Editorial: Soft computing and machine learning applications for healthcare systems

This special issue provides a platform for researchers, practitioners to present the theoretical and practical applications of soft computing and machine learning for efficiency and effectiveness in healthcare sector. Soft computing embraces the group of computational techniques that are based on Artificial Intelligence (AI) and natural selection to enable complex problems for which analytical formulations are not feasible to be solved in quick and effective way. Typical soft computing techniques include Artificial Neural Networks, Fuzzy logic, Evolutionary algorithms, Swarm intelligence, and other computational methods that are based on approximate reasoning and approximate modelling.Obagbuwa et al. presented a study on supervised machine learning for depression sentiment analysis. The study utilized machine learning model and sentiment analysis techniques to predict the level of depression earlier in social media users' post. Four machine learning models, namely Extreme Gradient Boosting (XGB) Classifier, Random Forest, Logistic Regression and Support Vector Machine (SVM) were employed for the prediction task. The study's findings highlighted the potential of utilizing machine learning models and sentiment analysis techniques for early detection of depression in social media users. The effectiveness of SVM and efficiency of logistic regression in terms of execution time, suggest their suitability for practical implementation in real-world scenarios.Asani et al. in their study developed a web-based plant diagnosis application called mobile-enabled Plant Diagnosis-Application (mPD-App). The mPD-App was proposed to serve as a veritable tool for farmers and agricultural stakeholders in Sub-Saharan Africa to detect and diagnose plant diseases effectively and efficiently. Convolutional neural network (CNN) model was engaged and overall accuracy of 93.91% was achieved.Morapedi and Obagbuwa presented air pollution particulate matter (PM2.5) prediction in South African cities using machine learning techniques. The study entails the use of machine techniques such as Cat Boost Regression, Extreme Gradient Boosting Regressor, Random Forest Classifier, Logistic Regression, Support Vector Machine, K-Nearest Neighbour and Decision Tree to identify air pollution in terms of time, cost and efficiency in different scenarios so that the system can select optimal way for their needs. The findings is that Cat Boost Regressor and Extreme Gradient Boosting Regressor predicted the latest PM2.5 concentrations for South African Cities with recording stations using past dated recordings while Random Forest Classifier, Logistic Regression, Support Vector Machine, K-Nearest Neighbour and Decision Tree predicted the Air Quality Index status for South African cities.Malik et al. in their study explore dermoscopic structures for melanoma lesions classification. The study explored of AI vulnerability in discerning melanoma from benign lesions using features of size, color, and shape. Tests with artificial and natural variations revealed a notable decline in accuracy, emphasizing the necessity for additional information, such as dermoscopic structures. The methodology used are Transformers and CNN-based models to classify these images based on dermoscopic structures. Classification results are validated using feature visualization. To assess model susceptibility to image variations, classifiers are evaluated on test sets with original, duplicated, and digitally modified images. Additionally, testing is done on ISIC 2016 images. The study focused on three dermoscopic structures crucial for melanoma detection: Blue-white veil, dots/globules, and streaks. In evaluating model performance, adding convolutions to Vision Transformers proves highly effective for achieving up to 98% accuracy. CNN architectures like VGG-16 and DenseNet-121 reach 50-60% accuracy, performing best with features other than dermoscopic structures. Vision Transformers without convolutions exhibit reduced accuracy on diverse test sets, revealing their brittleness. OpenAI Clip, a pre-trained model, consistently performs well across various test sets. To address brittleness, a mitigation method involving extensive data augmentation during training and 23 transformed duplicates during test time, sustains accuracy. Grassi et al. the authors presented a study on enhanced sleep staging with artificial intelligence: a validation study of software for sleeping scoring. The study investigated extensively Manual Sleep Staging (MSS) which uses polysomnography and STAGER which is a software program based on machine learning algorithm to perform automatic sleep staging using only ECG signals from polysomnography. The findings revealed several agreement statistics between the automatic sleep staging (ASS) and MSS, among the different MSSs, and their differences were calculated. Bootstrap resampling was used to calculate 95% confidence intervals and the statistical significance of the differences. STAGER's ASS was most comparable with, or statistically significantly better than the MSS, except for a partial reduction in the positive percent agreement in the wake stage. These promising results indicate that STAGER software can perform ASS of inpatient polysomnographic recordings accurately in comparison with MSS.Ferhi al. developed enhanced diagnostic accuracy in symptom-based health checker: a comprehensive machine learning approach with clinical vignettes and benchmarking. The study