of which is costly for the society in whole.
Personal health monitoring at home has also additional benefits for its user. As many basic measurements can be performed at home, less traveling to the healthcare providers is needed. For some handicapped patients or people living at remote rural locations this is a notable benefit. Travel has direct costs in form of gasoline or transportation tickets and also indirect costs in terms of travel time, delayed treatment, and lost productivity. In fact, travel has accounted for a significant proportion of the total cost in healthcare [Bre06]. However, it should also be noted that some patients may actually prefer frequent visits to the doctors because of social and related factors, and reduced travel would not be a benefit for them.
Health monitoring at home may also, in some cases, yield more reliable results. One reason for this is the so-called “white coat syndrome” [Cel04], i.e., the patient feels stressed when the doctor/nurse is performing measurements in a clinic and can obtain more reliable results when performing the measurement relaxed at home. This benefit is arguable, as many recordings can also produce false results if performed in the wrong manner without the strict guidance of a medical professional. Nonetheless, by performing the measurements at home they can be performed more frequently, generating a log of long-term follow-up data, kind of which is difficult to obtain otherwise. The user can also perform a recording at the instant he feels ill, thus giving very precise data right at the instant of an seizure or an attack, which is very helpful for the diagnosis purposes.
4.2 Measurements used for personal health
4.2 Measurements used for personal health monitoring at home 73
for electronic diaries automated data analysis and mining tools can be applied to generate automated alerts and notifications based on the data. The traditional telephone interview performed by a medical practitioner also remains an effective surveillance tool. Video-call and teleconferencing techniques enable visual interaction which enhances the possibilities of this manual surveillance [BL02]. Advances in IP-telephony and instant messaging technologies have made inexpensive video-calls available for all home computer users having wideband internet access.
4.2.1 Physiological measurements
Physiological measurements are the main source of information for personal health monitoring.
From the engineering or sensor designer viewpoint two types of physiological measurements can be identified: Distinct values and variables, and the recordings of biosignals. One can correctly argue that the second group is only raw data for which the calculation of values and parameters is yet to be done. In most cases this is true, the signal analysis being so complex that it is not possible or wise to perform in the measurement device. But it is also possible that we do not know beforehand what analysis method to use if the disease/condition to be identified is not known, or there isn’t yet an automated analysis method for the signal. In the latter case the analysis is usually done by the doctor visually, for which the graph generated from the raw data is needed. It is also useful to have raw signals in long-time monitoring when we do not know right from the start what to look for.
Distinct values and variables usually tell us immediately something about the person. The weight and blood pressure are easy examples of values that if out of predefined range give cause for immediate attention. Their long-term monitoring gives us trends which indicate the progress of person’s health. Other values and variables, like the amount of sleep per night or heart rate (or heart rate variability) are values which tell us the current state, but which may have large daily variations and no diagnosis can be given based on one measurement. A longer monitoring over a few weeks gives us enough data to make some diagnosis of the person’s health and well being.
For the recorded physiological signals a similar division exits. There are events and diseases for which we have known patterns of known changes in the signals. If we detect these known abnormalities in the recorded signal, then we can make immediate evaluation of the users health or diagnosis in case of the doctor. Also, if we have prior recorded “normal”, i.e., healthy, signal of the person, we can detect abnormal events with increased accuracy. In long-term monitoring we can observe how the state of the patient develops. If we record some less known biosignal, like the ballistocardiogram, we may not be able to evaluate the person’s health based on one recording, but in long-term monitoring we can detect abnormal changes and use them to generate alerts or evaluate the development of his condition. Table 4.1 lists the physiological parameters
Signal/Technique Source Range of parameter
Frequency range (Hz)
Primary sensor types
Ballistocardiography (BCG)
Heart (mechanical)
0-7 mg DC-40 Force, displacement
Blood pressure (BP), non-invasive
Arm (Blood vessel)
25-400 mm Hg DC-60 Cuff, auscultation
Electrocardiography (ECG)
Heart (electrical)
0.5-4 mV 0.01-250 Skin electrodes Electroencephalography
(EEG)
Brain 5-5000µV DC-150 Surface electrodes Electromyography
(EMG)
Muscle 0.1-5 mV DC-10000 Surface electrodes Eye potentials (EOG,
ERG)
Eye dipole field, eye retina
0-3500µV DC-50 Contact electrodes
Pulse oximetry (SpO2), noninvasive
Blood 30-100 % DC-2 Infrared
Respiratory rate Lungs 2-50 breaths per minute
0.1-10 Strain gage on chest, impedance, BCG
Temperature of body Body 32-40 ◦C DC-0.1 Thermistor, thermocouple Table 4.1: Physiological signals used in personal health monitoring, adapted from [JWC98, Bro95].
traditionally used in personal health monitoring applications at home.
In addition to physiological signals, other closely related signals and parameters can also be measured. Weighing scales are used to monitor the body weight, balance boards can be used to detect abnormal balance, and accelerometers and other wearable sensors [Axi05] to detect abnormal gait, agitation, motor activity, and special incidents like falling. Some sensors, like a bed mounted BCG sensor, can be used to measure biosignals (ballistocardiogram and respiratory & movement activity signals derived from it), average heart rate or respiration values, to recognize sleep/awakeness and calculate amount of daily sleeps, or used as proximity/location sensors to detect that a person is in the bed. This shows how a sensory device cannot always be classified just by its method of measurement, and also how a same sensory unit can have very different data communication needs based on how and to what purpose it is applied.
4.2 Measurements used for personal health monitoring at home 75
Sensor type Detect or measure Infra red sensor Presence, mobility
Infra red camera Surface temperature and temperature differences of objects, use of heat sources such as stove/oven
Light sensor Night/day and switching on/off of lights
Temperature sensor Ambient temperature and detection of use of shower, water flow, use of electrical appliances
Mechanical switches Alarm buttons, mobility measurements by detecting door openings and closings, identify use of major appliances as the refrigerator Central power sensor Daily power consumption, detect daily routines
Appliance power sensor Power consumption, measured from the power cord directly or in-directly, measure use of electrical appliance
Pressure sensor Presence, mobility, ballistocardiogram
Sound sensor Presence, activities, based on automatic recognition Capacitive sensors Presence, mobility
Microwave radar Presence, movement
Table 4.2: Sensors for ambient measurements used for personal health monitoring, adapted from [Cel95][P8].
4.2.2 Ambient measurements
Non-physiological signals and values can be monitored to support personal health monitoring and health status assessments. Changes in the health status change the patterns of daily living and use of household resources [Cel95, Cel96, Kor03, Bar05]. These “activities of daily living”
(ADL) signals give us indirect information about the users living habits and activities, functional health status, and possible functional deterioration [Cel95]. Table 4.2 presents sensors useful in ambient monitoring.
In addition, modern computerized and networked devices may provide state/usage infor-mation which can be used in activity monitoring. Home security & autoinfor-mation systems also provide ambient activity data useful in healthcare monitoring, and interaction with these systems is beneficial.
Daily activity information can be enhanced by using personal wearable or portable devices, which may or may not also provide physiological and related measurements. Accelerometers are used to implement pedometers to measure amount of daily movement. Personal mobile phones and GPS sensors can be used to track the person and to monitor the daily activity patters outside the home. [Axi05] [P8]