score and multiplied by 100.

Overall learning effect

The overall learning effect was defined as the learners’ scores in the final exam, which covered material from the entire course. This study aimed to calculate standardized scores to reflect the overall learning effect. To construct a standardized score that ranged from 0 to 100, the original score of the final exam was divided by the possible score and multiplied by 100.

Policy-defined MOOC developmental stages

This study examined the evolution of the MOOC across four policy-defined stages, which were determined based on the major shifts in China’s social-distancing policies during different periods [18–20]. Although these stages were externally shaped by public policy, they also indirectly correspond to the natural progression of the course’s development and implementation over time. The four stages included: the germination stage (2018.01 ~ 2019.12), strategic opportunity stage (2020.01 ~ 2020.06), maturity stage (2020.07 ~ 2022.12), and sustainable stage (2023.01 ~ 2023.12).

Ethical considerations

The platform has desensitized learner information, so neither teaching staff nor other learners could see the personal information of any individual learner. The dataset used in this study was fully anonymized by the MOOC platform prior to being provided to the research team. No identifiable personal information was accessible at any stage of the analysis. This study was approved by the Department of Education Administration and the ethics committee of Sun Yat-sen University.

Data analysis

All data were analyzed using Mplus V.8.0 and SPSS V.26.0. The tests described below were two-sided, with p < 0.05 indicating statistically significant results.

First, Latent Profile Analysis (LPA) with continuous indicators of the scores of seven-unit tests was performed in Mplus 8.0. LPA is a person-centered approach that identifies mutually exclusive subgroups of a population that share common characteristics [21]. To determine the optimal number of latent profiles, a single-profile model was computed at first, and then the number of latent profiles gradually increased to compare fitting indices between models. Model-fitting indicators were as follows: (1) Akaike information criteria (AIC), Bayesian information criteria (BIC), and adjusted Bayesian (aBIC) information criteria, smaller values indicated a better-fitting model [22]; (2) Lo-Mendell-Rubin (LMR) and Bootstrapped likelihood ratio test (BLRT). A significant p-value suggested that the k model was better than the k-1 model [22]; (3) entropy was an index for estimating the quality of profile assignments. Entropy above 0.8 implies accurate classification. A higher entropy indicates a more accurate classification of the model [22]. Overall, we sought a model with lower values for all criterion indices, but higher entropy values. Statistical indicators in combination with the substantive meaningfulness of profiles guided the optimal model selection [23]. Importantly, these indicators do not merely ensure statistical adequacy; they also enhance educational interpretability by supporting the identification of learner groups that reflect meaningful differences in learning performance. In this way, the selected model is both statistically robust and theoretically aligned with the goal of distinguishing heterogeneous learner patterns within the MOOC context.

Second, SPSS V.26.0 was used to analyze the relationship between the latent profiles, MOOC developmental stages, and overall learning effect. Chi-square analysis was conducted to assess whether profiles differed as a function of MOOC developmental stages. The correlation coefficient between different categorical variables was calculated using the Pearson contingency coefficient. Post-hoc analysis (Z-test) based on the Chi-square test was also performed to compare proportion-based measures. Kruskal-Wallis test was conducted to analyze the differences in overall learning effect between different profiles, and post-hoc multiple comparison tests were performed using the Kruskal-Wallis one-way ANOVA (k samples) test. The Bonferroni method was used to adjust p values for multiple variables.

Results

Overall course registrant number and completion rate

A total of 69,642 learners registered for the Medical Nursing MOOC. Among them, 4171 learners completed the course, yielding an overall course completion rate of 5.99%. The enrollment figures and learners’ completion rates for the 11 semesters are illustrated in Fig. 2.

Fig. 2.

Registrants number and completion rate of MOOC learners for 11 semesters

Course complete number, pass number, and pass rate

The number of completed MOOC learners, along with the passing rates across the four stages, is illustrated in Fig. 3. The maturity stage exhibited the highest pass rate at 75.15%, while data from the sustainable stage revealed similarly high results. In comparison to the germination stage, there was a significant increase in the passage rate over time.

Fig. 3.