focuses on evaluating and optimizing machine learning models using a dataset of 10 diseases and 9,572 samples. The authors selected and optimised the following models: Decision Tree, Random Forest, Naïve Bayes, Logistic Regression and K-Nearest Neighbors. The evaluation metrics used accuracy, F1 scores. ROC-AUC and precisionrecall curves to assess the model performance. The ROC-AUC curves revealed that model performance improved with increasing complexity. Precision-recall curves were particularly useful in evaluating model sensitivity in imbalanced dataset scenarios. Clinical vignettes demonstrated the robustness of the models in providing accurate diagnoses.Khan and O'Sullivan presented a study on a comparison of the diagnostic ability of large language models (LLM) in challenging clinical cases. The study compared different performance characteristics of common LLMS utility in solving complex clinical cases and assess the utility of a novel tool to grade LLM output. The authors performed a comparative analysis of three LLM models-Bing, Chat GPT, and Gemini-across a diverse set of clinical cases as presented in the New England Journal of Medicines case series. The results revealed that models performed differently when presented with identical clinical information, with Gemini performing best. The grading tool had low interobserver variability and proved a reliable tool to grade LLM clinical output.In the study, a multi-task AutoEncoder (AE) was proposed to automate anomaly detection of VMAT for lung cancer patients. Among the four tested AE models, the proposed multi-task AE model achieves the highest values in AUC(0.964), accuracy (0.821), precision (0.471), and F1 score (0.632), and the lowest value in FPR(0.206). The proposed multi-task AE model using two-dimensional (2D) feature maps can effectively detect anomalies in radiotherapy plans for lung cancer patients. Compared to the other existing AE models, the multi-task AE is more accurate and efficient.Liu et al. presented modelling disagreement in automatic data labelling for semisupervised learning in clinical natural language processing. The authors investigated the quality of uncertainty estimates from a range of current state-of-the-art predictive models applied to the problem of observation detection in radiology reports. The problem remains understudied for Natural Language Processing in the healthcare domain. The findings revealed that Gaussian Processes (GP) provides superior performance in quantifying the risk of three uncertainty labels based on negative log predictive probability (NLPP) evaluation metric and mean maximum predicted confidence levels (MMPCL), whilst retaining strong predictive performance.Wang et al. applied machine learning in intelligent systems: knowledge graph-enhanced ophthalmic contrastive learning with clinical profile prompt. The study investigated the application of machine learning techniques by integrating knowledge graphs with contrastive learning and utilizing "clinical profile" prompts to refine the performance of the ophthalmology-specific large language model, MeEYE, which is built on the CHATGLM3-6B architecture. The study employed a novel methodological framework that incorporates domain-specific knowledge through knowledge graphs and enhances feature representation using contrastive learning. The MeEYE model is fine-tuned with structured clinical knowledge, enabling it to better distinguish subtle yet significant ophthalmic features. The experimental findings demonstrate that integrating knowledge graphs and contrastive learning into the MeEYE model significantly improves both diagnostic accuracy and model interpretability. Comparative analyses against baseline models reveal that the proposed approach enhances the identification of ophthalmic conditions with higher precision and clarity.Finally, Hsu et al. explored the pivotal variables of tongue diagnostic between patients with chronic kidney disease and health participants. The study examined the relationship between tongue characteristics and chronic kidney disease (CKD) severity using an automatic tongue diagnosis system (ATDS) which captures tongue images noninvasively to provide objective diagnostic information. Cross-sectional, case-control study was conducted from July 1, 2019, to December 31, 2021. Participants were divided into three groups based on estimated glomerular filtration rate (eGFR): control (eGFR > 60 ml/min/1.732), CKD stage 3 (30 ≤ eGFR < 60 ml/min/1.732), and CKD stage 4-5 (eGFR < 30 ml/min/1.732). Tongue images were analyzed using ATDS to extract nine primary features: tongue shape, color, fur, saliva, fissures, ecchymosis, tooth marks, and red dots. The study revealed that significant differences in the fur thickness, tongue color, amount of ecchymosis, and saliva among three groups. Ordinal logistic regression indicated that pale tongue color (OR: 2.107, P < 0.001), bluish tongue color (OR: 2.743, P = 0.001), yellow fur (OR: 3.195, P < 0.001), wet saliva (OR: 2.536, P < 0.001), and ecchymoses (OR: 1.031, P = 0.012) were significantly associated with increased CKD severity.The study consists of the contributions of each author in the research topic. The limitations across the studies are the used of fewer soft computing techniques. The gaps observed in the study is the minimal coverage of applications of healthcare systems with soft computing and machine learning. The future research priority areas is the fusion of machine learning techniques for effective diagnosis of diseases in the healthcare system domain.