Abstract:Medical research offers potential for disease prediction, like Multiple Sclerosis (MS). This neurological disorder damages nerve cell sheaths, with treatments focusing on symptom relief. Manual MS detection is time-consuming and error prone. Though MS lesion detection has been studied, limited attention has been paid to clinical analysis and computational risk factor prediction. Artificial intelligence (AI) techniques and Machine Learning (ML) methods offer accurate and effective alternatives to mapping MS progression. However, there are challenges in accessing clinical data and interdisciplinary collaboration. By analyzing 103 papers, we recognize the trends, strengths and weaknesses of AI, ML, and statistical methods applied to MS diagnosis. AI/ML-based approaches are suggested to identify MS risk factors, select significant MS features, and improve the diagnostic accuracy, such as Rule-based Fuzzy Logic (RBFL), Adaptive Fuzzy Inference System (ANFIS), Artificial Neural Network methods (ANN), Support Vector Machine (SVM), and Bayesian Networks (BNs). Meanwhile, applications of the Expanded Disability Status Scale (EDSS) and Magnetic Resonance Imaging (MRI) can enhance MS diagnostic accuracy. By examining established risk factors like obesity, smoking, and education, some research tackled the issue of disease progression. The performance metrics varied across different aspects of MS studies: Diagnosis: Sensitivity ranged from 60 % to 98 %, specificity from 60 % to 98 %, and accuracy from 61 % to 97 %. Prediction: Sensitivity ranged from 76 % to 98 %, specificity from 65 % to 98 %, and accuracy from 62 % to 99 %. Segmentation: Accuracy ranged up to 96.7 %. Classification: Sensitivity ranged from 78 % to 97.34 %, specificity from 65 % to 99.32 %, and accuracy from 71 % to 97.94 %. Furthermore, the literature shows that combining techniques can improve efficiency, exploiting their strengths for better overall performance.

Misdiagnosis of Multiple Sclerosis: Past, Present, and Future

Misdiagnosis of Multiple Sclerosis: Frequency, Causes, Effects, and Prevention

Misdiagnosis and underdiagnosis of multiple sclerosis: A systematic review and meta-analysis

Current and future role of MRI in the diagnosis and prognosis of multiple sclerosis

Differential diagnosis of suspected multiple sclerosis: an updated consensus approach

Multiple sclerosis: clinical aspects

Shifting our attention earlier in the multiple sclerosis disease course

Assessment of lesions on magnetic resonance imaging in multiple sclerosis: practical guidelines

Clinical Characteristics and Misdiagnosis Analysis of Tumour-like Multiple Sclerosis

Multiple sclerosis diagnosis: Knowledge gaps and opportunities for educational intervention in neurologists in the United States

A Comprehensive Review on Multiple Sclerosis

Cortical lesions, central vein sign, and paramagnetic rim lesions in multiple sclerosis: emerging machine learning techniques and future avenues

An analytical review on the use of artificial intelligence and machine learning in diagnosis, prediction, and risk factor analysis of multiple sclerosis

Frequency of NMOSD misdiagnosis in a cohort from Latin America: Impact and evaluation of different contributors

Differential diagnosis of suspected multiple sclerosis: considerations in people from minority ethnic and racial backgrounds in North America, northern Europe, and Australasia

Current opinion: Racial and ethnic health disparities in multiple sclerosis: considering the social determinants of health

Predicting multiple sclerosis disease progression and outcomes with machine learning and MRI-based biomarkers: a review

Typical and Emerging Diagnostic MRI Features in Multiple Sclerosis

A Review on the Clinical Diagnosis of Multiple System Atrophy

Differential Diagnosis of Suspected Multiple Sclerosis in Pediatric and Late-Onset Populations: A Review

Recent advances in development of nanomedicines for multiple sclerosis diagnosis