Dec 3rd 2007 - Getting Started - Prof. Robert John
I'm very pleased to announce the commencement of this exciting new collaborative project between De Montfort University and the University of Missouri.
The developed world is expected to have an elderly population of roughly two billion people by 2050 which will have a huge impact on healthcare resources.
As a result of this we are likely to see technology playing a very significant part in supporting and monitoring people in their homes or in assisted living facilities, allowing individuals to lead more active lifestyles with a reduced dependence on social care.
We think that fuzzy logic will play an important role in ensuring this technology is as effective as possible.
This project will investigate how advanced models of uncertainty, developed at DMU, can help with important work being carried out by ElderTech.
CCI & ElderTech
Uncertainty Modelling in Technologies for the Elderly
Project Lab Book
Dec 7th 2007 - Visa and Planning - Dr. Simon Coupland
I have now received my visa to allow to undertake this research in the US. I have also completed my travel arrangement as will begin work on the project on January 7th 2008. We also have a new type-2 fuzzy logic development IDE thanks to a recently completed project. This will be an invaluable tool whilst developing systems on this project.
I have now received my visa to allow to undertake this research in the US. I have also completed my travel arrangement as will begin work on the project on January 7th 2008. We also have a new type-2 fuzzy logic development IDE thanks to a recently completed project. This will be an invaluable tool whilst developing systems on this project.
Dec 18th 2007 - Press Release - Dr. Simon Coupland
From the DMU Press Office:
From the DMU Press Office:
RESEARCH TO EXPLORE HOW COMPUTATIONAL INTELLIGENCE CAN BE USED IN THE CARE OF OLDER PEOPLE
Intelligent computers could help to raise the alarm when older people suffer medical emergencies such as serious falls.
Researchers at De Montfort University Leicester (DMU) are exploring the role that computational intelligence can play in the care of older people.
A joint research team from DMU and the University of Missouri in the US will look at how fuzzy logic, where computing is used to replicate the human decision making process, can be used in technology designed to assist the elderly.
Experts from DMU’s Centre for Computational Intelligence (CCI) are taking part in the six-month project with researchers from the University of Missouri’s Center for Eldercare and Rehabilitation Technology (CERT).
The £45,000 project will build on research currently being carried out at CERT, including a study monitoring older adults using sensors. Networks ranging from simple motion sensors to bed sensors which capture sleep restlessness and pulse and respiration levels have been installed in several apartments at an assisted living facility in Missouri.
The technology can be used for many health and social care purposes, including detecting when medical emergencies, such as serious falls, have happened and helping to diagnosis the early signs of serious health problems.
The joint project will look at how fuzzy logic can improve the effectiveness and reliability of the existing monitoring equipment and new technology which could be used in the future.
DMU’s Dr Simon Coupland will spend four months with the CERT team in Missouri to undertake the research.
Dr Coupland said: “There are many factors that can cause problems for assistive technologies in nursing and social care, from the noise of sensors on some equipment, to variations in how well technology performs in different environments, and of course, the uncertainty involved with human behaviour. This project will look at how computational intelligence, and in particular fuzzy logic, can be used to address these kinds of problems.”
He added: “One example of how this technology might be used is in sensors that detect the sound of somebody falling over. Fuzzy logic could be used to distinguish between the sound of a person falling to the floor and other noises, such as a door being slammed.
“Any sounds suspected to be that of a fall would trigger an alarm summoning medical assistance. In this way, the technology would help to ensure that genuine emergencies are never ignored while reducing the number of false alarms.”
Fuzzy logic is a branch of mathematics which can be used to help computers reach a conclusion based on vague information. It mimics the human approach to problem solving but arrives at a decision much more quickly than people do, and solves problems that cannot be addressed using conventional mathematics.
Professor Robert John, Director of the CCI, is also taking part in the project. He is a leading expert in the field of fuzzy logic with an extensive background in research in the area.
Professor John said: “The developed world is expected to have an elderly population of roughly two billion people by 2050 which will have a huge impact on healthcare resources.
“As a result of this we are likely to see technology playing a very significant part in supporting and monitoring people in their homes or in assisted living facilities, allowing individuals to lead more active lifestyles with a reduced dependence on social care. We think that fuzzy logic will play an important role in ensuring this technology is as effective as possible.”