Complete number, pass number, and pass rate of MOOC learners for 4 stages

LPA results for MOOC learners

The findings of the latent profile analysis for determining the optimal profile solution are presented in Table 1. Models ranging from one to eight profiles were estimated. First, model selection indices (i.e. AIC, BIC, and aBIC) were used to identify the most suitable number of latent MOOC profiles. The index values decreased as the number of latent profiles increased from one to eight, indicating that as the profiles increased, the model fitting degree also improved. However, there was a noticeable flattening out or diminishing improvement in fit between Profiles 7 and 8, suggesting that retaining less than 8 profiles may be more appropriate. Similarly, according to the LMR test results, the model with 7 profiles had a significantly better fit (P = 0.002) compared to the model with 6 profiles, with no further improvement in fit when additional profiles were considered (P = 0.575). Furthermore, entropy values for each model exceeded 0.98 which indicated highly discriminating profiles for all models.

Table 1.

Latent profile model fit comparison (n = 4171)

Profiles	AICa	BICb	aBICc	Entrophy	LMRd P-value	BLRTe P-value	Profile propertion (%)
1	301746.98	301835.68	301791.19	/	/	/	/
2	271114.23	271253.62	271183.72	0.989	P < 0.001	P < 0.001	40.71/59.29
3	263705.81	263895.89	263800.56	0.987	P < 0.001	P < 0.001	33.83/53.25/12.92
4	259520.15	259760.91	259640.17	0.990	P < 0.001	P < 0.001	32.10/56.34/51.67/10.59
5	256811.07	257102.52	256956.35	0.989	P < 0.001	P < 0.001	10.67/24.74/8.66/6.40/49.53
6	254419.86	254762.00	254590.41	0.990	P < 0.001	P < 0.001	9.72/24.64/ 7.24/ 6.35/3.51/ 48.53
7	252095.56	252488.38	252291.37	0.991	0.002	P < 0.001	9.66/22.51/4.77/3.79/3.62/48.43/7.22
8	250422.55	250643.64	250643.64	0.990	0.575	P < 0.001	3.64/22.73/6.31/4.48/2.59/7.72/4.68/47.85

Note:

^aAIC Akaike Information Criterion

^bBIC Bayesian Information Criterion

^caBIC Adjusted Bayesian Information Criterion

^dLMR Lo-Mendell-Rubin Test

^eBLRT Parametric Bootstrapped Likelihood Ratio Test

Lastly, based both on the model fit indices and theoretical plausibility, we have decided to retain 6 profiles for the following reasons: (a) Two profiles of the 7 latent profiles model were derived from one profile in the 6 latent profiles model, and these two profiles exhibited very similar characteristics. (b) Six profiles allowed for more meaningful and diverse interpretations compared to fewer profiles.

The individual profiles are illustrated in Fig. 4. Based on varying performance levels on the total score of the unit tests, all six profiles were first categorized into three levels. The six profiles were then named according to the different learning performances in the whole learning process. The detailed characteristics of each latent profile are presented in Table 2.

Fig. 4.

Latent Profile indicators mean values for the six-profile solution

Table 2.

Characteristics of each latent profile (n = 4171)

Latent profiles	n	%	mean ± SD
Level 1: Poor performance	1435	34.40%	15.585 ± 10.731
Early dropout learners	1027	24.62%	11.058 ± 5.948
Mid-term dropout learners	408	9.78%	26.981 ± 11.602
Level 2: Mediocre performance	709	17.00%	58.905 ± 13.231
Catch-up learners	264	6.33%	53.891 ± 14.137
Anticlimactic learners	146	3.50%	57.825 ± 12.568
Committed learners	299	7.17%	63.860 ± 10.732
Level 3: Excellent performance	2027	48.60%	95.308 ± 5.927
Committed learners

Note: SD standard deviation

Comparison of overall learning effect across latent profiles

The Kruskal-Wallis test revealed significant differences in the final exam scores of the learners comparing the six profiles (H = 2823.604, P < 0.001), and the results are shown in Table 3. Subsequent post-hoc tests revealed that each profile exhibited significantly different scores for the final exam, except for when comparing the “Poor performance: early dropout learners” and “mid-term dropout learners” profiles, “Mediocre performance: catch-up learners” and “Mediocre performance: committed learners” profiles. Additionally, it was found that learners characterized by dropout behaviors performed more poorly on the final exam than those who did not, representing the difference in the overall learning effect across latent profiles.

Table 3.

