Healthcare Technology in the Faculty of Engineering Professor Ian Craddock, Director SPHERE
Healthcare Technology in the Faculty of Engineering Professor Ian Craddock, Director SPHERE
Good news! 3
100 years old 70 years old 40 years old 10 years old 4
100 years old 70 years old 40 years old 10 years old 5
100 years old 70 years old 40 years old 10 years old 6
Healthcare Under Pressure: 1. Changing Demographics, e.g.: 1.4millionpeopleagedover85intheUK,by2035therewillbe3.6million. 2. Rise in Long term health conditions, e.g. Prescriptions for antidepressants have risen by 28% over three years. AquarterofadultsinEnglandareobese. Between 2006 and 2011 the number of people diagnosed with diabetes increased by 25%. 70% of the NHS budget is already consumed by long term health conditions.
Healthcare Under Pressure: We need to personalise healthcare. Weneedtokeeppeopleoutofhospitalsandclinics. To help clinicians to understand the whole patient: Integrating: - My genetic data - My metabolic data - My medical imaging data And, most difficult of all - My lifestyledata To determine the right medication, the right advice, the right diagnostic test, the right surgical procedure for me.
Science fiction?
Entirely fictional composite person #1 Williamis 55 and works in publishing. MRI of his heart, regular blood tests Personal risk profile/air quality Smartphone personalised alert, diary
Entirely fictional composite person #2 Oscaris a 63 year old executive. 2000 genome = predisposition to cardiovascular disease. Sensor = smartwatch/ecg Intervention, target Diary, exercise
Entirely fictional composite person #3 Lucasis Sensors: laptop webcam, keyboard Risk: exposure to electric light at night Intervention: rest/illumination
EPSRC Healthcare Grand Challenges 2015. 1. Data analytics and digital infrastructure for healthcare 2. Enabling technologies for regenerative medicine 3. Engineering healthy behaviours 4. Functional enhancement for safe and independent living 5. Infection prevention and control 6. Patient specific treatment 7. Prediction and early diagnosis 8. Smart surgeries and therapies 9. Systems to support and improve healthcare provision 10. Understanding and interventions in neurological function
What does that mean for Engineering? temperature, light levels, humidity, airquality base-station, +social media, +encryption analysis, pattern extraction, feature extraction data display activity monitoring via accelerometers (&others) Video Emotion, gait, activity, interaction water consumption, electrical consumption
Example project: SPHERE 12M over 5 years (plus 3M from industry and the Universities = 15M) Largest single research project in the Faculty. Led by Faculty of Engineering at the University of Bristol. In collaboration with Southampton University (Health Sciences), Southampton University (Electrical Engineering), Reading University (Cybernetics), the Elizabeth Blackwell Health Research Institute, Bristol Vision Institute, Department of Experimental Psychology, School of Social and Community Medicine, School of Oral and Dental Sciences, the Centre for Medical Ethics, the Centre for Public Engagement, School of Clinical Sciences, Communications Systems & Networks Group, Intelligent Systems Group, Bristol Heart Institute, Interaction & Graphics Group, Bristol Health Partners, ALSPAC (Children of 90s), Bristol City Council, KnowleWest Media Centre, Bristol NIHR Biomedical Research Unit in Nutrition, Diet & Lifestyle, Bristol NIHR Biomedical Research Unit for Cardiovascular Disease, IBM and Toshiba.
Environmental Sensors Sensors: PIR Water Door Contact Temperature Humidity Light Noise Air quality Electricity & gas usage (at meter and at appliance level)
Video Analytics Majid Mirmehdi, Dima Damen, Tilo Burghardt Key Challenges: 1. Tracking the movement of the subjects. 2. Identifying specific actions and activities of the subject. 3. Distinguishing the interaction of the subject with key objects in the environment. 4. Assessing the quality of specific actions. 5. Estimating the activity intensity levels. 6. Recognising mood and expressions of the subject. 7. Availability of data from real living scenarios. 20
Devices, Goals and Constraints Cameras Low-cost wide angle cameras RGB-D devices Goals Detect and track body movements Extract relevant features Analyse/Identify specific patterns relevant for clinicians Constraints Low Cost/Scalable acquisition system RT Processing Real-life environment (cluttered scene, poor illumination, nonoptimal camera position etc.)
Human Tracking Detect and track body movements Extract relevant features Analyse/Identify specific patterns relevant for clinicians
Online quality assessment of human movement from skeleton data* Injured Recovered *Paiment, Adeline and Tao, Liliand Hannuna, Sionand Camplani, Massimo and Damen, Dima and Mirmehdi, Majid (2014). Online quality assessment of human movement from skeleton data. British Machine Vision Conference (BMVC), Nottingham, UK
Measuring respiratory rate video
Ultra low power IoTs for ehealthcare Robert Piechocki Toshiba visit, 04/12/2014
Key Challenges Small size for wearability and portability Robust and reliable full-home wireless coverage Interference immunity Energy sustainability and seamless cooperation with energy harvesting Provision of RF indoor localisation information Large-scale deployment(> 100 houses)
Development: Ultra-Low Power Wearable Prototype SPW-1 SPW-1 Specifications: Dimensions 24x39mm Ultra-Low Power Design Dual Accelerometers Efficient PCB Antenna Multiple Power Options(Coin Cell, Li-Po) Energy-Harvesting Ready External Sensor Support External Antenna Support
RF Wireless Power Transfer RF Source RF Energy Receiver Advantages: Controllable remote power supply. Relatively wider range than inductive. Compact transceivers: GHz band. Flexible: multi-transmitter, multi-receiver. Issues: High attenuation by obstacles, body, etc. Unpredictable received power: receiver location, user behaviour, environment, etc. Low transmitter power: safety limitations. Low RF/DC conversion efficiency at low power levels. 28
Measured DC Power in Indoor Environment Orientation 1 Orientation 2 29
Typical User Case: Received DC Power Intermittency 30
Data fusion and data mining Peter Flach Bristol Key Challenges: 1. To recognise activities from environmental sensors.. in multiresident homes. 2. Enriched activity models with overlapping and hierarchical activity labels. 3. To achieve the best possible activity recognition from a minimal set of wearable 4. Uncertainty management and calibration. 5. Context-aware decision making. 31
Citizen Science for Future Health 32
Citizen Science for Future Health Person-centric? Supportive? Preventative? or Threatening? Invasive? 33
Citizen Science for Future Health Person-centric? Supportive? Preventative? or Threatening? Invasive? Are these absolutes? Or only a question of risk/benefit? 34
Citizen Science for Future Health Person-centric? Supportive? Preventative? or Threatening? Invasive? Are these absolutes? Or only a question of risk/benefit? Universities are the only place where we can really bring all these considerations together, informed by the risks and capability of emerging technologies. 35
Citizen Science for Future Health Ethnographic, qualitative research 36
Citizen Science for Future Health Making it real through additional investment and support. Testing technology but also understanding attitudes. A true living lab. Rolling out to 100 homes from 2016. To let 37
In Conclusion Healthcare is one of the Faculty s largest research topics. The Faculty s Healthcare Agenda is focussed on national research priorities: e.g. Internet of Things networks of battery-free, pervasive, wireless, wearable sensors Video Analytics, Data Mining. It is(hugely) interdisciplinary: Engineering + clinical and clinical science collaborators. But also a growing and widening community of allied research across all Faculties in Bristol And beyond the University 38
Thank you Ian.Craddock@bristol.ac.uk 39