Blockchain-based hybrid method for diabetic management on daily basis through insulin dosage prediction

【Author】 Laxmi, Nakirekanti; Mohana, J.

【Source】SOFT COMPUTING

【影响因子】3.732

【Abstract】The transparency facilitated by blockchain technology allows anyone on the network to view all the blockchain's data. This feature can be of much use in diabetes management. The daily activities of a diabetic patient need to be monitored to calculate the ideal medicine dosage to be administered. The diabetic patient dataset has different parameters, such as skin thickness, insulin level, blood pressure, diabetes pedigree function, demography, drug intake levels, age, glucose, BMI, and pregnancy, to calculate the dosage level. Diabetic patient activities are collected through a telegram to estimate the right medicine dosage for the diabetic patient accurately. These data obtained from the diabetic patient consist of both clinical and non-clinical information. So, a hybrid algorithm needs to be proposed for classifying diabetic disorders. In this paper, the authors apply hybrid methods such as fuzzy, KNN, neural networks, and multiple linear regressions (MLR) to accurately calculate the ideal medicine dosage for diabetic patients from the diabetic patient dataset. The diabetic patient dataset is fed to a pre-trained neural network algorithm to enhance accuracy. The proposed hybrid algorithm performs better than the other methods regarding regression, response, and accuracy. The accuracy obtained is about 94%, which is higher than other methods.

【Keywords】Blockchain technology; Drug dosage; Fuzzy; KNN; Neural network; Multiple linear regression (MLR)

【发表时间】2023 2023 JUL 7

【收录时间】2023-07-30

【文献类型】实验仿真

【主题类别】

区块链应用-实体经济-医疗领域