Next Article in Journal
Teams, Tools, Processes and Resources to Manage Oncologic Clinical Decision Support: Lessons Learned from City of Hope’s Multistate, Academic, and Community Oncology Enterprise
Previous Article in Journal
Ki-67 as a Prognostic Marker in Squamous Cell Carcinomas of the Vulva: A Systematic Review
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
JCM
Volume 14
Issue 6
10.3390/jcm14062046
jcm-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Low Back Pain Among Health Sciences Undergraduates: Results Obtained from a Machine-Learning Analysis

by
Janan Abbas
1,*,
Malik Yousef
2,
Kamal Hamoud
1 and
Katherin Joubran
1
1
Department of Physical Therapy, Zefat Academic College, Zefat 13206, Israel
2
Department of Information Systems, Zefat Academic College, Zefat 13206, Israel
*
Author to whom correspondence should be addressed.
J. Clin. Med. 2025, 14(6), 2046; https://doi.org/10.3390/jcm14062046
Submission received: 14 February 2025 / Revised: 14 March 2025 / Accepted: 15 March 2025 / Published: 17 March 2025
(This article belongs to the Section Orthopedics)
Browse Figures
Versions Notes

Abstract

:
Background and objective. Low back pain (LBP) is considered the most common and challenging disorder in health care. Although its incidence increases with age, a student’s sedentary behavior could contribute to this risk. Through machine learning (ML), advanced algorithms can analyze complex patterns in health data, enabling accurate prediction and targeted prevention of medical conditions such as LBP. This study aims to detect the factors associated with LBP among health sciences students. Methods. A self-administered modified version of the Standardized Nordic Questionnaire was completed by 222 freshman health sciences students from May to June 2022. A supervised random forest algorithm was utilized to analyze data and prioritize the importance of variables related to LBP. The model’s predictive capability was further visualized using a decision tree to identify high-risk patterns and associations. Results. A total of 197/222 (88.7%) students participated in this study, most of whom (75%) were female. Their mean age and body mass index were 23 ± 3.8 and 23 ± 3.5, respectively. In this group, 46% (n = 90) of the students reported having experienced LBP in the last month, 15% (n = 30) were smokers, and 60% (n = 119) were involved in prolonged sitting (more than 3 h per day). The decision tree of ML revealed that a history of pain (score = 1), as well as disability (score= 0.34) and physical activity (score = 0.21), were significantly associated with LBP. Conclusions. Approximately 46% of the health science students reported LBP in the last month, and a machine-learning approach highlighted a history of pain as the most significant factor related to LBP.

1. Introduction

Low back pain (LBP) is commonly recognized as a health, social, and economic burden in Western countries. The condition covers different types of pain and involves a spectrum of structures (e.g., muscles and discs). It has been reported that mechanical LBP, which is considered the most common type, could lead to pain chronicity [1]. It is also accepted that LBP increases with age [1,2], and its prevalence according to the National Center of Health Statistics at younger (18–44 years) and older ages (45–65 years) is 23% and 35%, respectively [2]. It is worth noting that LBP is a multifactorial component that includes biological, environmental, and psychological factors [3]. Growing evidence suggests that student lifestyles, particularly those in health science programs, lead to a higher prevalence of LBP than their counterpart in the general population [4,5]. For example, the 12-month prevalence of LBP among nursing students was found to be high (65%), whereas the reported rate among student athletes was 39% [6]. Furthermore, some studies reported that being a physiotherapy student was considered to be a risk factor for LBP [7,8]. These findings could be associated with the fact that health science students possess highly demanding curricula (e.g., loss of sleep hours), poor working posture, manual handling activities during training, and stressful experiences [8,9,10]. However, data regarding the specific variables that greatly stimulate LBP among health science students are still unclear [10,11,12,13]. For example, two prospective cohort studies [10,11] among nursing students found diverse predictive variables for LBP.
Machine learning (ML) is continually gaining importance in medical predictive analytics and has numerous advantages compared to conventional analyses such as logistic regression [14]. ML algorithms can expose highly complex interactions among variables that allow for more accurate medical prognosis, as well as other novel applications [15,16]. ML can be divided mainly into supervised and unsupervised models. A common supervised model is that of decision tree (DT) learning, whose classification and regression implement a grouping or a regression task, which is more visible and easier to understand than other modalities [17]. Additionally, DT can learn highly irregular patterns, with low bias and high variance for the training set, boosting the model’s performance [18]. Most studies on ML in spinal disorders are in the domain of diagnosis, followed by prognosis, prediction, and biomechanics for spinal applications [19]. A recent systematic review showed that ML has been used only for the recovery prediction and treatment of LBP [20]. The authors concluded that ML could also enhance the ability to detect patterns of clinical characteristics in LBP [20] even though little existing ML research predicts its onset. Understanding the risk factors of LBP among undergraduates of health science, who are more likely to engage in unhealthy behaviors that adversely affect their well-being, can help in implementing suitable preventive or treatment measures to attenuate LBP chronicity in future daily function and practice.
Thus, this study aims to investigate the factors contributing to LBP among health science students using the ML approach. By applying a random forest algorithm, the study seeks to uncover complex variable interactions, prioritize risk factors, and provide actionable insights for preventive strategies tailored to this population.

