Paper Title
MACHINE LEARNING BASED MONITORING SYSTEM FOR ELDERLY HOME ALONE

Abstract
Withtheincreasingnumberofolderpersonsall over the world comes the growing trend of the elderly living alone. Although independence is crucial for psychologicalandemotionalhealth,livingindependently can pose severe health and safety hazards such as falls, medicalemergencies,andinactivityforextendedperiods of time. There are seriouslyissues with traditional monitoring systems such as rein on active usage, limited two-way response and dealing with emergencies. To alleviate these issues, the present study has developed a multi-facetedmachinelearningbasedmonitoringsystem which aims to improve safety, health monitoring and quality of life for elderly living alone.The system’s key feature is based on deep learning algorithms such as ConvolutionalNeuralNetworks(CNNs)forthepurposeof video-based fall detection and the Recurrent Neural Networks (RNNs) for the purpose of time-series data analysis. These models utilize considerably large databases in order to have a higher precision level of normal activity and critical occurrences such that there are a few false alarms and the application is effective in realsituations.Further,aspecificalgorithmisalsotrained forabnormalitydetectionforidentifyingslightchangesin health parameters such as changes in heart rates or breath rates that signify health threats. Toensureuserconfidentiality,thedesignofthesystemis such that it incorporates end to end encryption features, local data storage and complies with data protection normsaswellasgeneralethicalissues.Thealertsaresent to caregivers or emergency responders using an application that allows chronic monitoring of the user’s status remotely for instant updates through a secure medium. Moreover, the system incorporates features of adaptive learning that enable it to learn an individual user’scharacteristicsovertimeforimproved efficiency in monitoring. Thedesignedsystememploys multipletypesofinternetof things (IOT) including motion sensors, wearable sensors and ambient sensors to continuously capture data about activity, vital signsand surrounding environment. Medical alert system is based on nurses taking such a number of algorithms as Once medical warning system detects danger, the devices notifying responsible persons such as caregiverscontactshavealreadybeencreated.Asystemof time series analysis, anomaly recognition and predictive modelling approacheshasbeenemployed inthispaperin order to present the monitoring system for elderly people. Keywords - Elderly Care Technology, Machine Learning inHealthcare,ElderlyMonitoringSystem,FallDetection