Professor James Keller is the main researcher on the project from the University of Missouri and CERT.
He holds the University of Missouri’s Curators’ Professorship in the Electrical and Computer Engineering and Computer Science Departments, is the university’s R.L. Tatum Professor in the College of Engineering and is also a Principal Investigator at CERT.
Professor John added: “We are very excited to be working on this project with Professor Keller, who is a world-renowned expert in the field of fuzzy logic. We hope this research will lead to future collaborations with the Center for Eldercare and Rehabilitation Technology and the University of Missouri.”
The funding has been provided by the Engineering and Physical Sciences Research Council.
Dec 19th 2007 - Press Coverage of the Project - Dr. Simon Coupland
This afternoon I was interviewed by BBC Radio Leicester. A short news item about the project will be broadcast on New Years Day.
Other coverage:
British Journal of Healthcare Information Management
Higher Education Research Opportunities
This afternoon I was interviewed by BBC Radio Leicester. A short news item about the project will be broadcast on New Years Day.
Other coverage:
British Journal of Healthcare Information Management
Higher Education Research Opportunities
Dec 20th 2007 - More Press Coverage of the Project - Dr. Simon Coupland
This afternoon I was interviewed by Tim Fergerson from Silicon.com. Click here to read the piece.
This afternoon I was interviewed by Tim Fergerson from Silicon.com. Click here to read the piece.
Jan 6th 2008 - Missouri Leg of the Project Begins - Dr. Simon Coupland
Last night I arrived in the University town of Columbia, Missouri, which is located around 120 miles east of Kansas City. I met with Dr Popescu last night and will be meeting up with Prof. Keller later today.
Last night I arrived in the University town of Columbia, Missouri, which is located around 120 miles east of Kansas City. I met with Dr Popescu last night and will be meeting up with Prof. Keller later today.
Jan 7th 2008 - First Day at the Lab - Dr. Simon Coupland
Today I met other members of the ElderTech team, including the project leader Marjorie Skubic and a number of graduate students. We discussed the best area to begin working on. We have agreed to begin by looking at reinterpreting some earlier work on voxel processing of silhouette data, using some of the advanced uncertainty modelling techniques provided by the type-2 fuzzy paradigm. I will be chairing a seminar on type-2 to get the whole team up to speed with the cutting edge of the field, date TBA.
Today I met other members of the ElderTech team, including the project leader Marjorie Skubic and a number of graduate students. We discussed the best area to begin working on. We have agreed to begin by looking at reinterpreting some earlier work on voxel processing of silhouette data, using some of the advanced uncertainty modelling techniques provided by the type-2 fuzzy paradigm. I will be chairing a seminar on type-2 to get the whole team up to speed with the cutting edge of the field, date TBA.
Jan 9th 2008 - Visit to TigerPlace - Dr. Simon Coupland
Today I went to the TigerPlace for the first time and met more of the ElderTech team, including a number of staff from nursing and social care. It’s a diverse team taking a multi-angled approach to the task of utilising technology for improving the lives of the elderly.
Today I went to the TigerPlace for the first time and met more of the ElderTech team, including a number of staff from nursing and social care. It’s a diverse team taking a multi-angled approach to the task of utilising technology for improving the lives of the elderly.
Jan 10th 2008 - Ethics Compliance Training - Dr. Simon Coupland
Today I completed my IRB (Institutional Review Board) and HIPAA (Health Insurance Portability and Accountability Act) compliance training. This is concerned with best ethical practice when conducting research and is understandably of huge importance on this project where we are dealing some of the most vulnerable people in society.
Today I completed my IRB (Institutional Review Board) and HIPAA (Health Insurance Portability and Accountability Act) compliance training. This is concerned with best ethical practice when conducting research and is understandably of huge importance on this project where we are dealing some of the most vulnerable people in society.
Jan 14th 2008 - First Meeting - Dr. Simon Coupland
The first team meeting took place today. Present were Prof. Keller, Prof. John (via a web link), Dr. Coupland, Dr. Popescu and two MIZZOU students Derek Anderson and Robert Luke. The main item up for discussion was how to move forward with the voxel based fall detection system built by Derek and Robert. More data will shortly become available from a stunt actor who has been trained by nurses to replicate the way the elderly fall. In the short term I will continue exploring the current system and find ways of enhancing the uncertainty models used. Longer term we need to think about learning the sets/rules in the fuzzy system and also look at an audio detection system.