Comparison of the mean scores in the final exam across latent profiles (n = 4171)

		Median score (25th percentile, 75th percentile)	Kruskal-Wallis test		Post-hoc comparison
			H	P
Learners’ performance profiles			2823.604	< 0.001	a < d < c **
Poor performance	Early dropout learnersa	0.00 (0.00, 0.00)			a < e < f **
	Mid-term dropout learnersb	0.00 (0.00, 0.00)			b < d < c **
Mediocre performance	Catch-up learnersc	66.00 (0.00, 84.00)			b < e < f **
	Anticlimactic learnersd	0.00 (0.00, 68.50)			d < e < f **
	Committed learnerse	64.00 (36.00, 76.00)			c < f **
Excellent performance	Committed learnersf	96.00 (88.00, 98.00)

Note: Superscript letters (a–f) are used to denote latent profiles. **: P < 0.01

Associations between MOOC developmental stages and profiles

The chi-squared test revealed significant differences between the proportions of the six profiles regarding MOOC developmental stages (χ² = 695.464, P < 0.001), and the contingency coefficient revealed moderate correlation (C = 0.378, P < 0.001). Post hoc analysis (Z-test) based on the Chi-square test was also performed to compare proportion-based measures and the results are shown in Table 4. The proportion of different profiles has changed significantly across various MOOC developmental stages, which are mainly manifested in the following trends:

Table 4.

Z-test for differences in latent profiles for different stages (n = 4171)

Learners’ performance profiles			MOOC developmental stages				Total
			Germination stage	Strategic opportunity stage	Maturity stage	Sustainable stage
Poor performance	Early dropout learners	Count	86a	96b	191c	35b	408
		% within column	16.90%	7.40%	10.80%	5.90%	9.80%
	Mid-term dropout learners	Count	210a	306b	399b	112b	1027
		% within column	41.30%	23.70%	22.50%	18.80%	24.60%
Mediocre performance	Catch-up learners	Count	21a	204b	54a	20a	299
		% within column	4.10%	15.80%	3.00%	3.40%	7.20%
	Anticlimactic learners	Count	49a	158a	53b	4c	264
		% within column	9.60%	12.20%	3.00%	0.70%	6.30%
	Committed learners	Count	38a	54b	44c	10c	146
		% within column	7.50%	4.20%	2.50%	1.70%	3.50%
Excellent performance	Committed learners	Count	105a	472b	1034c	416d	2027
		% within column	20.60%	36.60%	58.30%	69.70%	48.60%
Total		Count	509	1290	1775	597	4171
		% within column	100.00%	100.00%	100.00%	100.00%	100.00%

Note: Different subscript letters indicate significant differences in the proportion of profiles in different MOOC developmental stages (P < 0.05)

For early dropout and mid-term dropout learners who performed poorly, their proportion decreased significantly since the strategic opportunity stage of MOOC and remained stable. This shows that in the early and middle stages of course development, the phenomenon of dropping out is more likely, but as the course enters the later stage of development, the proportion of dropping out gradually shrinks.

For the three types of learners with mediocre performance, Catch-up learners and Anticlimactic learners had relatively low proportions in the germination stage (4.1% and 9.6%, respectively). Their proportions increased in the strategic opportunity stage (15.8% and 12.2%) but dropped significantly in the maturity stage (3.0% and 3.0%) and the sustainable stage (3.4% and 0.7%). Committed learners accounted for 7.5% in the germination stage but showed a decreasing trend in the later stages—4.2% in the strategic opportunity stage, 2.5% in the maturity stage, and 1.7% in the sustainable stage. Overall, the proportion of mediocre-performing learners was relatively high in the early stages of MOOC development but gradually declined as courses became more mature.

Lastly, there was a significant increase in the proportion of excellent performance: committed learners over time, ontinuing from the initial 20.60% and reaching 69.74% during the sustainable stage. This indicates that with the development and maturity of MOOCs, the proportion of learners who persist and perform well increases significantly.

Discussion

The findings of this study provide valuable insights into the evolution of coverage, learner engagement, retention and performance in MOOCs at different stages of development. By analyzing the changes in completion rates, passing rates, and learner profiles, this study reveals clear trends that highlight the progressive refinement of MOOCs over time. Throughout the development of MOOCs, there has consistently been skepticism regarding the role of MOOCs in modern education [24], especially nowadays, as traditional offline classes regain prominence in mainstream higher education. However, our findings could significantly enhance the confidence among the developers and administrators of MOOCs, although there are some areas that need further improvement.