2. Materials and Methods

Study design. A cross-sectional study of 222 freshman undergraduates in health sciences (nursing, physical therapy, medical laboratory science, and emergency medical services) from the Zefat Academic College in the north of Israel who were enrolled between May and June 2022. There were no exclusion criteria other than the inability to fill out the hardcopy self-reported questionnaire. We considered only the freshman class, as we plan to follow the participants throughout their academic years. The participants were recruited in their classrooms by one of the study’s assessors at the end of that course in the presence of their instructor. A consent form, which included the purpose of the study and the right of the student to withdraw at any time, was received from each participant.
The study was conducted in adherence to the Helsinki Declaration, and it was approved by the Departmental Research Ethics Committee at Zefat Academic College (no. 19-2022).

2.1. Instruments and Measures

A structured and anonymous questionnaire according to the modified validated Standardised Nordic Questionnaire [21] that seeks information on sociodemographic characteristics and factors such as sedentary behavior and smoking habits was used [22] (Table 1). Physical activities were also recorded and were adapted from the American College of Sports Medicine guidelines [23]. These were graded from 0 = no physical activity to 5 = high frequency of physical activity. The students were asked if they had ever suffered LBP at some point in their life (lifetime prevalence) and if they had ever been hospitalized due to LBP. Those who had experienced LBP in the last year (based on the response “yes” to the question) were asked specific questions about pain frequency (graded from 1 = daily to 4 = rarely/never) and medication consumption, as well as whether they were seeking care and disability according to the Oswestry Disability Index (ODI) [24].
In addition, the participants were asked about variable-related stress [21] in the last month (e.g., study and personal issues), and were divided into two categories based on their response to the question: “What is your daily stress level?” (a = very high to quite high, b = low to none). Their LBP in the last month was also recorded, and this was considered positive if it had lasted at least 12 h and its intensity scale according to the numeric rating [25] was above five.

2.2. Statistical and Supervised Machine-Learning Analyses

In analyzing our data, we first used statistical methods with IBM SPSS version 25. We checked whether the numerical data followed a normal distribution and conducted descriptive analyses for both numerical and categorical variables. Next, we applied machine learning (ML) to identify key factors that influence LBP. Specifically, we used the Random Forest (RF) algorithm, a powerful ML method for complex data analysis [26]. RF comprises many DTs, which are simple models that mimic human decision-making [27]. Each DT asks a series of yes/no questions to classify data. For example, a tree might first ask a person if they sit for long hours and then ask their age before predicting their risk for LBP. RF combines multiple DTs to improve accuracy and find the most critical factors related to LBP. Potential variables such as age, activity level, and sedentary behavior were identified from previous studies. To ensure our results were reliable and generalizable, we implemented a rigorous validation approach using 100-fold Monte Carlo cross-validation [28]. We randomly split the dataset into training (70%) and testing (30%) sets for each fold, maintaining the class distribution across splits. This technique helps evaluate a model’s stability by repeatedly training and testing it on different subsets of the data, minimizing potential bias from a single arbitrary split.
We evaluated how well the model performed using several measures: sensitivity—how well it identified people with LBP, specificity—how well it identified those without LBP, precision—how accurate the positive predictions were, accuracy—the overall correctness of the predictions, and F-measure—the balance between precision and sensitivity. The area under the receiver operating characteristics curve (AUROC) was also considered for the test dataset to be a strong indicator of performance for classifiers in imbalanced datasets.
A total of 197 students participated in this study. For the DT learning process, we divided the participants into two groups: 90 with LBP and 107 without LBP. The optimal feature was selected recursively, and the training data were segmented according to this feature to ensure the best classification process. This process corresponds to the division of the feature space and the construction of the DT.
First, we constructed the root node, where all training data were placed. An optimal feature was chosen, and the training dataset was divided into subsets according to this feature, ensuring the best classification under the given conditions. If these subsets were correctly classified, they were assigned to leaf nodes. If any subset remained unclassified, a new optimal feature was selected, and the process continued until all data were classified or no more suitable features were available. Finally, each subset was assigned to the leaf nodes, completing the DT structure [29]. The KNIME platform was used with default parameters, which was deemed sufficient after initial testing showed stable performance.

3. Results

3.1. Student Characteristics

One hundred ninety-seven students (88.7%) responded and completed the questionnaire (Figure 1). The demographic and sedentary features of the participants are presented in Table 1. The mean age and body mass index (BMI) values for all the participants were 23 ± 3.8 and 23 ± 3.5, respectively. Seventy-five percent were females (n = 148) and 45% (n = 89) were from the nursing department. The majority (93%) of participants were single, and 15% were habitual smokers. Additionally, 73% defined themselves as secular and traditional, 60% were involved in prolonged daily sitting (more than 3 h), and most of them (85%, n = 167) reported that they experienced study-related stress.
In our series, 55% (n = 108) of the students took part in sports activities with diverse frequencies: 7% (n = 14) with low frequency (at least 20 min to 1 h weekly) and 19% (n = 37) with high practice frequency (more than 3 h).

