Skip to main content

Use of artificial neural networks in the prognosis of musculoskeletal diseases—a scoping review

Abstract

To determine the current evidence on artificial neural network (ANN) in prognostic studies of musculoskeletal diseases (MSD) and to assess the accuracy of ANN in predicting the prognosis of patients with MSD. The scoping review was reported under the Preferred Items for Systematic Reviews and the Meta-Analyses extension for Scope Reviews (PRISMA-ScR). Cochrane Library, Embase, Pubmed, and Web of science core collection were searched from inception to January 2023. Studies were eligible if they used ANN to make predictions about MSD prognosis. Variables, model prediction accuracy, and disease type used in the ANN model were extracted and charted, then presented as a table along with narrative synthesis. Eighteen Studies were included in this scoping review, with 16 different types of musculoskeletal diseases. The accuracy of the ANN model predictions ranged from 0.542 to 0.947. ANN models were more accurate compared to traditional logistic regression models. This scoping review suggests that ANN can predict the prognosis of musculoskeletal diseases, which has the potential to be applied to different types of MSD.

Peer Review reports

Introduction

Artificial intelligence (AI) has emerged as an opportunity allowing numerous practical applications [1, 2]. Artificial neural network (ANN), as an important branch of modern artificial intelligence technology, have been widely used in modern medicine due to their strong learning capability and stable feature recognition and prediction of functions [3, 4]. ANN is an information processing system established by imitating the structure and function of the neural network of the brain. Dr. Robert H. Nielsen, the inventor of the neural computer, defines a neural network as a computational system consisting of many simple, highly interconnected processing elements that can handle real-world problems by dynamically reacting to external inputs [5]. ANN can extract feature information from existing clinical experience and massive external input data and then perform self-learning, so ANN does not require a detailed description of the disease, but only basic information about the patients to obtain the corresponding diagnosis and treatment plan [6].

The application of ANN may be particularly useful in areas such as MSD as various clinical indicators related to musculoskeletal diseases are suitable for processing in ANN. Since ANN is able to use a large amount of those clinical data for machine learning, it may generate a stable clinical prediction model. In addition, musculoskeletal diseases (MSD) are associated with high morbidity and mortality and also lead to high healthcare costs [7, 8]. Globally, MSDs account for 21% of total morbidity and affect more than 25% of the population [9]. In the United States, approximately 130 million health care visits and approximately 70 million physician visits are associated with MSD each year [10], and MSD patients account for more than 25% of emergency department visits [11]. MSD is also the second most common cause of disability worldwide, with an estimated 45% increase in disability due to MSD disease, particularly osteoarthritis (OA), from 1990 to 2010, and the number of people suffering from MSD is expected to continue to increase with the impact of obesity, sedentary immobility, and an aging population [12,13,14].

Given the high prevalence and variety of MSDs (from tendon injuries in young athletes to degenerative diseases in the elderly), and the fact that some disease types are chronic or even incurable, finding a method that can effectively determine prognosis can help MSD patients better manage their disease and alleviate the burden of disease [15, 16], and may also reduce healthcare costs.

There is emerging evidence, that ANN can be applied to predict injury rates or treatment outcome in MSD. In postmenopausal women, An ANN model was used to predict fragility fractures using the bone strain index (BSI) and dual-energy x-ray absorptiometry (DXA), achieving a prediction accuracy of 79.56% [17]. In patients with chronic plantar fasciitis (CPF), ANN was used to determine the predictive factors for minimum clinically successful therapy (MCST) after extracorporeal shockwave treatment and found that the overall accuracy of the predictive model was 92.5% [18]. In patients with lumbar disc herniation (LDH), ANN model could predict the efficiency of hospitalization satisfaction with an accuracy of 96% [19]. Prognosis studies aim to predict the likelihood of disease progression related to different events (e.g., bone fracture, conjunctivitis) or explore the factors influencing disease outcomes [20, 21]. The use of ANN in prognosis may help doctors and patients to better understand the status and progression of the patient’s disease, resulting in individualized and more appropriate clinical decisions, which may reduce medical costs and improve recovery outcomes.

ANN has the potential to predict the prognosis of MSD by various variables such as patient's age, gender, treatment modality and disease severity. Therefore, the aims of this scoping review were threefold: (1) compile articles on the prognostic application of ANN in MSD, (2) investigate the accuracy of ANN in predicting the prognosis of MSD, (3) whether ANN has better predictive ability than other models.