Firstly, the overall course completion rate of 5.99% is consistent with prior researches on MOOCs [11], highlighting the ongoing challenge of retaining MOOCs learners. Moreover, it is important to acknowledge that not all registrants necessarily intend to complete the comprehensive evaluation. Some learners may enroll with the goal of acquiring only specific knowledge or skills, and therefore achieve their personal learning objectives without participating in the final assessment. This phenomenon is common in MOOCs and may partly explain the discrepancy between the number of registrants and the number of course completers. Another challenge is that the total number of learners has decreased over time, which may relate to the return of traditional classrooms and the diversion of similar courses in the market. However, it is important to highlight that the number of learners participating in MOOCs significantly exceeds that of those in traditional classroom settings, further underscoring the advantages of MOOCs in attracting large-scale learners and providing equitable access to high-quality educational resources [6]. At the same time, the data reveal a significant Long Tail Effect. Certain niche online courses, such as Medical Nursing tailored for specific segments, may have a relatively small audience. However, due to their ability to meet the needs of particular groups, their cumulative benefits should not be underestimated. The course pass rates have remained consistently high, this indicates an improvement in both learner success and course effectiveness. It also suggests that the demographic composition of MOOC participants may have evolved. Specifically, while overall enrollment figures may be lower, the remaining learners exhibit greater focus and engagement. This observation is further corroborated by subsequent latent profile analyses of underlying learner performance characteristics.

Secondly, this study adopts a human-centered approach to identify performance profiles of MOOC learners who have completed the course, categorizing them into six profiles at three levels based on both statistical criteria and theoretical interpretability. Compared to previous study [15], our findings offer a more comprehensive and personalized understanding of learning behaviors. The results indicate that learning persistence is a key factor in MOOC success [25]. Learners who consistently engage with the course and perform well in unit tests tend to achieve higher final grades, while those who disengage early or gradually reduce participation perform significantly worse. Specifically, we identified an “Excellent performance: committed learners” profile, which demonstrated the highest level of engagement and achieved the best scores in both unit tests and the final exam, while learners with dropout behavior characteristics performed poorly. This finding aligns with prior research [15, 26], reinforcing the importance of sustained effort in online learning. From the perspective of the Community of Inquiry (CoI) framework [27], the Medical Nursing MOOC’s structured design, which inherits and adapts elements from traditional classroom teaching, aligns closely with teaching presence. This structured instructional approach—including sequenced unit content, integrated tests, and comprehensive learning materials—may promote sustained cognitive presence by guiding learners’ attention, supporting goal-setting, and facilitating active engagement throughout the course. Compared to less structured online courses that merely provide learning resources, this approach appears more effective in fostering continuous learner participation. Moreover, the course design intentionally supports learners in setting and monitoring learning goals, thereby enhancing self-directed learning and improving overall learning outcomes in line with cognitive presence principles. Looking forward, the extensive real-time learning data available on MOOC platforms could be leveraged to further enhance course effectiveness. Some studies have indicated the importance of identifying learners at high risk of withdrawal as early as possible, this is particularly crucial when learners exhibit ongoing engagement in their coursework but demonstrate a tendency to disengage [28]. Course managers should develop personalized learning pathways by leveraging real-time data mining technology [29], and machine learning models [30] to continuously monitor student progress, identify at-risk learners early, and provide targeted interventions [31]. Integrating structured pedagogical design with these adaptive, data-informed strategies represents a promising direction for improving learner engagement and sustaining cognitive presence in MOOCs.