3.2. Low Back Pain

A high prevalence of students (74%, n = 146) have experienced LBP at some point in their lives, and 4% (n = 8) have been hospitalized because of such pain. In addition, 46% (n = 90) reported that they had suffered lower back pain in the last month (Table 2). The features of LBP during the last year were dispersed. For example, the frequency of LBP was as follows: 17% (n = 34) suffered almost daily, 24% (n = 47) once a week, 22% (n = 43) once a month, and 37% (n = 73) rarely or never. It was also found that 33% of the students (n = 65) reported disability, and about 20% (n = 39) were referred to a physician and/or used medication.
The assessment of the model performance such as accuracy and recall are presented in Table 3.
The DT represents the variables that are associated with LBP. As depicted in Figure 2a, the tree comprises internal nodes (conditions), branches (decisions), and leaves (outcomes). In general, variables that appear at the top of the tree (e.g., frequency of LBP) have a greater impact on LBP than those in the bottom nodes, such as department or BMI. To understand the DT output in Figure 2a, we illustrated its main findings in Figure 2b. We found that 90 out of the 197 students reported LBP in the last month. We also noticed that the “frequency of LBP in the last year” was considered to be at the “peak” of the tree, having the greatest impact on LBP (score = 1, Table 4). Additionally, among those who suffered from LBP more regularly (daily), 32 students out of 34 had LBP (94%), and when the frequency was reported as “rarely”, this percentage decreased to 8% (6 out of 73). This result implies that once the person begins to experience LBP more often, the likelihood of its recurrence increases. We also observed that a history of “disability” and physical activity (PA) influenced the onset of LBP, with scores of 0.34 and 0.21, respectively (Table 4). For example, 11 students out of 12 reported LBP if they met the following conditions: (a) had pain frequency at least “once a month”, (b) were traditionally religious, and (c) had experienced disability. Notably, the contribution of variables appearing in the later example’s internal nodes, such as religion (traditional), is partial since only 13 students out of 21 reported LBP (Figure 2b) (score = 0.13, Table 4). In addition, 20 students out of 25 suffered from LBP if their pain frequency was “once a week” and they did not engage in PA. The scores of each variable associated with LBP in the DT are depicted in Table 4.

4. Discussion

This cross-sectional study aimed to verify the associated variables for LBP among freshman health sciences students using machine learning. Contrary to previous studies that applied logistic regression analysis with well-known limitations (e.g., unstable variable selection), here, we used a random forest of ML algorithms. To the best of our knowledge, few studies have addressed the risk factors for LBP using this approach [30,31,32].
Although the utilization of ML for detecting LBP is still in its infancy, it was recently reported that supervised ML compared with other analyses (e.g., regression) could predict LBP among college students from a questionnaire with high accuracy [31]. Shim and colleagues (2021) also noted that ML could effectively identify populations at high risk of chronic LBP [32].
This study showed lifetime and 1-month prevalences of LBP of 74% and 46%, respectively. This finding falls within the range (31% to 84%) previously reported among health sciences undergraduates worldwide [7,13,33,34,35,36,37,38,39]. The lifetime prevalence among these undergraduates, such as those studying nursing and physiotherapy, was about 50% [7,13]. It was reported that about 84% of a freshman class in health sciences (n = 153) suffered from LBP [36]. Nyland and Grimmer also reported a 31% 1-month prevalence among first-class physiotherapy students [7].
Our results confirm that a history of pain (frequency) has the greatest relationship (score = 1) with LBP. However, while this finding is in agreement with some studies [10,40,41], it contradicts others [11,13,34,35,42,43,44]. A systematic review study in the general population showed that a history of previous episodes of LBP was the only predicting factor for the recurrence of LBP [40]. A prospective cohort study (over 4 years) among nursing students also reported that a history of LBP was considered a prediction factor for new episodes [10]. In contrast, other studies revealed that smoking, advanced physical activity, higher stress levels, a sedentary lifestyle, and mal-posture were independent factors for LBP among nursing students [11,35,43]. Notably, factors associated with such back pain are inconsistent due to high-risk biases, such as heterogeneity of the methodology and sample features (e.g., students’ department, general population) and/or a small study sample.
This study also found that disability in the last year and a lack of physical activity can predict LBP. This association between a disability history and LBP was previously confirmed [44,45]. For example, in a prospective cohort study in a working population with a 3-year follow-up period, it was found that high disability due to LBP was a prognostic factor for its recurrence [44]. As mentioned, data regarding physical activity and LBP are still conflicting [12,46,47,48]. For example, Felemban et al. [48] and Amelot et al. [12] found that exercise and walking at least 30 min a day significantly improved LBP among a group of dental and medical students. Conversely, an Australian cross-sectional study among nursing students revealed that vigorous physical activity contributed to LBP [46]. It is widely recognized that physical activity has numerous health benefits related to physiological and psychological health [49,50,51]. Moreover, it is well-accepted that physical activity can reduce the risk factors of various chronic diseases such as those that are cardiovascular and musculoskeletal, as well as diabetes [50,52]. Thus, the impact of physical activity on LBP may depend on the activity type and the degree of loading on the lumbar spine [53,54]. One related review [55] suggested that the type and intensity of physical activity should be considered when measuring the association with LBP. A recent systematic and meta-analysis review reported that a medium activity level was associated with a lower prevalence of LBP [56], and the authors recommended moderate doses of physical activity when managing and preventing LBP.
The fact that 55% of our participants were involved in physical activity is in line with a review from Keating and colleagues that 40–50% of college students are physically inactive [57]. In addition, sufficient evidence has been found to indicate that most university students have an unhealthy lifestyle, with poor levels of physical activity (PA) [58,59]. These findings are not surprising and highlight what has previously been reported regarding the need for PA intervention programs for these students [60,61].