Methods

This scoping review complies with all of the Preferred Items for Systematic Reviews and the Meta-Analyses extension for Scope Reviews (PRISMA-SCR) guidelines [22] and reports the required information accordingly. In addition, we also implemented the stages set out by Arksey and O'Malley [23] in the current scoping review. The protocol for this scoping review (https://doi.org/10.17605/OSF.IO/7UGFV) was registered at the Open Science Framework Registries (OSF).

Identifying the research question

This scoping review examined peer-reviewed articles on the use of ANN for prognosis prediction in MSD. Our scoping review identified the following questions: 1. Can ANN use basic clinical information of patients with musculoskeletal disorders to predict the prognosis of patients? 2. How effective and accurate was the ANN model used in prognosis studies? 3. Was ANN more effective than other machine learning methods or logistic regression? 4. What metrics were used in the included studies to predict prognostic outcomes?

Identifying relevant studies

Inclusion and exclusion criteria were determined on the basis of our study objectives.

Inclusion criteria

Studies applying ANN to predict the prognosis of musculoskeletal diseases, written in English.

Exclusion criteria

Studies not related to musculoskeletal diseases, not written in English, duplicate publications, unpublished studies, literature review papers, letters to the editor, conference abstracts and animal model studies.

Study selection

To ensure an extensive search for the inclusion of relevant articles, we searched the Cochrane library, Embase, Pubmed and the Web of science core collection. The retrieval time range was from the establishment of the database to January 7th, 2023. We use medical subject headings (MeSH) to facilitate literature retrieval, with three main subject headings: artificial neural networks; musculoskeletal diseases; prognosis. We applied different search strategies in different databases, and the full documentation of the final search strategy is available in the supplementation file. The database search results were imported into Endnote X9 (Thomson Reuters, NY, USA) and duplicates were removed. In order to include as much relevant literature as possible, we first performed Mesh terms searches with artificial neural networks, musculoskeletal diseases, and prognosis combined strategy, and then performed keywords researches. Titles and abstracts of retrieved articles were independently read and reviewed by FJQ and JFL, and any disagreement during the screening process was resolved through discussion and consensus with the third reviewer (RRZ). After the full text was obtained, the data was extracted from the paper.

Charting the data

Microsoft Excel (Version 2019) was used for the extraction of study data. Data charting was performed according to our proposed questions and the information extracted included (1) basic information about the study (authors, region, year of publication, sample size, study purpose, study design), (2) characteristic information of patients (age, disease type) and (3) ANN effect evaluation method, accuracy, and platform for modeling.

Collating, summarizing, and reporting the results

A total of 294 records were retrieved from the four databases, leaving 246 articles after removing duplicates and non-English studies. After reviewing the titles and abstracts according to inclusion and exclusion criteria, 205 articles were excluded. After reading the full text, 23 articles were excluded. The flowchart of the article retrieval and screening is shown in Fig. 1. We did not perform a quality assessment due to inconsistencies in the types of studies included in our study.

Fig. 1
figure 1

Study selection process (according to the PRISMA-ScR guidelines [22])

Results

Eighteen papers were finally included in the systematic analysis (Table 1).

Table 1 Basic information of included studies

Characteristics of included studies

The included studies were from sixteen journals, with only 2 articles each were published in Clinical Orthopaedics and Related Research [24, 25] and Medicina [26, 27]. Thirty-three percent of these studies (7/18) were conducted in the United States [24, 25, 28,29,30,31,32], with 17% (3) in China [26, 33, 34]. Study designs included cohort study (83%; 15) and cross-sectional database study (17%; 3). The included studies investigated 16 different musculoskeletal diseases. The detailed information of study characteristics was shown in Table 2.

Table 2 Demographics of subjects

Characteristics of participants

The mean age of the patients ranged from 12.5 to 100.0 years [35, 36] (Table 2). Two of the studies included exclusively female or male participants [24, 37]. The sample size ranged from 58 to 10534 [26, 36]. Two studies [28, 38] did not provide data on the age and sex of participants.

Effects of ANN in prognosis

The areas under the curve (AUC) served as a metric to evaluate the accuracy of ANN. The overall accuracy ranged from 0.542 to 0.947 (Table 3). 80.0% (8/10) of the studies showed that ANN had a better prediction accuracy than logistic regression (LR) or other prediction models. Eight studies did not compare ANN with other models, and two study found that ANN model had lower prediction accuracy than gradient boosting machine (GBM) and extreme gradient boosted machine (XGBoost). MATLAB was the most frequently (3 times) used platform in ANN modeling.

