Clinical Prediction using Machine Learning-based IoT for E-Healthcare Systems

Abstract

Machine learning is one of the fundamental approaches to nature the data produced by Internet of Things (IoT) and has drastic enhancements to enhance the decision-making fields such as education, security, business, and especially the healthcare sector. This integration of ML with IoT makes it possible for healthcare systems to generate patterns from big data which can improve the efficiency of predictive analytics and referrals. It also helps in the development of templates for medical records and real-time patient care, and the identification of patient diagnosis electronically. Though, performance of the ML algorithms can be affected by the dataset on the basis of which clinical decisions are made. Such variations clearly demonstrate the requirement to gain a deeper understanding of how several ML approaches work in terms of IoT data analysis for healthcare. In the proposed study we are testing the Adaptive Neuro-Fuzzy Inference System (ANFIS) algorithms for tracking health parameters. We first employ the above-mentioned approach for training and validation of the ML models on the database available at UCI. In the testing phase, the IoT devices record actual data of heart rate, blood pressure, posture, and temperature to forecast a health anomaly. Subsequently, an examination of accuracy of the aforementioned prediction is performed on the cloud stored data with reference to those produced by K-nearest neighbour (KNN) algorithm. Such an approach makes it possible to give adequate consideration to the offered assessment of the role of ML in increasing the efficiency of IoT-based healthcare systems with regard to early detection and comprehensive intervention practices. In this way, this research help to consider variability in prognostic outcomes for more accurate and precise clinical decision-making processes, and thus contribute in enhancing patient care and health in IoT environment.