5. Limitations of the Study

Despite the widespread use and good validation of the Standardized Nordic Questionnaire (SNQ), it has some notable disadvantages. While it effectively screens for musculoskeletal symptoms such as LBP, it lacks diagnostic specificity and cannot differentiate between various pathologies causing similar pain patterns [62]. The SNQ relies entirely on self-reporting, introducing potential recall bias and subjective interpretation of symptoms, especially when respondents report pain over the previous 12-month period [63]. Its simplistic pain assessment fails to capture the multidimensional nature of pain, missing important aspects like pain duration, patterns, and psychological factors that influence pain perception. Cultural and linguistic aspects as well as socioeconomic factors can significantly affect how pain is reported and understood across different populations, potentially compromising cross-cultural validity despite translations. Additionally, the SNQ lacks psychometric refinement compared to newer instruments, noting that while broadly applicable, it has less robust reliability and validity metrics than more recently developed specialized assessment tools [64]. The SNQ also overlooks important contextual factors like work environment, biomechanical exposures, and psychological considerations that contribute to musculoskeletal disorders. It was also pointed out that the questionnaire provides insufficient information for clinical decision-making or treatment planning without supplementary assessments [65]. Finally, the instrument may not reflect current musculoskeletal disorders or contemporary occupational health concerns.
Notably, as the participants lack diversity (e.g., health science undergraduates, environmental), this may cause the model to miss relevant predictors in underrepresented groups [66]. The fact that the selected variables were based on previous studies could be misleading to favor variables aligned with existing clinical beliefs about LBP [67]. Additionally, causal relationships are challenging in cross-sectional studies, where decision trees might misidentify the consequences of LBP as predictors rather than actual risk factors [68]. Additionally, the study was conducted at a single academic institution in the north of the State of Israel, which limits its generalizability. As LBP is a multifactorial phenomenon, some related factors such as spinal deformity, depression, socioeconomic background, and medical history were not addressed in this study.
We believe that our findings underscore the potential of ML to predict and prioritize LBP risk factors, enabling targeted interventions. For instance, academic institutions could design preventive measures, such as tailored PA programs or therapeutic exercise protocols including automobilization [69] for high-risk students identified by the model. One can also assume that stakeholders should dedicate “physical activity time” to the student curricula to reduce the risk and chronicity of future LBP. We anticipate that further studies are needed using ML methods in different academic centers and cultures to check for repeatable results and to establish basic predictive factors for LBP among undergraduates.

6. Conclusions

The current study found that the 1-month prevalence of LBP among freshman health sciences undergraduates was 46%. The random forest algorithm indicates that a history of pain frequency has the greatest impact on LBP. In addition, physical activity and a history of disability were also associated with LBP.

Author Contributions

Conceptualization K.J. and J.A.; methodology M.Y. and K.H.; validation and data collection K.J.; investigation M.Y. and J.A.; original draft preparation JA; writing—review and editing J.A., K.H. and M.Y.; supervision J.A.; administration J.A. and K.J. All authors have read and agreed to the published version of the manuscript.

Funding

The authors declare that they have no sources of funding for research reported.

Institutional Review Board Statement

The ethical committee of Zefat Academic College (no. 19-2022) has approved this research (Code: 19-2022; Date: 12 May 2022).

Informed Consent Statement

Informed and written consent was obtained from all subjects involved in the current study in their classroom through a statement in the initial questionnaire: “anyone who intends to fill out the questionnaire constitutes a reference that they agree to participate in the research”.

Data Availability Statement

The data presented in this study are available upon request from the corresponding author due to privacy reasons.