Table 3 Characteristics of studies in terms of ANN modeling and accuracy

Discussion

In the treatment and rehabilitation of musculoskeletal diseases, the consideration of different symptoms and demographic data to accurately predict clinical outcomes can aid in the clinical decision-making process to provide effective and adequate treatment for patients. The main findings of this scoping review were that in different types of musculoskeletal diseases, ANN can provide accurate predictions regarding the prognosis of patients and more accurate compared to other models.

In the prognostic studies of musculoskeletal diseases, ANN was able to make accurate predictions using demographic characteristics and patient clinical characteristics as the main parameters (features). Using bone mineral density and the bone strain index as parameters, ANN predicted the occurrence of vertebral fractures (VF) in postmenopausal women in 79.56% of cases [17]. When trabeculae microstructure parameters were used as the main variable, ANN models (AUC = 0.928) were more accurate than LR and random forest (RF) in predicting marginal bone loss [39]. Using disease history and lifestyle habits of 1419 patients as parameters to predict the risk of osteoporosis in adults, the ANN model was able to accurately predict the risk of osteoporosis (AUC = 0.901), outperforming the predictive power of Deep Belief Network (DBN), Support Vector Machine (SVM) and combinatorial heuristic method (Genetic Algorithm—Decision Tree) [40]. It is encouraging that in different types of MSD, ANNs are more accurate and outperform other prediction models in disease risk prediction.

ANN can also be applied in the area of predicting rehabilitation decisions and rehabilitation outcomes, with predictions being accurate. In a study using the demographic and clinical characteristics of 170 patients to predict rehabilitation options for patients with osteoarthritis of the knee, the developed medical decision support system was able to accurately predict treatment options for 87% of patients, thus effectively assisting clinical rehabilitation staff to develop OA rehabilitation plans [41]. A study using ANN to predict patient function one year after spinal cord injury found that ANN were highly accurate in predicting walking status (AUC range between 0.86 and 0.90) and moderately accurate when used to predict non-walking outcomes (AUC between 0.70 and 0.82), and that models generated by artificial neural networks performed better than LR [42]. The application of ANN in the rehabilitation can simplify the cumbersome manual assessment process and allow for accurate prediction of some parameters that are difficult to assess quantitatively in the clinic, saving clinical diagnosis and treatment time and reducing the workload of rehabilitation physicians and therapists.

The accuracy of prognostic prediction using ANN models varied among diseases. In patients undergoing elective adult spinal deformity procedures [43], the accuracy of ANN in predicting venous thromboembolism (VTE) and wound incidence can be considered as poor and failed (AUCVTE = 0.542; AUCwound = 0.606) according to generally accepted AUC accuracy classification practice [44, 45]. Compared to the other included studies, in which the ANN model prediction accuracy was higher than 0.7, a relatively low number of features was used in the ANN model. While in this study [43] 8 features were used to predict 4 different symptoms, other studies used 10 [24] to 25 [37] features to predict one symptom. A possible reason for the discrepancy in accuracy may thus be that insufficient relevant features were included in ANN models.

ANN has higher accuracy compared to traditional logistic regression models. Research in the field of bioengineering has demonstrated that ANN are superior to traditional statistical models in terms of their ability to analyze nonlinear relationships, their ability to handle relevant independent variables, and their classification accuracy [46]. The advantages of ANN are mainly in the following three aspects:

  • 1. Multi-layer network structure: ANN individual neurons cooperate with each other and form a network synergy when processing information, maintaining their independence while sharing and cascading the output results with other neurons, and making the results more reliable through the use of multiple hidden layers [47, 48];

  • 2. Adaptive: According to the characteristics of the information in the input neural network, ANN can continuously establish new structures consistent with external changes through learning, extracting, and collecting information required for specific tasks from the data, and summarize the acquired knowledge, thereby improving the ability of data processing [49];

  • 3. Accommodating data deficiencies: In contrast to traditional models that require data completeness, artificial neural networks can maintain the validity of the model even when the patient's data is incomplete [50,51,52].

Limitations

Only 56% (10/18) of the studies included in this scoping review compared the accuracy between ANN and other models, which may have limited the judgment of the effectiveness of ANN applications. This study also found that although ANN has shown excellent accuracy in its application, applying it to construct predictive models may be problematic and cause over-fitting. As a consequence, the results in the receiver operating curve may be better than actual, as patients are highly selected for inclusion. Therefore, the ANN model still needs to be externally validated after its construction to demonstrate its generalizability in patients.

