Australasian Science: Australia's authority on science since 1938

Catch ‘Em Before They Fall

Elderly man falling

Falls-risk assessment is currently a very active research area that is spurred on by the ageing populations in developed nations.

By Stephen Redmond

Biomedical engineers are developing a sensor that can predict when elderly people are going to fall.

One in three Australians aged over 65 years fall each year, and the injuries they suffer cost the Australian healthcare system more than $500 million (and rising) per annum. This ratio increases to one in every two people aged over 80 years, and in this age group injuries due to falls are seven times more numerous than all other causes of injury combined.

However, there is something quite impersonal about these cold hard facts. What is much harder to quantify is the personal cost of falls.

A lady aged in her mid-90s once told me of the ordeal she suffered after she fell down the stairs at home, lost consciousness and injured her hip. When she woke she could not stand up. She crawled to the next room to reach the telephone, and called for help. She now uses a walking frame wherever she goes.

Her story is all too common. Could something possibly have been done to help this woman if we had known that she was at risk of falling before she fell?

The answer to that question is “yes”. If we had known she was at risk of falling at an earlier stage, depending on her risk factors, we could have enrolled her in a muscle strengthening program, addressed any vision problems she might have had, or simply provided her with a walking aid. But how can we estimate her risk of falling in the future?

Falls-risk assessment is currently a very active research area that is spurred on by the ageing populations in developed nations. Many clinical falls-risk tests are based on questionnaires that ask the person to recall how many falls they have had in the past few months or years, or observe their gait and balance as they perform normal activities. Often these methods require bulky, specialised equipment and the presence of a trained assessor, which limits their ability to monitor the general population.

But wireless systems now pervade our lives, and motion-sensing technologies have become very cheap, very small, and are available for purchase online. This combination of reliable wireless information transmission and low-cost sensing electronics has allowed us to build small, wearable motion-sensing devices that make it feasible to bring objective falls-risk assessment into the home.

We hope to condense all the equipment, subjective assessment and personnel associated with how we currently predict falls, into a single wearable sensor that automates the entire process. Microelectromechanical sensors can measure acceleration, rate of rotation and barometric air pressure. The sensors and a wireless transmitter are housed in a single small case that is worn on the hip, which enables the reliable characterisation of human movement.

In order to assess how likely someone is to fall, our group at the University of NSW has developed a short series of physical assessment tests. These tests are performed unsupervised in the home and can be characterised using our wearable motion-sensing device, which is clipped onto a belt around the subject’s waist.

These tests consist of three physical assessments:

• rise from a chair, walk 3 metres and return to the chair;

• place the left foot on a small step, then back to the floor, then the same with the right foot, and repeat four times; and

• stand up and sit down five times.

The resulting movement signals are analysed to extract information relating to speed, strength and balance. All this information is then reduced to a single falls-risk score using some clever mathematics.

We have evaluated our tests on 68 elderly subjects, and found that the falls-risk scores generated by our system are almost identical to the same scores generated by another popular clinical falls-risk test.

To speculate as to how such falls-risk monitoring technologies might find their way into the home environment, we need to examine the current state of fall detection devices, which recognise the moment a person actually falls by the large impact acceleration generated. These technologies are beginning to reach maturity, although many generate a large number of false alarms. The marketplace for such technologies is competitive and already crowded, with several monitoring devices available.

It is my belief that fall detection devices, in whatever final form they assume, will become as commonplace in the homes of elderly people as televisions and refrigerators. Therefore, it is a small leap of imagination to predict that fall detection technologies in the home will eventually include some type of falls prevention strategy, similar to the method being proposed by our research group.

While the use of wearable sensors to objectively estimate one’s risk of falling in the future would be a very positive step forward, one of the stumbling blocks facing the use of wearable sensors is user compliance. There are two issues here.

The first issue is that if we require the subject to undertake contrived physical assessments at home, they may simply refuse to do it. This is a problem for which we are currently working towards a solution. We hope to be able to estimate someone’s risk of falling by observing how they move as they perform normal daily activities, such as walking or standing up from a chair, but this would require them to wear the device for most of the day. The concern here is that the subject will forget or refuse to wear the device continuously – a particular problem for those suffering from Alzheimer’s disease, who are prone to falls.

A similar problem faces the successful implementation of wearable falls detection systems, especially during night-time visits to the bathroom, when someone is unlikely to put on a wearable sensor before leaving bed. Many research groups around the world, including our own, are working towards various solutions to this problem. Efforts vary from embedding intelligent monitoring systems within the home, to the use of tiny motion sensors implanted in the person’s body. This latter proposal is not as far-fetched as it might initially sound – strain gauge sensors are already used in some artificial hip replacements to detect loosening of the prosthesis.

Underpinning all these exciting developments is the internet. The ability to monitor someone’s health from a great distance while they remain in their own home, act on this information and advise them on how to best maintain their wellbeing, will open up a world of possibilities for healthcare. This is especially relevant for elderly people, who often have a strong desire to continue living independently at home for as long as possible.

The transition will be painful for the healthcare system, but change is undoubtedly coming. As the Australian government prepares to contribute $27.5 billion towards a total estimated cost of $36 billion to install fibre optic cable carrying broadband speeds of approximately 100 Mb/s to every home in the country, the migration of primary healthcare to the home is proving to be one of the critical issues driving the debate. Irrespective of the successes of Australia’s National Broadband Network, the internet will continue to permeate into our everyday lives, and it is only a matter of time before healthcare systems change the way they do business and begin to make full use of this incredible communication resource.

Over the next year we plan to trial our falls-risk assessment technology on a larger scale and in a completely unsupervised setting. If these trials prove successful we hope to commercialise our device and make it available to the Australian public. In the coming years, it is hoped that this form of monitoring technology will significantly improve the quality of life of an ageing population by reducing their incidence of falls.

Stephen Redmond is a lecturer in the University of NSW Graduate School of Biomedical Engineering.