The first team meeting took place today. Present were Prof. Keller, Prof. John (via a web link), Dr. Coupland, Dr. Popescu and two MIZZOU students Derek Anderson and Robert Luke. The main item up for discussion was how to move forward with the voxel based fall detection system built by Derek and Robert. More data will shortly become available from a stunt actor who has been trained by nurses to replicate the way the elderly fall. In the short term I will continue exploring the current system and find ways of enhancing the uncertainty models used. Longer term we need to think about learning the sets/rules in the fuzzy system and also look at an audio detection system.
Jan 16th 2008 - Press Article - Prof. Robert John
Electronics Weekly have published an article on the project, click here to read the article.
Electronics Weekly have published an article on the project, click here to read the article.
Jan 18th 2008 - Faculty Retreat - Dr. Simon Coupland
The Center for Eldercare and Rehabilitation Technology’s faculty retreat took place today. Most members of the group attended and a wide range of topics were discussed, including current project status, upcoming areas of work (including the CCI input), future publishing and grant opportunities. Additionally a talk was given by Marc Hamilton about inactivity, exercise and the effect of prolonged sitting.
The Center for Eldercare and Rehabilitation Technology’s faculty retreat took place today. Most members of the group attended and a wide range of topics were discussed, including current project status, upcoming areas of work (including the CCI input), future publishing and grant opportunities. Additionally a talk was given by Marc Hamilton about inactivity, exercise and the effect of prolonged sitting.
Jan 21th 2008 - Press Article - Prof. Robert John
Computer Weekly have published an article on the project, click here to read the article.
Computer Weekly have published an article on the project, click here to read the article.
Jan 23rd 2008 - On Data - Dr. Simon Coupland
I attended the weekly meeting at TigerPlace this morning. A presentation was given by Steve Miller, it was concerned with the collected data and patient events, such as hospitalisations and falls. The graphics and numbers provided gave a real context to the difficulty of the problems being tackled by the Missouri team. The data is seemingly full of noise and trends are not easily seen. It is becoming increasingly clear that handling data correctly and understanding what the data means is critically important.
I attended the weekly meeting at TigerPlace this morning. A presentation was given by Steve Miller, it was concerned with the collected data and patient events, such as hospitalisations and falls. The graphics and numbers provided gave a real context to the difficulty of the problems being tackled by the Missouri team. The data is seemingly full of noise and trends are not easily seen. It is becoming increasingly clear that handling data correctly and understanding what the data means is critically important.
Jan 28th 2008 - EPSRC - Prof. Robert John
The Engineering and Physical Sciences Research Council has featured this project on their website: http://www.epsrc.ac.uk/InTheNews/FuzzyLogic.htm
The Engineering and Physical Sciences Research Council has featured this project on their website: http://www.epsrc.ac.uk/InTheNews/FuzzyLogic.htm
Feb 5th 2008 - GPGPU: General-Purpose Computation Using Graphics Hardware - Dr. Simon Coupland
An interesting side-project has emerged while working on the type-2 fuzzy system for fall detection. A number of people in the ElderTech group have expertise in GPGPU, a methodology for creating highly parallelised systems using high-end graphics hardware. Until very recently deploying general computing applications on graphics hardware required detailed knowledge of the hardware and of a shading language (such as OpenGL SL) that is implemented on that hardware. Recently the graphics hardware vendor NVIDIA released the C-like programming library for it's cards called CUDA. CUDA allows graphics hardware to be programmed in a C-like syntax without knowledge of a shading language. A knowledge and appreciation of the characteristics of the hardware is essential, as well as a degree patience and an empirical approach to writing the software.
Together with MIZZOU grad student, Derek Anderson, I have been working on a CUDA implementation of a type-1 Mamdani Fuzzy system, with a view to extending this to interval type-2 and potentially generalised type-2 as well. The massively parallel nature of graphics hardware is ideal for fuzzy systems, this promises to be an interesting a-side to the main thrust of the project.