Prospective

  • 1. Promote the application of ANN in the MSD. Incorporating artificial neural networks into clinical settings can enable clinicians to predict disease progression and functional recovery faster and more accurately.

  • 2. Optimize the quality of the data set. The model should be built by selecting samples with different etiology, disease duration, age, ethnicity, and sex, the number of layers and complexity of the algorithm model should be determined according to the amount of data to ensure that the trained ANN models have better clinical adaptability and benefit the clinical treatment.

  • 3. Adjust legal regulations. The artificial intelligence technology represented by ANN requires a large amount of data input related to clinical parameters (e.g., images, and videos). Adequate laws related to the use of ANN in healthcare are necessary to ensure the protection of patient privacy, while reasonably allocating the responsibility in case of errors in artificial neural network models.

Conclusion

This scoping review provides preliminary evidence that ANN can provide accurate prognosis prediction for MSDs by demographic information of patients and clinical characteristics of diseases. ANN models are superior to other traditional prediction models such as LR and deserve to be tested and replicated in other MSD types. The weaknesses highlighted must be addressed in future studies to enable ANNs models to better contribute to clinical decision making.

Availability of data and materials

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

References

  1. Contreras I, Vehi J. Artificial intelligence for diabetes management and decision support: literature review. J Med Internet Res. 2018;20(5):e10775.

    Article  Google Scholar 

  2. Nikseresht A, et al. Using artificial intelligence to make sustainable development decisions considering VUCA: a systematic literature review and bibliometric analysis. Environ Sci Pollut Res Int. 2022;29(28):42509–38.

    Article  Google Scholar 

  3. Garcia-Vidal C, et al. Artificial intelligence to support clinical decision-making processes. EBioMedicine. 2019;46:27–9.

    Article  Google Scholar 

  4. Shaban-Nejad A, Michalowski M, Buckeridge DL. Health intelligence: how artificial intelligence transforms population and personalized health. NPJ Digit Med. 2018;1:53.

    Article  Google Scholar 

  5. Dande P, Samant P. Acquaintance to Artificial Neural Networks and use of artificial intelligence as a diagnostic tool for tuberculosis: A review. Tuberculosis (Edinb). 2018;108:1–9.

    Article  Google Scholar 

  6. Scott R. Artificial intelligence: its use in medical diagnosis. J Nucl Med. 1993;34(3):510–4.

    CAS  Google Scholar 

  7. Intriago M, et al. Bone Mass Loss and Sarcopenia in Ecuadorian Patients. J Aging Res. 2020;2020:1072675.

    Article  CAS  Google Scholar 

  8. Coll PP, et al. The prevention of osteoporosis and sarcopenia in older adults. J Am Geriatr Soc. 2021;69(5):1388–98.

    Article  Google Scholar 

  9. Cottrell MA, et al. Real-time telerehabilitation for the treatment of musculoskeletal conditions is effective and comparable to standard practice: a systematic review and meta-analysis. Clin Rehabil. 2017;31(5):625–38.

    Article  Google Scholar 

  10. National Research Council (US) and Institute of Medicine (US) Panel on Musculoskeletal Disorders and the Workplace. Musculoskeletal Disorders and the Workplace: Low Back and Upper Extremities. Washington (DC): National Academies Press (US); 2001. Executive Summary. Available from: https://www.ncbi.nlm.nih.gov/books/NBK222440/.

  11. Matifat E, et al. Benefits of Musculoskeletal Physical Therapy in Emergency Departments: A Systematic Review. Phys Ther. 2019;99(9):1150–66.

    Article  Google Scholar 

  12. Hotez PJ, et al. The global burden of disease study 2010: interpretation and implications for the neglected tropical diseases. PLoS Negl Trop Dis. 2014;8(7):e2865.

    Article  Google Scholar 

  13. Hoy D, et al. The global burden of neck pain: estimates from the global burden of disease 2010 study. Ann Rheum Dis. 2014;73(7):1309–15.

    Article  Google Scholar 

  14. Bucki FM, et al. Scoping Review of Telehealth for Musculoskeletal Disorders: Applications for the COVID-19 Pandemic. J Manipulative Physiol Ther. 2021;44(7):558–65.

    Article  Google Scholar 

  15. Raeissadat SA, et al. Autologous conditioned serum applications in the treatment of musculoskeletal diseases: a narrative review. Future Sci OA. 2022;8(2):Fso776.