Conflicts of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

References

  1. Itz, C.J.; Geurts, J.W.; van Kleef, M.; Nelemans, P. Clinical course of non-specific low back pain: A systematic review of prospective cohort studies set in primary care. Eur. J. Pain 2013, 17, 5–15. [Google Scholar] [CrossRef] [PubMed]
  2. Gaonkar, A.P.; Kulkarni, R.; Caytiles, R.D.; Iyengar, N. Classification of Lower Back Pain Disorder Using Multiple Machine Learning Techniques and Identifying Degree of Importance of Each Parameter. Int. J. Adv. Sci. Technol. 2017, 105, 11–24. [Google Scholar] [CrossRef]
  3. Lall, M.P.; Restrepo, E. The biopsychosocial model of low back pain and patient-centered outcomes following lumbar fusion. Orthop. Nurs. 2017, 36, 213–221. [Google Scholar] [CrossRef] [PubMed]
  4. Crawford, R.J.; Volken, T.; René Schaffert, R.; Bucher, T. Higher low back and neck pain in final year Swiss health professions’ students: Worrying susceptibilities identified in a multi-centre comparison to the national population. BMC Public Health 2018, 8, 1188. [Google Scholar] [CrossRef]
  5. Cakmak, A.; Yücel, B.; Ozyalçn, S.N. The frequency and associated factors of low back pain among a younger population in Turkey. Spine 2004, 29, 1567–1572. [Google Scholar] [CrossRef]
  6. Wong, A.Y.L.; Chan, L.L.Y.; Lo, C.W.T.; Chan, W.W.Y.; Lam, K.C.K.; Bao, J.C.H.; Ferreira, M.L.; Armijo-Olivo, S. Prevalence/Incidence of Low Back Pain and Associated Risk Factors Among Nursing and Medical Students: A Systematic Review and Meta-Analysis. PM&R 2021, 13, 1266–1280. [Google Scholar]
  7. Nyland, L.J.; Grimmer, K.A. Is undergraduate physiotherapy student a risk factor for low back pain? A prevalence study of low back pain in physiotherapy students. BMC Musculoskelet. Disord. 2003, 4, 22. [Google Scholar] [CrossRef]
  8. Falavigna, A.; Teles, A.R.; Mazzocchin, T.; de Braga, G.L.; Kleber, F.D.; Barreto, F.; Santin, J.T.; Barazzetti, D.; Lazzaretti, L.; Steiner, B.; et al. Increased prevalence of low back pain among physiotherapy students compared to medical students. Eur. Spine J. 2011, 20, 500–505. [Google Scholar] [CrossRef]
  9. Algarni, A.D.; Al-Saran, Y.; Al-Moawi, A.; Bin Dous, A.; Al-Ahaideb, A.; Kachanathu, S.J. The prevalence of and factors associated with neck, shoulder, and low-back pains among medical students at university hospitals in Central Saudi Arabia. Pain Res. Treat. 2017, 1, 1235706. [Google Scholar] [CrossRef]
  10. Feyer, A.M.; Herbison, P.; Williamson, A.M. The role of physical and psychological factors in occupational low back pain: A prospective cohort study. Occup. Environ. Med. 2000, 57, 116–120. [Google Scholar] [CrossRef]
  11. Mitchell, T.; O’Sullivan, P.B.; Burnett, A. Identification of modifiable personal factors that predict new-onset low back pain: A prospective study of female nursing students. Clin. J. Pain 2010, 26, 275–283. [Google Scholar] [CrossRef]
  12. Amelot, A.; Mathon, B.; Haddad, R.; Renault, M.C.; Duguet, A.; Steichen, O. Low Back Pain among Medical Students: A Burden and an Impact to Consider! Spine 2019, 44, 1390–1395. [Google Scholar] [CrossRef]
  13. AlShayhan, F.A.; Saadeddin, M. Prevalence of low back pain among health sciences students. Eur. J. Orthop. Surg. Traumatol. 2018, 28, 165–170. [Google Scholar] [CrossRef]
  14. Khan, O.; Badhiwala, J.H.; Wilson, J.R.F.; Jiang, F.; Martin, A.R.; Fehlings, M.G. Predictive modeling of outcomes after traumatic and nontraumatic spinal cord injury using machine learning: Review of current progress and future directions. Neurospine 2019, 16, 678–685. [Google Scholar] [CrossRef]
  15. Senders, J.T.; Staples, P.C.; Karhade, A.V.; Zaki, M.M.; Gormley, W.B.; Broekman, M.L.; Smith, T.R.; Arnaout, O. Machine learning and neurosurgical outcome prediction: A systematic review. World Neurosurg. 2018, 109, 476–486.e1. [Google Scholar]
  16. LeCun, Y.; Bengio, Y.; Hinton, G. Deep learning. Nature 2015, 521, 436–444. [Google Scholar] [CrossRef]
  17. Ren, G.; Yu, K.; Xie, Z.; Wang, P.; Zhang, W.; Huang, Y.; Wang, Y.; Wu, X. Current Applications of Machine Learning in Spine: From Clinical View. Glob. Spine J. 2021, 28, 521, 436–444. [Google Scholar] [CrossRef]
  18. Galbusera, F.; Casaroli, G.; Bassani, T. Artificial intelligence and machine learning in spine research. JOR Spine 2019, 2, e1044. [Google Scholar] [CrossRef]
  19. Azimi, P.; Yazdanian, T.; Benzel, E.C.; Aghaei, H.N.; Azhari, S.; Sadeghi, S.; Montazeri, A. A Review on the Use of Artificial Intelligence in Spinal Diseases. Asian Spine J. 2020, 14, 543–571. (In English) [Google Scholar]
  20. Tagliaferri, S.D.; Angelova, M.; Zhao, X.; Owenet, P.J.; Miller, C.T.; Wilkin, T.; Belavy, D.L. Artificial intelligence to improve back pain outcomes and lessons learnt from clinical classification approaches: Three systematic reviews. NPJ Digit. Med. 2020, 3, 93. [Google Scholar]