Thirdly, the relationship between learner performance profiles and the developmental stages of MOOCs reveals a clear trend in how online learning behaviors have evolved over time. Notably, the proportion of “Excellent performance: committed learners” has significantly increased over time, rising from 20.60% to 69.74%. This suggests more learners can fully engage with courses, improving autonomy and learning strategies. Conversely, the proportion of “Mediocre performance: committed learners” has declined over time, with only 1.70% remaining in the sustainable stage. This shift may indicate that learners who previously struggled but persisted are now either better prepared for online learning or are filtering out earlier in the process. At the same time, dropout behaviors have also evolved. Our study identified three key profiles— “Poor performance: early dropout learners,” “Poor performance: mid-term dropout learners,” and “Mediocre performance: anticlimactic learners”—who exhibited a gradual decline in unit test scores and eventually disengaged at different points in the course. The proportion of these profiles has decreased over time, suggesting that learners now enter MOOCs with more realistic expectations and a clearer understanding of their own learning needs [25]. A unique profile, “Mediocre performance: catch-up learners,” emerged during the strategic opportunity stage of MOOCs, when many learners were first exposed to online learning due to social distancing measures. These learners struggled in the early units due to a lack of preparedness but improved later in the course, though their final performance remained moderate. This indicates that when implementing new technologies and innovative educational approaches, educators should proactively consider the adaptability of learners. Furthermore, they should dedicate greater efforts to guiding students in adjusting to these novel teaching and learning methods.

Limitations

First, the study’s reliance on data from a single MOOC course inherently constrains the generalizability of our results. While this dataset provided valuable insights, the restricted sampling frame limits representativeness across diverse MOOC platforms, course subjects, pedagogical approaches, and learner demographics. Consequently, the findings should be interpreted as context-specific to this particular course environment. Because of the Personal Privacy Protocol of the MOOC platform, the personal information of the learners has been desensitized, preventing us from using the data of the learners’ occupational/education identity, education background, and geographical location to conduct more in-depth research. This absence of granular demographic and contextual data hindered our ability to conduct deeper subgroup analyses that could reveal nuanced differential effects or tailor interventions. At the same time, traditional variable-centered approaches often overlook heterogeneity among learners and may lead to biased parameter estimates. In contrast, person-centered methods such as Latent Profile Analysis (LPA) enable researchers to capture this heterogeneity by identifying distinct subgroups of individuals with similar response patterns. The classification schemes derived from LPA are conceptually meaningful and methodologically valuable for data-driven development and integration of learner typologies. Given the diversity of learner engagement and performance in MOOCs, LPA offers a theoretical only anonymized assessment-related data were accessible for this study. Variables that could provide deeper insights into learners’behavioral processes, such as real-time engagement, interaction patterns, or participation in discussion forums, were not available. This absence prevented us from conducting more comprehensive analyses of how learning behaviors evolve or how instructional elements may influence engagement. Future research must validate these patterns by incorporating data from multiple, heterogeneous MOOCs to enhance external validity and establish broader applicability. In addition, complementary qualitative studies exploring learners’ experiences, motivations, and challenges could further illuminate how different learner types interact with online courses and support the development of more effective, adaptive MOOC designs.

Conclusions

This article presents the results of an evaluation that is based on an individual-centered approach to categorize various types of MOOC learners who have completed the course. This can help educators gain a better understanding of learners’ behavioral patterns and needs, and then the enhancement of MOOC design and management can be facilitated. The study also uncovers the trends of MOOC learners’ performance styles at different developmental stages of a Medical Nursing MOOC from 2018 to 2023. Our results highlight the evolving role of MOOCs in education. While skepticism about their long-term viability remains, the increasing proportion of highly engaged learners and the declining dropout rate suggest that MOOCs are becoming more effective and sustainable. To further strengthen this trend, educators should focus on enhancing learner adaptability, improving course engagement, and integrating instructional strategies that support different learner profiles.

Authors’ contributions

YLJ and JJZ designed the study; YLJ and JJZ wrote the original manuscript; KL, JZ, LMY, XWZ, WZ, XC, and YB collected the data; YLJ, XXW, WXW, SMH, and CYW analyzed the data; and all authors read and approved the final version.

Funding

This study was supported by the Sun Yat-sen University 2024 Teaching Quality Engineering Project (No. 53000-12253004) and the 2024 Guangdong Provincial Higher Education Teaching Research and Reform Project (Department of Education of Guangdong Province〔2024〕30)sectors. The funder had no role in the survey design, implementation or analysis.

Data availability

The datasets used and analysed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Department of Education Administration and the ethics committee of Sun Yat-sen University. All participating students provided informed consent. This study was conducted in accordance with the principles of the Declaration of Helsinki. The platform has desensitized learner information, so neither teaching staff nor other learners could see the personal information of any individual learner. The dataset used in this study was fully anonymized by the MOOC platform prior to being provided to the research team. No identifiable personal information was accessible at any stage of the analysis.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.