    Article  CAS  Google Scholar 

  16. Gheno R, et al. Musculoskeletal disorders in the elderly. J Clin Imaging Sci. 2012;2:39.

    Article  Google Scholar 

  17. Ulivieri FM, et al. Bone Strain Index predicts fragility fracture in osteoporotic women: an artificial intelligence-based study. Eur Radiol Exp. 2021;5(1):47.

    Article  Google Scholar 

  18. Yin M, et al. Use of artificial neural networks to identify the predictive factors of extracorporeal shock wave therapy treating patients with chronic plantar fasciitis. Sci Rep. 2019;9(1):4207.

    Article  Google Scholar 

  19. Matis GK, et al. Prediction of Lumbar Disc Herniation Patients’ Satisfaction with the Aid of an Artificial Neural Network. Turk Neurosurg. 2016;26(2):253–9.

    Google Scholar 

  20. Altman DG. Systematic reviews of evaluations of prognostic variables. BMJ. 2001;323(7306):224–8.

    Article  CAS  Google Scholar 

  21. Moons KG, et al. Critical appraisal and data extraction for systematic reviews of prediction modelling studies: the CHARMS checklist. PLoS Med. 2014;11(10):e1001744.

    Article  Google Scholar 

  22. Tricco AC, et al. PRISMA Extension for Scoping Reviews (PRISMA-ScR): Checklist and Explanation. Ann Intern Med. 2018;169(7):467–73.

    Article  Google Scholar 

  23. Arksey H, O’Malley L. Scoping studies: towards a methodological framework. Int J Soc Res Methodol. 2005;8(1):19–32.

    Article  Google Scholar 

  24. Alfieri KA, et al. Preventing Heterotopic Ossification in Combat Casualties—Which Models Are Best Suited for Clinical Use? Clin Orthop Relat Res. 2015;473(9):2807–13.

    Article  Google Scholar 

  25. Bevevino AJ, et al. A Model to Predict Limb Salvage in Severe Combat-related Open Calcaneus Fractures. Clin Orthop Relat Res. 2013;472(10):1–8.

  26. Chen CY, et al. Artificial Neural Network and Cox Regression Models for Predicting Mortality after Hip Fracture Surgery: A Population-Based Comparison. Medicina (Kaunas). 2020;56(5):243.

    Article  Google Scholar 

  27. Shin H, et al. Assessment of Therapeutic Responses Using a Deep Neural Network Based on 18F-FDG PET and Blood Inflammatory Markers in Pyogenic Vertebral Osteomyelitis. Medicina (Kaunas). 2022;58(11):1693.

    Article  Google Scholar 

  28. Jalali A, et al. Machine Learning Applied to Registry Data: Development of a Patient-Specific Prediction Model for Blood Transfusion Requirements During Craniofacial Surgery Using the Pediatric Craniofacial Perioperative Registry Dataset. Anesth Analg. 2021;132(1):160–71.

    Article  Google Scholar 

  29. Kim JS, et al. Predicting Surgical Complications in Patients Undergoing Elective Adult Spinal Deformity Procedures Using Machine Learning. Spine Deformity. 2018;6(6):762–70.

    Article  Google Scholar 

  30. Lu Y, et al. Understanding Anterior Shoulder Instability Through Machine Learning: New Models That Predict Recurrence, Progression to Surgery, and Development of Arthritis. Orthop J Sports Med. 2021;9(11):23259671211053330.

    Article  Google Scholar 

  31. Norgeot B, et al. Assessment of a Deep Learning Model Based on Electronic Health Record Data to Forecast Clinical Outcomes in Patients With Rheumatoid Arthritis. JAMA Netw Open. 2019;2(3):e190606.

    Article  Google Scholar 

  32. Scheer JK, et al. Development of a preoperative predictive model for major complications following adult spinal deformity surgery. J Neurosurg Spine. 2017;26(6):736–43.

    Article  Google Scholar 

  33. Su Q, Xu G. Endoscopic Surgical Treatment of Osteoarthritis and Prognostic Model Construction. Comput Math Methods Med. 2022;2022:1799177.

    Article  Google Scholar 

  34. Wang Y, et al. Modeling using clinical examination indicators predicts interstitial lung disease among patients with rheumatoid arthritis. PeerJ. 2017;2017(2):1–15.

    Google Scholar 

  35. Bowman A, et al. A prognostic model for the patient-reported outcome of surgical treatment of carpal tunnel syndrome. Muscle Nerve. 2018;58(6):784–9.

