Optimizing Control of IOT Device using Traditional Machine Learning Models and Deep Neural Networks

ABSTRACT

With the increasing threat of Covid-19 and now omicron infection across the world among people, there has been a significant surge in the demand for a fully-automated, self-controlled or mechanized ventilator which can provide sufficient air-pressure to weak human lungs continuously. It is our humble endeavor to mitigate the effects caused due to handful of trained-physicians over countless untreated patients and lack of enough health-infrastructure facilities to support in the time of dire need. We all dread losing another precious life on earth due to any one of the above mentioned reason. We have tried simulating the observations obtained from a lab-developed mechanical ventilator system under different lung settings. After preprocessing this dataset using NLP, training data is analysed to study the correlation between observations from numerous attributes. A couple of Machine Learning (LR, RF, SVM, LGBM) and Deep Learning (MLP, LSTM, Bi-LSTM) algorithms have been deployed to train our model individually, out of which Bi-LSTM performed exceptionally well above others. However, only after exhaustive clinical trials and recommendations a large of number of patients on life-support can get a new life through the large practical application of this device, in the near future. © 2022 IEEE.