  21. Kuorinka, I.; Jonsson, B.; Kilbom, A. Standardised Nordic questionnaires for the analysis of musculoskeletal symptoms. Appl. Ergon. 1987, 18, 233–237. [Google Scholar] [CrossRef]
  22. Jessor, R.; Turbin, M.S.; Costa, M.F. Survey of Personal and Social Development at CU; Institute of Behavioral Sciences, University of Colorado: Boulder, CO, USA, 2003; Volume 19, p. 2016. Available online: http://www.colorado.edu/ibs/jessor/questionnaires/questionnaire_spsd2.pdf (accessed on 15 December 2021).
  23. Thompson, W.R.; Gordon, N.F.; Pescatello, L.S. American College of Sport Medicine. In ACSM’s Guidelines for Exercise Testing and Prescription, 9th ed.; Lippincott Williams and Wilkins: Philadelphia, PA, USA, 2014. [Google Scholar]
  24. Fairbank, J.C.; Pynsent, P.B. The Oswestry Disability Index. Spine 2000, 25, 2940–2952. [Google Scholar] [CrossRef]
  25. Dionne, C.E.; Dunn, K.M.; Croft, P.R. A consensus approach toward the standardization of back pain definitions for use in prevalence studies. Spine 2008, 33, 95–103. [Google Scholar] [CrossRef]
  26. Breiman, L. Random forests. Mach. Learn. 2001, 45, 5–32. [Google Scholar] [CrossRef]
  27. Biau, G.; Scornet, E. A random forest guided tour. Test 2016, 25, 197–227. [Google Scholar] [CrossRef]
  28. Picard, R.R.; Cook, R.D. Cross-validation of regression models. J. Am. Stat. Assoc. 1984, 79, 575–583. [Google Scholar] [CrossRef]
  29. Feng, Y.; Wang, Z.; Cui, R.; Xiao, M.; Gao, H.; Bai, H.; Delvoux, B.; Dekker, A.; Romano, A.; Wang, S.; et al. Clinical analysis and artificial intelligence survival prediction of serous ovarian cancer based on preoperative circulating leukocytes. J. Ovarian Res. 2022, 15, 64. [Google Scholar] [CrossRef]
  30. Muhaimil, A.; Pendem, S.; Sampathilla, N.; Priya, P.S.; Nayak, K.; Chadaga, K.; Goswami, A.; Chandran, O.M.; Shirlal, A. Role of Artificial intelligence model in prediction of low back pain using T2 weighted MRI of Lumbar spine. F1000Research 2024, 13, 1035. [Google Scholar] [CrossRef]
  31. Dasgupta, P.; Kuhn, L.M.; Sekeroglu, K.; Prasai, A. A mathematical modelling approach for analysing the risk of low back pain in college students using student desks. Int. J. Hum. Factors Ergon. 2023, 10, 85–112. [Google Scholar] [CrossRef]
  32. Shim, J.G.; Ryu, K.H.; Cho, E.A.; Ahn, J.H.; Kim, H.K.; Lee, Y.-J.; Lee, S.H. Machine learning approaches to predict chronic lower back pain in people aged over 50 years. Medicina 2021, 57, 1230. [Google Scholar] [CrossRef]
  33. Mitchell, T.; O’Sullivan, P.B.; Burnett, A.F. Low back pain characteristics from undergraduate student to working nurse in Australia: A cross-sectional survey. Int. J. Nurs. Stud. 2008, 45, 1636–1644. [Google Scholar] [CrossRef]
  34. Nordin, N.; Singh, D.; Kanglun, L. Low back pain and associated risk factors among health science undergraduates. Sains Malays. 2014, 43, 423–428. [Google Scholar]
  35. Hafeez, K.; Memon, A.A.; Jawaid, M. Back pain-are health care undergraduates at risk? Iran. J. Public Health 2013, 42, 819–825. [Google Scholar] [PubMed]
  36. Yucel, H.; Torun, P. Incidence and Risk Factors of Low Back Pain in Students Studying at a Health University. Bezmialem Sci. 2016, 4, 12–18. [Google Scholar] [CrossRef]
  37. Anggiat, L.; Hon, W.H.C.; Baait, W.S.N. The incidence of Low back pain among university students. J. Pro-Life 2018, 5, 677–687. [Google Scholar]
  38. Smedley, J.; Trevelyan, F.; Inskip, H. Impact of ergonomic intervention on back pain among nurses. Scand. J. Work. Environ. Health 2003, 29, 117–123. [Google Scholar] [CrossRef]
  39. Maul, I.; Läubli, T.; Klipstein, A.; Krueger, H. Course of low back pain among nurses: A longitudinal study across eight years. Occup. Environ. Med. 2003, 60, 497–503. [Google Scholar] [CrossRef]
  40. Da-Silva, T.; Mills, K.; Brown, B.T. Risk of Recurrence of Low Back Pain: A Systematic Review. J. Orthop. Sports Phys. Ther. 2017, 47, 305–313. [Google Scholar] [CrossRef]
  41. Machado, G.C.; Maher, C.G.; Ferreira, P.H.; Latimer, J.; Koes, B.W.; Steffens, D.; Ferreira, M.L. Can recurrence after an acute episode of low back pain be predicted? Phys. Ther. 2017, 97, 889–895. [Google Scholar] [CrossRef]
  42. Alshagga, M.A.; Nimer, A.R.; Yan, L.P.; Ibrahim, I.A.A.; Al-Ghamdi, S.S.; Al-Dubai, S.A.R. Prevalence and factors associated with neck, shoulder and low back pains among medical students in a Malaysian Medical College. BMC Res. Notes 2013, 6, 244. [Google Scholar] [CrossRef]
  43. Citko, A.; Górski, S.; Marcinowicz, L.; Górska, A. Sedentary Lifestyle and nonspecific low back pain in medical personnel in North-East Poland. BioMed Res. Int. 2018, 2018, 1965807. [Google Scholar] [CrossRef] [PubMed]
  44. van den Heuvel, S.G.; Ariëns, G.A.; Boshuizen, H.C.; Hoogendoorn, W.E.; Bongers, P.M. Prognostic factors related to recurrent low-back pain and sickness absence. Scand. J. Work. Environ. Health 2004, 1, 459–467. [Google Scholar] [CrossRef] [PubMed]