    Article  Google Scholar 

  36. Yahara Y, et al. A deep convolutional neural network to predict the curve progression of adolescent idiopathic scoliosis: a pilot study. BMC Musculoskelet Disord. 2022;23(1):610.

    Article  Google Scholar 

  37. Eller-Vainicher C, et al. Recognition of morphometric vertebral fractures by artificial neural networks: analysis from GISMO Lombardia Database. PLoS ONE. 2011;6(11):e27277.

    Article  CAS  Google Scholar 

  38. Almhdie-Imjabbar A, et al. Prediction of knee osteoarthritis progression using radiological descriptors obtained from bone texture analysis and Siamese neural networks: data from OAI and MOST cohorts. Arthritis Res Ther. 2022;24(1):66.

    Article  CAS  Google Scholar 

  39. Zhang H, et al. Trabeculae microstructure parameters serve as effective predictors for marginal bone loss of dental implant in the mandible. Sci Rep. 2020;10(1):18437.

    Article  CAS  Google Scholar 

  40. Wang Y, et al. Prediction model for the risk of osteoporosis incorporating factors of disease history and living habits in physical examination of population in Chongqing, Southwest China: based on artificial neural network. BMC Public Health. 2021;21(1):991.

    Article  Google Scholar 

  41. Hawamdeh ZM, et al. Development of a decision support system to predict physicians’ rehabilitation protocols for patients with knee osteoarthritis. Int J Rehabil Res. 2012;35(3):214–9.

    Article  Google Scholar 

  42. Belliveau T, et al. Developing Artificial Neural Network Models to Predict Functioning One Year After Traumatic Spinal Cord Injury. Arch Phys Med Rehabil. 2016;97(10):1663-1668.e3.

    Article  Google Scholar 

  43. Kim JS, et al. Predicting Surgical Complications in Patients Undergoing Elective Adult Spinal Deformity Procedures Using Machine Learning. Spine Deform. 2018;6(6):762–70.

    Article  Google Scholar 

  44. Metz CE. Basic principles of ROC analysis. Semin Nucl Med. 1978;8(4):283–98.

    Article  CAS  Google Scholar 

  45. Li F, He H. Assessing the Accuracy of Diagnostic Tests. Shanghai Arch Psychiatry. 2018;30(3):207–12.

    Google Scholar 

  46. Renganathan V. Overview of artificial neural network models in the biomedical domain. Bratisl Lek Listy. 2019;120(7):536–40.

    CAS  Google Scholar 

  47. Ozkan O, et al. A Study on the Effects of Sympathetic Skin Response Parameters in Diagnosis of Fibromyalgia Using Artificial Neural Networks. J Med Syst. 2016;40(3):54.

    Article  Google Scholar 

  48. Cao B, et al. Status quo and future prospects of artificial neural network from the perspective of gastroenterologists. World J Gastroenterol. 2021;27(21):2681–709.

    Article  Google Scholar 

  49. Menke NB, et al. A retrospective analysis of the utility of an artificial neural network to predict ED volume. Am J Emerg Med. 2014;32(6):614–7.

    Article  Google Scholar 

  50. Moon S, et al. Artificial neural networks in neurorehabilitation: A scoping review. NeuroRehabilitation. 2020;46(3):259–69.

    Article  Google Scholar 

  51. Sharpe PK, Solly RJ. Dealing with missing values in neural network-based diagnostic systems. Neural Comput Appl. 1995;3(2):73–7.

    Article  Google Scholar 

  52. Śmieja, M., et al., Processing of missing data by neural networks. Advances in neural information processing systems, 2018. 31.

Download references

Acknowledgements

Not applicable.

Funding

Open Access funding enabled and organized by Projekt DEAL. Fanji Qiu is supported by a grant from the China Scholarship Council (grant no. 202106520004). The article processing charge was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – 491192747 and the Open Access Publication Fund of Humboldt-Universität zu Berlin.

Author information

Authors and Affiliations

Authors

Contributions

F.Q. had the idea for the study conception and design. F.Q., J.L. and R.Z. selected the studies for inclusion and extracted data. F.Q. performed the statistical analyses and wrote the first draft. K.L. critically revised the paper for important intellectual content. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication. All authors have read and agreed to the published version of the manuscript.

Corresponding author

Correspondence to Fanji Qiu.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Qiu, F., Li, J., Zhang, R. et al. Use of artificial neural networks in the prognosis of musculoskeletal diseases—a scoping review. BMC Musculoskelet Disord 24, 86 (2023). https://doi.org/10.1186/s12891-023-06195-2

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12891-023-06195-2

Keywords