11:20 - 11:40 amSunday, September 29
LK 120
A cloud-based system for predicting and preventing falls in older people
LK 120
A cloud-based system for predicting and preventing falls in older people
Austrian Institute of Technology
Background Falls in older people are common. More than 30% of the people older than 65, and more than 50% in the age group above 80 fall at least once a year [1]. In order to develop an individualized... Read more

Description

Background

Falls in older people are common. More than 30% of the people older than 65, and more than 50% in the age group above 80 fall at least once a year [1]. In order to develop an individualized fall prevention strategy for older people living in the community, the first step is to assess their fall risk to determine the starting points of the intervention. Until now, assessing fall risk is done in a clinical environment, supervised by a physician. Only people with an active fall history will be screened for fall risk [2]. Since health care resources are limited, but the number of older people is increasing, this work focuses on the feasibility of a self-assessment fall prediction, prevention and feedback approach to empower older people living at home.

Methods

In the project iStoppFalls [3] a Microsoft KINECT depth camera and a wearable sensor is used to assess fall risk in the living room. Patients are encouraged to answer a questionnaire and to perform unsupervised physical exercises in front of the system. Data are transmitted to an internet-based fall prediction service, which automatically reasons new data, uses machine learning algorithms and historical data to predict future fallers. Data from a previous prospective fall risk factor study in Australia with 449 participants were analysed to discover risk factors with a high predictive power.People can use the iStoppFalls system to perform exercises to reduce their risk of falling. Individual health advices and feedback are provided to them. Patients have been engaged during the whole design and development cycle. In a first phase end-user needs were assessed with empirical interviews and workshops. The system functionality was tested in a first pilot study with 30 older people in Germany. A second study focusing on the reliability and validity of the risk assessment will start in July. The baseline assessment of this study will be conducted in the lab and fall-related assessments [4] will be performed to compare the iStoppFalls prediction results with. In the final phase starting in October 2013 the system will be tested in a 6-months randomized controlled trial in Germany, Finland, Spain and Australia, involving more than 200 people.

Results

The number of previous falls in the past year, total number of medications, painful feet, a lower reaction time, age of a person and a higher fear of falling have been shown as high predictors for falls. A preliminary prediction model based on historical data can discriminate between high-risk fallers and non-fallers with an area under the receiver operating characteristic curve of 0.75. The first results from the iStoppFalls risk assessment yield to a good discrimination between fallers and nonfallers compared to the lab tests.

Conclusions

A machine learning approach is suitable to detect high-risk fallers. Paired with a technology supported self-assessment and an internet-based fall prediction service a wide range of people living in the community can be screened. High-risk fallers can be detected automatically for further medical interventions. The users will get individual feedback and suggestions to engage them in their own healthcare.

Andreas Ejupi, PhD candidate was born in Vienna, Austria. He holds a Bachelor degree (BSc) in Business Informatics, a MBA and a Master degree (MSc) in Medical Informatics from the Vienna University of Technology. Currently he is working on his PhD thesis at the Austrian Institute of Technology. He is interested in initiatives and research related to the future of medicine and healthcare. He focusses on assistive technologies, especially on solutions for older people and people with (chronic) diseases. In particular, he is working on fall detection and prevention solutions. He develops supportive devices for people with Parkinson’s disease. Recently he received awards at the Health and Wellness innovation competition at MIT and got awarded at the Ambient Assisted Living – Forum in Europe.

Contact Us

We're not around right now. But you can send us an email and we'll get back to you, asap.

Not readable? Change text. captcha txt

Start typing and press Enter to search