  45. Estlander, A.M.; Takala, E.P.; Viikari-Juntura, E. Do psychological factors predict changes in musculoskeletal pain? J. Occup. Environ. Med. 1998, 5, 445–453. [Google Scholar] [CrossRef]
  46. Mitchell, T.; O’Sullivan, P.B.; Smith, A. Biopsychosocial factors are associated with low back pain in female nursing students: A cross-sectional study. Int. J. Nurs. Stud. 2009, 46, 678–688. [Google Scholar] [CrossRef]
  47. Skoffer, B.; Foldspang, A. Physical activity and low-back pain in schoolchildren. Eur. Spine J. 2008, 17, 373–379. [Google Scholar] [CrossRef]
  48. Felemban, R.A.; Sofi, R.A.; Alhebshi, S.A.; Alharbi, S.G.; Farsi, N.J.; Abduljabbar, F.H.; Farsi, J.M. Prevalence and predictors of musculoskeletal pain among undergraduate students at a dental school in Saudi Arabia. Clin. Cosmet. Investig. Dent. 2021, 17, 39–46. [Google Scholar] [CrossRef]
  49. Warburton, D.E.; Nicol, C.W.; Bredin, S.S. Health benefits of physical activity: The evidence. Can. Med. Assoc. J. 2006, 174, 801–809. [Google Scholar] [CrossRef]
  50. O’Donovan, G.; Lee, I.M.; Hamer, M.; Stamatakis, E. Association of “Weekend Warrior” and Other Leisure Time Physical Activity Patterns with Risks for All-Cause, Cardiovascular Disease, and Cancer Mortality. JAMA Intern. Med. 2017, 177, 335–342. [Google Scholar] [CrossRef]
  51. Sadarangani, K.P.; Hamer, M.; Mindell, J.S.; Coombs, N.A.; Stamatakis, E. Physical activity and risk of all-cause and cardiovascular disease mortality in diabetic adults from Great Britain: Pooled analysis of 10 population-based cohorts. Diabetes Care 2014, 37, 1016–1023. [Google Scholar] [CrossRef]
  52. National Institute for Health and Clinical Excellence. Physical Activity: Exercise Referral Schemes; National Institute for Health and Clinical Excellence: London, UK, 2014. [Google Scholar]
  53. Iwamoto, J.; Abe, H.; Tsukimura, Y.; Wakano, K. Relationship between radiographic abnormalities of lumbar spine and incidence of low back pain in high school and college football players: A prospective study. Am. J. Sports Med. 2004, 32, 781–786. [Google Scholar] [CrossRef]
  54. Lundin, O.; Hellstrom, M.; Nilsson, I.; Sward, L. Back pain and radiological changes in the thoracolumbar spine of athletes: A long term follow-up. Scand. J. Med. Sci. Sports 2001, 11, 103–109. [Google Scholar] [CrossRef]
  55. Heneweer, H.; Staes, F.; Aufdemkampe, G.; van Rijn, M.; Vanhees, L. Physical activity and low back pain: A systematic review of recent literature. Eur. Spine J. 2001, 20, 826–845. [Google Scholar] [CrossRef] [PubMed]
  56. Alzahrani, H.; Martin Mackey, M.; Stamatakis, E.; Zadro, J.R.; Shirley, D. The association between Physical activity and low back pain: A systematic review and meta-analysis of observation studies. Sci. Rep. 2019, 9, 8244. [Google Scholar] [CrossRef] [PubMed]
  57. Keating, X.D.; Guan, J.; Piñero, J.C.; Bridges, D.M. A meta-analysis of college students’ physical activity behaviors. J. Am. Coll. Health 2005, 54, 116–126. [Google Scholar] [CrossRef] [PubMed]
  58. Jakubiec, D.; Kornafel, D.; Cygan, A.; Gorska-Klek, L.; Chromik, K. Lifestyle of students from different universities in Wroclaw, Poland. Rocz. Państwowego Zakładu Hig. 2015, 66, 337–344. [Google Scholar]
  59. Almutairi, K.M.; Alonazi, W.B.; Vinluan, J.M.; Almigbal, T.H.; Batais, M.A.; Al Odhayani, A.; Alsadhan, N.; Tumala, R.B.; Moussa, M.; Aboshaiqah, A.E.; et al. Health promoting lifestyle of university students in Saudi Arabia: A cross-sectional assessment. BMC Public Health 2018, 8, 1093. [Google Scholar] [CrossRef]
  60. Sigmundová, D.; ChmelÍk, F.; Sigmund, E.; Feltlová, D.; Frömel, K. Physical activity in the lifestyle of Czech university students: Meeting health recommendations. Eur. J. Sport Sci. 2013, 13, 744–750. [Google Scholar] [CrossRef]
  61. Deliens, T.; Deforche, B.; De Bourdeaudhuij, I.; Clarys, P. Determinants of physical activity and sedentary behaviour in university students: A qualitative study using focus group discussions. BMC Public Health 2015, 15, 201. [Google Scholar] [CrossRef]
  62. Descatha, A.; Roquelaure, Y.; Chastang, J.F.; Evanoff, B.; Melchior, M.; Mariot, C.; Ha, C.; Imbernon, E.; Goldberg, M.; Leclerc, A. Validity of Nordic-style questionnaires in the surveillance of upper-limb work-related musculoskeletal disorders. Scand. J. Work. Environ. Health 2007, 33, 58–65. [Google Scholar] [CrossRef]
  63. Baron, S.; Hales, T.; Hurrell, J. Evaluation of symptom surveys for occupational musculoskeletal disorders. Am. J. Ind. Med. 1996, 29, 609–617. [Google Scholar] [CrossRef]
  64. Crawford, J.O. The Nordic Musculoskeletal Questionnaire. Occup. Med. 2007, 57, 300–301. [Google Scholar] [CrossRef]
  65. Palmer, K.; Smith, G.; Kellingray, S.; Cooper, C. Repeatability and validity of an upper limb and neck discomfort questionnaire: The utility of the standardized Nordic questionnaire. Occup. Med. 1999, 49, 171–175. [Google Scholar] [CrossRef]