An interesting side-project has emerged while working on the type-2 fuzzy system for fall detection. A number of people in the ElderTech group have expertise in GPGPU, a methodology for creating highly parallelised systems using high-end graphics hardware. Until very recently deploying general computing applications on graphics hardware required detailed knowledge of the hardware and of a shading language (such as OpenGL SL) that is implemented on that hardware. Recently the graphics hardware vendor NVIDIA released the C-like programming library for it's cards called CUDA. CUDA allows graphics hardware to be programmed in a C-like syntax without knowledge of a shading language. A knowledge and appreciation of the characteristics of the hardware is essential, as well as a degree patience and an empirical approach to writing the software.
Together with MIZZOU grad student, Derek Anderson, I have been working on a CUDA implementation of a type-1 Mamdani Fuzzy system, with a view to extending this to interval type-2 and potentially generalised type-2 as well. The massively parallel nature of graphics hardware is ideal for fuzzy systems, this promises to be an interesting a-side to the main thrust of the project.
Feb 8th 2008 - Data Collection - Dr. Simon Coupland
I took part in latest data capture exercise for the fall detection system. Yesterday, a stunt actor, trained by the nursing staff to replicate the physical nature of an elderly person falling, was videoed in a variety of scenarios. To complement this positive fall data a number of false alarm scenarios were filmed this morning. The video below shows me acting out one such scenario where a person bends down and gets back up again. This data will be processed into silhouettes before features are extracted and processed to test the fall detection system.
I took part in latest data capture exercise for the fall detection system. Yesterday, a stunt actor, trained by the nursing staff to replicate the physical nature of an elderly person falling, was videoed in a variety of scenarios. To complement this positive fall data a number of false alarm scenarios were filmed this morning. The video below shows me acting out one such scenario where a person bends down and gets back up again. This data will be processed into silhouettes before features are extracted and processed to test the fall detection system.
Feb 12th 2008 - CIS Talk - Dr. Simon Coupland
Today I gave a talk the University of Missouri IEEE CIS chapter. The seminar was well received by a group of around 40 academics. The talk covered all the basic definitions and operations of interval and generalised type-2 fuzzy logic.
Today I gave a talk the University of Missouri IEEE CIS chapter. The seminar was well received by a group of around 40 academics. The talk covered all the basic definitions and operations of interval and generalised type-2 fuzzy logic.
Feb 15th 2008 - Dealing with Data - Dr. Simon Coupland
I have run around 15 minutes of footage through the type-2 fuzzy fall detection system that I have been working on. The system seems to be working quite well. However, with this amount of data made up of hundreds of frames, visualising and assessing system performance becomes a difficult task in itself. I have been working on an OpenGL program that outputs the results from the fuzzy system alongside the camera images, with pause, fast-forward and rewind functionality. This tool is proving to be difficult to build, but will be very useful in assessing system performance.
I have run around 15 minutes of footage through the type-2 fuzzy fall detection system that I have been working on. The system seems to be working quite well. However, with this amount of data made up of hundreds of frames, visualising and assessing system performance becomes a difficult task in itself. I have been working on an OpenGL program that outputs the results from the fuzzy system alongside the camera images, with pause, fast-forward and rewind functionality. This tool is proving to be difficult to build, but will be very useful in assessing system performance.
Feb 21th 2008 - Arriving in Missouri - Prof. Robert John
Am on my visit to the University of Missouri Columbia on our EPSRC project (Uncertainty Modelling in Technologies for the Elderly) with Jim Keller. An eventful trip (!) meant it took 2 days to get there but I was made very welcome on arrival. Our project is to do some work using type-2 fuzzy logic on fall prediction for the elderly in their home. The project is a fascinating one that is interesting from a fuzzy logic perspective but also a project that could have a real impact on improving the lives of the elderly. The first working day of the visit I gave a presentation on some nursing research I have done in the past. This was to the Eldertech team at TigerPlace where various types of sensors will be implemented in situ. This will eventually lead to large amounts of data for analysis to monitor the residents behaviour. This is an impressive project that is difficult from a practical point of view but also from a 'human' aspect in that putting sensors in peoples accommodation has to be handled with care and the role of the nurses is central.
After the visit to TigerPlace we have had various discussions about how to proceed with our work. Simon arrived in January and has used that time to explore a new algorithm for finding the centroid of his geometric type-2 systems, implemented an interval type-2 fuzzy logic system and explored the existing type-1 fuzzy systems on some test data. We have a number of ideas to take the type-2 work forward including investigating hierarchical type-2 fuzzy systems. We need to firm these ideas up but they are exciting and new and, most importantly, will help tackle a real problem.
