Artificial Intelligence-Based Cyber-Physical System for Severity Classification of Chikungunya Disease

Abstract

Background: Artificial intelligence techniques are widely used in solving medical problems. Recently, researchers have used various deep learning techniques for the severity classification of Chikungunya disease. But these techniques suffer from overfitting and hyper-parameters tuning problems. Methods: In this paper, an artificial intelligence-based cyber-physical system (CPS) is proposed for the severity classification of Chikungunya disease. In CPS system, the physical components are integrated with computational algorithms to provide better results. Random forest (RF) is used to design the severity classification model for Chikungunya disease. However, RF suffers from overfitting and poor computational speed problems due to complex architectures and large amounts of connection weights. Therefore, an evolving RF model is proposed using the adaptive crossover-based genetic algorithm (ACGA). Results: ACGA can efficiently optimize the architecture of RF to achieve better results with better computational speed. Extensive experiments are performed by utilizing the Chikungunya disease dataset. Conclusion: Performance analysis demonstrates that ACGA-RF achieves higher performance as compared to the competitive models in terms of F-measure, accuracy, sensitivity, and specificity. The proposed CPS system can prevent users from visiting hospitals and can render services to patients living far away from hospitals.

Review Of Models & Techniques For Identification Of Human Health Diseases

“…While absolute values for accuracy & precision were available, other metrics like delay, complexity and scalability were evaluated in terms of fuzzy range sets of Low (L), Medium, High (H), and Very High (VH), which were decided based on their internal configurations and performance across different scenarios. Based on this strategy, table 1 showcases parameters for these models, Based on this analysis, it can be observed that DF [43], TPA GAN [4], PCA NET [17], THS GAN [50], SSDP [24], MA Net [30], TSA CNN [26], and RL [42] showcase high accuracy, while DLN [40], Fuzzy GBDT [22], PD Res Net [21], GERF [39], PC SVM [41], SSDP [24], SALL [25], THS GAN [50], ACGA [23], TSA CNN [26], and RL [42] showcase high precision, which makes them useful for a wide variety of real-time clinical use cases.…”

“…Their suggested model performs with 95.51% accuracy, 94.44% precision, 96.59% recall, 94.44% specificity, and an F1-score of 95.51% on the clinical gait dataset. According to [22], heart disease has a detrimental influence on people's lives since it often results in hospitalizations and fatalities. Early diagnosis and treatment depend on accurate prognosis.…”

“…TSA CNN [26], and RL [42] showcase high accuracy, whereas DLN [40], Fuzzy GBDT [22], PD Res Net [21], GERF [39], PC SVM [41], SSDP [24], SALL [25], THS GAN [50], ACGA [ In terms of complexity, CT [6], SVM [15], SSDP [24], and LR [27] are able to achieve higher performance. On the other hand, in terms of delay, UESR [5], JHCP MMP [13], SVM [15], and SVM [47] are able to display quicker performance, and hence may be utilized for high-speed clinical deployments.…”

Guest Editorial Artificial Intelligence-Driven Biomedical Imaging Systems for Precision Diagnostic Applications

“…Despite these advances, problems such as the requirement for huge, annotated datasets, ethical considerations, and regulatory frameworks must be addressed to assure the responsible and ethical deployment of AI-driven biomedical imaging systems [3]. Collaboration among researchers, healthcare professionals, and politicians is critical to navigating these hurdles and promoting wider acceptance of innovative technologies.…”