ABSTRACT-Sensor systems can be used to assist elderly peopleliving alone at home. Sensor systems are able to perform different functionsranging from monitoring of older adults to predicting their functional healthstatus.Movement tracking of elderly people can be performed by severaldifferent types of sensor equipments. Sensor systems canalso be deployed in the homes of older adults living alone for functionalhealth assessments.
I.INTRODUCTIONThere are many people of accidentaldeaths at home about 13 thousand every year, and about 80%, 10 thousand iselderly people. In comparison, look up traffic death that is about 5 thousandpeople,which means elderly people should pay attention to take bath and when anaccident occurs they would need some help.In such cases sensor systems can beused to assist elderly people living alone at home. They can be used foralarming or visualisation of information.
Direct automated health assessments canalso be made based on sensor data.II.METHODOLOGY1. Sensor systems for monitoring elderlypeople living alone at homeDoppler radarThe Doppler effects means phenomenon ofwave frequency is observed differentlyby the presence of relative speed between source of wave and observer3.Therefore we use the Doppler radar by observing the shift of the frequency dueto the Doppler effects to measure the moving speed of the observation targetonly rather than position.1) Bathroom Install at a wall of high position ofdressing room toward a bathroom to detect fall down or drown in a bathtub bymomentum calculation of the Doppler sensor.
2) Living roomInstall at a corner of high position ofa room covered whole spaces to detect fall down by momentum calculation of theDoppler sensor.3) X-band Doppler Sensor Module This specification covers the generalrequirements for Xband microwave Doppler module. This module is designed formotion sensing applications. It consists of DRO (Dielectric ResonatorOscillator), balanced Schottky Barrier Diode mixer and Micro- strip PatchAntennas.2.
Sensor systems for predictingfunctional health status of elderly people living alone at homeInthis approach first a large feature set is created from the sensor data.Subsequently, machine learning methods are used for the selection of the bestfeatures and the best model. For feature selection and model selection, twooff-the-shell algorithms for regression are compared: linear regressioncombined with correlation-based ranking and regression forests.
Fig. 1.Floorplan that indicate sensor placement in a house 1) SENSOR DATAEach house is equippedwith a sensor system of approximately 16 sensors mostly passive infrared motionsensors. Requirements are that the main areas in the house should be covered bypassive infrared motion sensors, making it possible to track a person throughthe house. For detecting specific activities additional sensors can beinstalled: passive infrared motion sensors for presence, a float sensor fortoilet’s flush and contact switches on doors and cabinets.
When the behaviourof the resident triggers the sensors, events are generated and stored remotelyas triples(label, timestamp, value),where label is the sensor id and value=0,1. 2)FEATUREEXTRACTIONFeatureextraction method defines location related features that are calculated formany time intervals throughout the day. For each time interval followingfeatures are calculated:· Duration of stay in an area· Number of transitions to an area· Total number of transition between areas· Total agitation in the form of total number of sensorevents 3) REGRESSIONMODELS Modelling the relation between the sensor dataand the health status is a regression problem as the features extracted fromthe sensor data as the health metrics are on a continuous scale. Featureextraction method will probably result in a redundancy in the features.
Severalfeatures will refer to the same concept. Instead of manually choosing relevantconcepts, a feature selection mechanism should be part of the solution. Becausethe complexity of problem is unknown two off-shell regression methods are chosenthat differ in expressive power. a)Linear regression b)Regression forest3)LINEAR REGRESSIONLinear regression(usingordinary least squares) is used in combination with correlation based rankingas a feature selection procedure1.The goal of OLSis to minimise the differences between the observed responses in some arbitrarydataset and the responses predicted by the linear approximation of the data.The equation forOLS is: For the linear regression model the traindata are used to rank the features based on their correlation with the healthmetric 4)REGRESSION FOREST Regression forest is chosen as a non-linear method.
The advantage isthat they generalize well and have internal feature selection, which makes themsuitable for this problem. Forest has the advantage that it generalizes betterthan a single tree, no explicit feature selection has to be done and the modelcan capture non-linear relations in the data. The regression forest is anensemble method of many trees, where each regression tree is generated from asample drawn with replacement (bootstrap sample), and for a subset of features2.A regression tree is a variant of a decision tree that is suited for(nonlinear) regression problems instead of classi?cation problems. At each nodethe data are split such that a simpler model (or weak learner) can handle thedata.
Atypical objective function that should be minimized to ?nd the optimalsplit is the mean squared error (MSE).For the regression forest before usingthe training data to train a model,2-fold cross validation is used to optimizethe parameters, namely forest size and number of features for each tree. Withthese parameters, a forest is trained and test data are used to calculate theMean Absolute Error(MAE). III.
CONCLUSIONIntegrated sensor networks realizegiving reassurance for a family and elderly people as a whole service. Thefeatures of this service are inexpensive even initial and running cost becausethinking about a way of data communication, and wide range cover in home thatseveral different types of sensor using depends on circumstances, and easy tomounting and maintenance that sensors are solar cell so no need of changingbattery and no need of wiring bywireless, and new construction work is not needed for dedicated communicationline of this service.It is possible to predict health fromdomestic sensor data, even with little assumptions on the features and with asimple feature selection and modelling scheme. Although prediction of healthmetrics is possible, changes in functional health — which are at least asvaluable to a caregiver — can be predicted with significantly better precision.This opens up opportunities to better and faster detection of problems orhealth degradation, and is likely to have a great impact on clinical practice.