Am on my visit to the University of Missouri Columbia on our EPSRC project (Uncertainty Modelling in Technologies for the Elderly) with Jim Keller. An eventful trip (!) meant it took 2 days to get there but I was made very welcome on arrival. Our project is to do some work using type-2 fuzzy logic on fall prediction for the elderly in their home. The project is a fascinating one that is interesting from a fuzzy logic perspective but also a project that could have a real impact on improving the lives of the elderly. The first working day of the visit I gave a presentation on some nursing research I have done in the past. This was to the Eldertech team at TigerPlace where various types of sensors will be implemented in situ. This will eventually lead to large amounts of data for analysis to monitor the residents behaviour. This is an impressive project that is difficult from a practical point of view but also from a 'human' aspect in that putting sensors in peoples accommodation has to be handled with care and the role of the nurses is central.
After the visit to TigerPlace we have had various discussions about how to proceed with our work. Simon arrived in January and has used that time to explore a new algorithm for finding the centroid of his geometric type-2 systems, implemented an interval type-2 fuzzy logic system and explored the existing type-1 fuzzy systems on some test data. We have a number of ideas to take the type-2 work forward including investigating hierarchical type-2 fuzzy systems. We need to firm these ideas up but they are exciting and new and, most importantly, will help tackle a real problem.
Feb 27th 2008 - Leaving Missouri - Prof. Robert John
The week I spent with the Eldertech team is now over. This proved be both an interesting and enjoyable trip. The team are doing some really innovative things with sensor information and it seems type-2 fuzzy logic should help. We had a long brain storming meeting about our side of things where Simon demo'd his first working interval type-2 fuzzy system. This has led to some interesting ideas relating to 'blurring' a type-1 system and how we cope with hierarchical type-2 fuzzy systems. So, a very enjoyable trip with some great meals and beer! Looking forward to Jim and Mihail's visit in May.
The week I spent with the Eldertech team is now over. This proved be both an interesting and enjoyable trip. The team are doing some really innovative things with sensor information and it seems type-2 fuzzy logic should help. We had a long brain storming meeting about our side of things where Simon demo'd his first working interval type-2 fuzzy system. This has led to some interesting ideas relating to 'blurring' a type-1 system and how we cope with hierarchical type-2 fuzzy systems. So, a very enjoyable trip with some great meals and beer! Looking forward to Jim and Mihail's visit in May.
March 14th 2008 - First Prototype- Dr. Simon Coupland
The first prototype of the type-2 fuzzy logic system for fall detection is now operational.
The video above shows the inputs to the system, video images which are translated in to silhouettes before being used to form voxel person, the 3D reconstruction you can see in the bottom right of the video.
A set of features are extracted from voxel person and fed in to the type-2 fuzzy system.
A number of rules fire, an output is arrived at, the green region in the video represents the output for standing.
The black line is the equivalent output from the existing type-1 system.
The next stage of the project involves tuning the type-2 system so that the area of the green region increases and decreases with the uncertainty propagated through the system.
April 1st 2008 - Last Day in Columbia, Missouri - Dr. Simon Coupland
Today marks the completion of the first phase of this project at the University of Missouri Columbia. The next time the team meets it will be in at De Montfort University in Leicester in May. The current status of the work is as follows:
Today marks the completion of the first phase of this project at the University of Missouri Columbia. The next time the team meets it will be in at De Montfort University in Leicester in May. The current status of the work is as follows:
- Type-2 Fuzzy System for Fall Detection. The fuzzy system has been implemented and can be run on data extracted from sequences of video, some system tuning is still required. Data from recent video sequences is still being processed. Once this is complete we can compare the performance of the two systems. The problem of fall identification is a relatively crisp one, we expect any performance improvements to be in the reduction of the false alarm rate.
- Audio System for Fall Detection. During this phase of the project data has been gathered for identifying falls from audio signals. Features, sets and rules still need to be identified.
- Real-Time Implementation. One of the technical difficulties in this project is the processing of large amounts of data in real-time. The enabling technology to allow real-time processing is Graphical Processing Units (GPUs). During the 3 months I have been here we have made significant progress in the implementation of image processing and fuzzy logic applications on NVidia GPUs. One paper has been submitted on this work and a number of others are currently in the pipeline.