Evaluating the efficacy of deep learning models for knee osteoarthritis prediction based on Kellgren-Lawrence grading system

•The idea behind transfer learning for KOA grade image classification is that if the existing models are already trained on a large and general dataset, these models will effectively serve as generic models to fulfil the study's objectives.•The authors have focused on the novel approach for grading knee osteoarthritis using KL grading and deep learning techniques.•This study aimed to explore the use of transfer learning convolutional neural network for medical image classification applications using KOA as a clinical scenario, comparing twelve different transfer learning DL models.•The evaluation is done for 12 different DL models and this comparison is done for a maximum of 8 models only in the previously published articles.•The findings of this study will pave the way for future development in the field, with the development of more accurate models and tools.•These tools can improve medical image classification by machine learning and will give valuable insight into orthopaedic disease pathology.•The number of images in the dataset is also increased when compared with the existing published articles.•The study paves the way for extrapolating the learning using ML to develop an automated KOA classification tool and enable healthcare professionals with better decision-making. Osteoarthritis of the knee, also known as OA has been determined that osteoarthritis of the knee is the leading cause of activity limitations and the development of disability, particularly in people who are older. The utilisation of artificial intelligence (AI) methodologies grounded in deep learning (DL) has yielded promising outcomes in the realm of radiographic interpretation. The utilisation of deep learning in the healthcare industry has yielded remarkable outcomes and elevated the benchmark for the quality of medical treatment. This study used knee OA as a clinical scenario to compare twelve transfer learning DL models for detecting the grade of KOA from a radiograph, compared their accuracy, and determined the best model for detecting KOA. The models exhibited a range of 30% to 98% in detecting the KOA. It was determined that MobileNet was responsible for the highest level of accuracy, which came in at 98.36%. It has high training and validation accuracy. The maximum loss was observed for EfficientNetB7. DL approaches created by skilled radiologists and orthopaedic specialists could help smaller hospitals learn and make more emergency room. This would be especially helpful in