  66. Moons, K.G.; Kengne, A.P.; Woodward, M.; Royston, P.; Vergouwe, Y.; Altman, D.G.; Grobbee, D.E. Risk prediction models: I. Development, internal validation, and assessing the incremental value of a new (bio) marker. Heart 2012, 98, 683–690. [Google Scholar] [CrossRef]
  67. Wolff, R.F.; Moons, K.G.; Riley, R.D.; Whiting, P.F.; Westwood, M.; Collins, G.S.; Reitsma, J.B.; Kleijnen, J.; Mallett, S.; PROBAST Group. PROBAST: A tool to assess the risk of bias in prediction model studies. Ann. Intern. Med. 2019, 170, 51–58. [Google Scholar] [CrossRef]
  68. Hancock, M.J.; Maher, C.G.; Latimer, J.; Herbert, R.D.; McAuley, J.H. Can predictors of response to NSAIDs be identified in patients with acute low back pain? Clin. J. Pain 2011, 27, 535–540. [Google Scholar] [CrossRef]
  69. Cuenca-Zaldívar, J.N.; Fernández-Carnero, J.; Sánchez-Romero, E.A.; Álvarez-Gonzalo, V.; Conde-Rodríguez, R.; Rodríguez-Sanz, D.; Calvo-Lobo, C. Effects of a Therapeutic Exercise Protocol for Patients with Chronic Non-Specific Back Pain in Primary Health Care: A Single-Group Retrospective Cohort Study. J. Clin. Med. 2023, 12, 6478. [Google Scholar] [CrossRef]
Jcm 14 02046 g001
Figure 1. Flow chart for the study sample.
Figure 1. Flow chart for the study sample.
Jcm 14 02046 g001
Jcm 14 02046 g002
Figure 2. (a): Decision tree output of the random forest for predicting 1-month LBP obtained by the given variables. Neg = negative, pos = positive. PA—physical activity; (b): the main variables significantly associated with LBP according to the decision tree outcome.
Figure 2. (a): Decision tree output of the random forest for predicting 1-month LBP obtained by the given variables. Neg = negative, pos = positive. PA—physical activity; (b): the main variables significantly associated with LBP according to the decision tree outcome.
Jcm 14 02046 g002
Table 1. Sample size characteristics.
Table 1. Sample size characteristics.
% (n)/or Mean ± SD
89 (197)Response rate
25 (49)Male
75 (148)Female
Department:
45 (89)Nursing
21 (41)Physical Therapy
19 (37)Medical Lab
15(30)Emergency Medical Services
23 ± 3.8Mean age (year)
23 ± 3.5Mean BMI (kg/m2)
91 (179)Dominant right hand
Marital status:
93 (184)Single
7 (13)Other
15 (30)Smoking
Religion & faith:
30 (60)Secular
43 (84)Traditional
13 (26)Religious/orthodox
14 (27)Other
55 (108)Those with physical activity (wk):
7 (14)20 min to 1 h
19 (37)1–2 h
10 (20)2–3 h
9 (18)3–4 h
10 (19)>4 h
Prolonged daily sitting:
40 (78)Up to 3 h
23 (45)3–5 h
37 (74)5 h>
Total daily sitting:
32 (63)Up to 6 h
30 (60)6–8 h
38 (74)>8 h
* Study-related stress:
85 (167)Very high–quite high
15 (30)Little–none
* We present only study-related stress as this variable was most common compared to other types of stress.
Table 2. Low back pain (LBP) characteristics for the study group.
Table 2. Low back pain (LBP) characteristics for the study group.
N (%)
146 (74)Those who experienced LBP at some point in their life
8 (4)Those who had been hospitalized due to LBP
LBP in the last year:
1. Frequency of LBP:
34 (17)
  • Every day
47 (24)
  • Once a week
43 (22)
  • Once a month
73 (37)
  • Rarely/never
65 (33)2. Disability
39 (20)3. Seeking care
39 (20)4. Medication consumption
90 (46)1-month LBP
Table 3. Means and standard deviation (SD) of the model performance.
Table 3. Means and standard deviation (SD) of the model performance.
AccuracyRecallSpecificityPrecisionF1Area Under Curve
Mean0.840.930.760.790.840.88
SD0.090.090.180.130.070.07
Table 4. Scores of the significant variables associated with LBP by the decision tree.
Table 4. Scores of the significant variables associated with LBP by the decision tree.
VariableScore
Frequency of LBP1
History of Disability0.34
Frequency of physical activity (PA)0.21
Weight0.18
Religion and faith0.13
Department0.12
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Abbas, J.; Yousef, M.; Hamoud, K.; Joubran, K. Low Back Pain Among Health Sciences Undergraduates: Results Obtained from a Machine-Learning Analysis. J. Clin. Med. 2025, 14, 2046. https://doi.org/10.3390/jcm14062046

AMA Style

Abbas J, Yousef M, Hamoud K, Joubran K. Low Back Pain Among Health Sciences Undergraduates: Results Obtained from a Machine-Learning Analysis. Journal of Clinical Medicine. 2025; 14(6):2046. https://doi.org/10.3390/jcm14062046

Chicago/Turabian Style

Abbas, Janan, Malik Yousef, Kamal Hamoud, and Katherin Joubran. 2025. "Low Back Pain Among Health Sciences Undergraduates: Results Obtained from a Machine-Learning Analysis" Journal of Clinical Medicine 14, no. 6: 2046. https://doi.org/10.3390/jcm14062046

APA Style

Abbas, J., Yousef, M., Hamoud, K., & Joubran, K. (2025). Low Back Pain Among Health Sciences Undergraduates: Results Obtained from a Machine-Learning Analysis. Journal of Clinical Medicine, 14(6), 2046. https://doi.org/10.3390/jcm14062046

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop