ARISES: An Adaptive, Real-time, Intelligent System to Enhance Self-care of chronic disease
Lead Research Organisation:
Imperial College London
Department Name: Electrical and Electronic Engineering
Abstract
This research brings together a multidisciplinary collaborative team of engineers, clinicians and patients to deliver a user driven, patient centric, bespoke technology to treat chronic health-conditions. The proposal will develop an Adaptive, Real-time, Intelligent System (ARISES) that will run on a smart phone locally and collect data from multiple sources to deliver an intervention to the patient that allows self-management of chronic disease. The core of ARISES will use Case-based-reasoning (CBR), a consolidated artificial intelligence technique which can solve problems in much the same way as a human does, using historical data and scenarios as a reference to recommend a current solution which can treat the patient. CBR is also powerful in that it has the capability to be adaptive according to patient lifestyle and behavior and always provide the most optimum solutions for a given set of resources. ARISES will have the capability to collect data from wearable devices such as smart-watches, activity monitors, hear rate monitors and continuous glucose meters and using smart-algorithms will be able to extract meaningful information to provide to the CBR system. Underpinning this will be energy efficient algorithms which always make ARISES aware of what sensors are connected to the patients local area network, safety systems that minimise the risk of any possible undesired event related to the management of the disease, and a data security to make sure information is protected against non-authorised access. ARISES will provide a generic framework which can be used to treat many chronic diseases such as asthma, chronic obstructive airways disease, hypertension, heart failure, ischaemic heart disease, arrhythmias and chronic neurological conditions. Given the global incidence, as an exemplar chronic condition to demonstrate its use we have chosen diabetes which currently affects 3% of the world's population, and we will target improvement in glycaemic control which can reduce micro- and macrovascular complications associated with the disease. In this context the system will promote the self-management of diabetes by optimizing glucose control through insulin dosage recommendations, exercise and physical activity support, carbohydrate recommendations to prevent hypoglycaemia, and behavioral change through education.
Planned Impact
There are three main groups of beneficiaries from this project:
1. Academic/Scientific (researchers working on this project, PhD/MSc students working on related projects, undergraduate students studying Electronic Engineering, Bioengineering, computing and similar courses)
2. Society/Healthcare (individuals who suffer with a chronic condition which will be eventually using the systems and products whose development was initiated in this project, as well as their carers and hospitals)
3. Economic/Commercial (companies and services which offer healthcare products for disease management)
We also anticipate the specific impact of this work will be realised in the following areas:
1. Knowledge- This proposal will generate a novel framework, combining CBR and real-time data fusion, information driven algorithms and safety to create new advances in each of these respective fields we have never before been brought together on a real-time portable system.
2. Society- The proposed platform will offer a novel solution for chronic disease management outside of the hospital, reducing potential ill-health complications and progressive degeneration of conditions. It will improve health and quality of life, reducing the burden on healthcare services.
3. People- Academic and research staff working on this project will develop new skills, and become experienced in working across disciplines and development and translation of healthcare technology. There will also be knowledge dissemination impacting PhD/MSc students working on related projects, undergraduate students studying Electronic Engineering, Bioengineering, computing and similar courses. People suffering from chronic disease will also have improved quality of life, impacting themselves and their careers, and provide the capability to self-manage their condition without the need to frequently go to hospital.
4. Economy - The proposed solution has the potential to reduce hospital visits and make chronic care of disease more efficient through self-management. This will eventually save money and resource to the NHS. Additionally through translation, new products and companies may be formed which adopt this methodology contributing to wealth creation within the UK.
1. Academic/Scientific (researchers working on this project, PhD/MSc students working on related projects, undergraduate students studying Electronic Engineering, Bioengineering, computing and similar courses)
2. Society/Healthcare (individuals who suffer with a chronic condition which will be eventually using the systems and products whose development was initiated in this project, as well as their carers and hospitals)
3. Economic/Commercial (companies and services which offer healthcare products for disease management)
We also anticipate the specific impact of this work will be realised in the following areas:
1. Knowledge- This proposal will generate a novel framework, combining CBR and real-time data fusion, information driven algorithms and safety to create new advances in each of these respective fields we have never before been brought together on a real-time portable system.
2. Society- The proposed platform will offer a novel solution for chronic disease management outside of the hospital, reducing potential ill-health complications and progressive degeneration of conditions. It will improve health and quality of life, reducing the burden on healthcare services.
3. People- Academic and research staff working on this project will develop new skills, and become experienced in working across disciplines and development and translation of healthcare technology. There will also be knowledge dissemination impacting PhD/MSc students working on related projects, undergraduate students studying Electronic Engineering, Bioengineering, computing and similar courses. People suffering from chronic disease will also have improved quality of life, impacting themselves and their careers, and provide the capability to self-manage their condition without the need to frequently go to hospital.
4. Economy - The proposed solution has the potential to reduce hospital visits and make chronic care of disease more efficient through self-management. This will eventually save money and resource to the NHS. Additionally through translation, new products and companies may be formed which adopt this methodology contributing to wealth creation within the UK.
Organisations
Publications
Cappon G
(2019)
Classification of Postprandial Glycemic Status with Application to Insulin Dosing in Type 1 Diabetes-An In Silico Proof-of-Concept.
in Sensors (Basel, Switzerland)
Daniels J
(2022)
A Multitask Learning Approach to Personalized Blood Glucose Prediction.
in IEEE journal of biomedical and health informatics
Guemes A
(2020)
Predicting Quality of Overnight Glycaemic Control in Type 1 Diabetes Using Binary Classifiers.
in IEEE journal of biomedical and health informatics
Herrero P
(2018)
Automatic Adaptation of Basal Insulin Using Sensor-Augmented Pump Therapy.
in Journal of diabetes science and technology
Herrero P
(2017)
A coordinated control strategy for insulin and glucagon delivery in type 1 diabetes.
in Computer methods in biomechanics and biomedical engineering
J Chen
(2018)
Dilated recurrent neural network for short-time prediction of glucose concentration
in CEUR Workshop Proceedings
| Description | From the project there where the following notable achievements - 1. Machine learning algorithms using deep learning have been developed and validated with in-silico and retrospective clinical data to be able to reliably predict blood glucose with a 30 and 60 minute prediction horizon. These have been demonstrated in both software and hardware. Using these we have made a smartphone based decision support system, the ARISES diabetes management system, to optimise glucose management in subjects with diabetes. 2. A visual interface for the ARISES diabetes management system has been designed by involving user feedback through three focus groups with patients. The interface has been designed to reduce the cognitive load on people with diabetes when they interact with the software. |
| Exploitation Route | The machine learning algorithm is a generalised framework for blood glucose prediction. It can be used as a research tool to develop diabetes management systems or be used by patients to help predict their blood glucose so that they can optimise their insulin dosing. The visual interface can be used as a generalised framework for researchers developing smart-phone apps for diabetes, as this has been designed with user feedback in mind to reduce cognitive burden and improve east of use for people with diabetes. |
| Sectors | Digital/Communication/Information Technologies (including Software),Electronics,Healthcare |
| Title | Apps and Wearables for physiological and spatio-temporal data collection. |
| Description | The project has designed a study tool for the data collection phase of the project. This includes mobile apps to ,track user location and ambient temperature (retrospective) to understand effects on glucose levels - measured with a Dexcom G5 - on T1DM participants. The project uses wearable technology, Empatica E4 band, to measure in the same instance, physiological data such as blood volume pulse, electro-dermal activity, skin temperature, and physical activity. This is then compared with CGM data to draw relationships between these variables and the glucose levels of participants. |
| Type Of Material | Physiological assessment or outcome measure |
| Year Produced | 2017 |
| Provided To Others? | No |
| Impact | This provides an easy to use platform to collect vital physiological data and assess it's impact on diabetes control. This will generate a database for the research community to then access and develop machine learning approaches to diabetes management. |
| URL | https://drive.google.com/file/d/1HVZwNIwPvy5Fe87BK7bOx_9sOhslosLO/view |
| Title | Circles of Affordance |
| Description | Invention by Robert Spence and Leah Redmond of the 'Circles of Affordance' representation of an interactive system, expected to be especially of value in the evaluation of the usability of an interactive system |
| Type Of Material | Model of mechanisms or symptoms - human |
| Year Produced | 2019 |
| Provided To Others? | No |
| Impact | This research tool will help understand user interaction with any medically related smartphone app and through this, optimise it's use for best outcome. |
| Title | Smartphone application and interface for glucose prediction and diabetes management |
| Description | The project has developed app on android and iOS systems designed aiming at make precise glucose prediction and diabetes management. People can use the app freely on smart phones. It is able to record people's meal, exercises, and other life style by manually input. Given the previous measured glucose, the app shows the predictive glucose level in the next 30 and 60 minutes with a dash line. |
| Type Of Material | Physiological assessment or outcome measure |
| Year Produced | 2018 |
| Provided To Others? | No |
| Impact | This provides an easy to use platform to collect vital physiological data and shows the future glucose levels. This allows people with diabetes to act in advance in order to avoid hypo or hyperglycaemia. |
| URL | https://github.com/danwells96/ARISES |
| Title | CRNN, a convolutional and recurrent neural network deep learning model for blood glucose prediction |
| Description | CRNN, a deep neural network that combines convolutional and recurrent neural network models to predict glucose level using as inputs historical glucose CGM data, meal data and insulin data (exercise, stress data are optional). |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2018 |
| Provided To Others? | No |
| Impact | A more accurate approach to forecasting blood glucose levels in the next 30, 60 and 120 minutes. It will help people with diabetes to manage their glucose more easily and efficiently. |
| Title | GluNet, a recurrent neural network deep learning model for blood glucose prediction |
| Description | GluNet, a recurrent neural network deep learning model to predict glucose level using as inputs historical glucose CGM data, meal data and insulin data (exercise, pressure data are optional). |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2018 |
| Provided To Others? | No |
| Impact | We will have a more accurate approach to predict the glucose level in the next 30 mins. It will help people with diabetes to manage their glucose more easily and efficiently. |
| Title | The DeepMed platform |
| Description | We have developed a deep-learning platform for analysis of biomedical data, and demonstrated its application using a diabetes dataset. All components are built using open-source software libraries. |
| Type Of Material | Data handling & control |
| Year Produced | 2018 |
| Provided To Others? | Yes |
| Impact | This platform provides healthcare professionals with an easy-to-use system to manage the heterogeneous clinical data with non-relational databases. With little specification, medical and clinical scientists with minimal expertise in data science can build their own deep learning models, and the system can train, update and deploy deep learning models autonomously. |
| Description | Advancement of the ARISES app |
| Organisation | University of Waikato |
| Country | New Zealand |
| Sector | Academic/University |
| PI Contribution | We have provided our finalised ARISES app for diabetes management to be analysed by the research partner and advance this with additional features for the user interface. |
| Collaborator Contribution | Analysis and advancement of the ARISES app for improved usability. |
| Impact | None yet. |
| Start Year | 2020 |
| Description | Machine Learning for classifying post-prandial glucose outcomes in type 1 diabetes |
| Organisation | University of Padova |
| Country | Italy |
| Sector | Academic/University |
| PI Contribution | This collaboration with the Department of Information Engineering of the University of Padova was for a Ph.D. student, Giacomo Cappon, to spend three months with us and conduct research on Machine Learning for classifying post-prandial glucose outcomes in type 1 diabetes. |
| Collaborator Contribution | Prof Andrea Facchinetti and Prof Giovanni Sparacino provided co-supervision for the PhD student. |
| Impact | Two papers have been published: Cappon, G., Facchinetti, A., Sparacino, G., Georgiou, P., & Herrero, P. (2019). Classification of postprandial glycemic status with application to insulin dosing in type 1 diabetes-An in silico proof-of-concept. Sensors, 19(14), 3168. Guemes, A., Cappon, G., Hernandez, B., Reddy, M., Oliver, N., Georgiou, P., & Herrero, P. (2019). Predicting Quality of Overnight Glycaemic Control in Type 1 Diabetes using Binary Classifiers. IEEE Journal of Biomedical and Health Informatics. |
| Start Year | 2019 |
| Title | ARISES smartphone app. |
| Description | This relates to our developed ARISES app for diabetes management. The application and machine learning algorithms have been protected by copyright. |
| IP Reference | |
| Protection | Copyrighted (e.g. software) |
| Year Protection Granted | 2019 |
| Licensed | No |
| Impact | The ARISES app will now be validated with a clinical study on people with Type 1 diabetes. Improvement in glucose control for these individuals will be a primary outcome. |
| Title | Predicting Physiologica Parameters |
| Description | The present disclosure relates to the prediction of physiological parameters. In particular, the present disclosure relates to the prediction of blood glucose levels using neural network techniques. |
| IP Reference | PCT/GB2019/053115 |
| Protection | Patent application published |
| Year Protection Granted | 2020 |
| Licensed | No |
| Impact | None yet. We are in discussions with commercial entities for licensing at present. |
| Title | Phase 1 observational study for ARISES interface |
| Description | Clinical physiological and environmental data collection (Phase 1) The adaptive real time intelligent system to enhance self-care of chronic disease (ARISES) is a novel platform capable of collecting data from multiple sources to empower people to effectively self-manage type 1 diabetes (T1DM) through therapeutic and lifestyle decision support (figure 1). Combining wearable sensors and smartphone technology, a range of biological, environmental and behavioural data will feedback into an artificial intelligence algorithm to output real-time decision support. The first phase of clinical trials will recruit 12 individuals with T1DM for a six-week observational study. This study will serve as a training set using wearable technologies to collect data and evaluate blood glucose correlations against physiological and environmental case parameters. Useful associations will assist the development of the ARISES algorithm and identify wearable devices for the final ARISES platform. This study will also provide questionnaire and feasibility outcomes for combining both CGM and a physiological data acquisition sensor. Glucose levels will be monitored using Dexcom G6 real-time continuous glucose monitoring (RT CGM). Physiological parameters (heart rate, blood volume pulse, skin temperature, motion and sympathetic activity, and electrodermal activity) will be monitored using the Empatica E4 wristband. Data from both wearable devices will be uploaded anonymously to a safe cloud-based server for analysis. Participants will be asked to log manual data inputs (e.g. exercise details, meal macronutrients, alcohol content and illness) into the mySugr smartphone applications for retrospective analysis. On completion of the study, location and ambient temperature data will be retrospectively accessed from myTracks app and smartphone location features. |
| Type | Management of Diseases and Conditions |
| Year Development Stage Completed | 2019 |
| Development Status | Under active development/distribution |
| Impact | We will have physiological data including glucose, heart rate, activity, skin temperature over a 6 week period in 10 people with diabetes. This study will serve as a training set using wearable technologies to collect data and evaluate blood glucose correlations against physiological and environmental case parameters. Useful associations will assist the development of the ARISES algorithm and identify wearable devices for the final ARISES platform. This data will allow us to quantify the life style of someone with diabetes and improve care through more accurate prediction of glucose and therefore insulin needed. |
| Description | 3rd year EEE student project proposal |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | Local |
| Primary Audience | Undergraduate students |
| Results and Impact | The ARISES team presented to a third year undergraduate cohort from the Department of Electrical and Electronic Engineering, Imperial College London. From this a group of 4 students carried our a group project where they researched a new design and implementation for the ARISES interface. They completed their work with usable interface in iOS system, and demonstrated their work in the entrance of Imperial College in front of a large audience. |
| Year(s) Of Engagement Activity | 2018 |
| Description | ATTD 2020 Poster Presentation - A NOVEL HAND-HELD INTERFACE SUPPORTING THE SELF-MANAGEMENT OF TYPE 1 DIABETES |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Industry/Business |
| Results and Impact | Our group presented at the 13th international conference on Advanced Technologies and Treatments for Diabetes. This is an annual conference attended by 4000 participants from including academia, clinicians and industry. We showed our work as a result of this grant entitled 'A NOVEL HAND-HELD INTERFACE SUPPORTING THE SELF-MANAGEMENT OF TYPE 1 DIABETES'. |
| Year(s) Of Engagement Activity | 2020 |
| URL | https://attd.kenes.com |
| Description | ATTD 2020 Poster Presentation - PERSONALIZED MEAL INSULIN BOLUS FOR TYPE 1 DIABETES USING DEEP REINFORCEMENT LEARNING |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Industry/Business |
| Results and Impact | Our group presented at the 13th international conference on Advanced Technologies and Treatments for Diabetes. This is an annual conference attended by 4000 participants from including academia, clinicians and industry. We showed our work as a result of this grant entitled 'PERSONALIZED MEAL INSULIN BOLUS FOR TYPE 1 DIABETES USING DEEP REINFORCEMENT LEARNING'. |
| Year(s) Of Engagement Activity | 2020 |
| URL | https://attd.kenes.com |
| Description | ATTD 2020 Poster Presentation -AN ADVANCED CLINICAL DECISION SUPPORT PLATFORM FOR THE MANAGEMENT OF TYPE 1 DIABETES |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Industry/Business |
| Results and Impact | Our group presented at the 13th international conference on Advanced Technologies and Treatments for Diabetes. This is an annual conference attended by 4000 participants from including academia, clinicians and industry. We showed our work as a result of this grant entitled 'AN ADVANCED CLINICAL DECISION SUPPORT PLATFORM FOR THE MANAGEMENT OF TYPE 1 DIABETES'. |
| Year(s) Of Engagement Activity | 2020 |
| URL | https://attd.kenes.com |
| Description | Demonstration of the Bio-inspired Artificial Pancreas at the Science Museum Night Owls event, for Autism. |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Public/other audiences |
| Results and Impact | We gave a live demonstration of the Bio-inspired Artificial Pancreas at the Science Museum Night Owls event. This was an event run to showcase technologies to people with Autism. |
| Year(s) Of Engagement Activity | 2017 |
| Description | Demonstration of the Dilated Recurrent Neural Network (RNN) for Short-Time Prediction of Glucose Concentration |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | We introduced our latest deep neural network model to make accurate glucose prediction using dilated RNN. This was a workshop for healthcare data science. |
| Year(s) Of Engagement Activity | 2018 |
| Description | Diabetes Patient focus groups to design the ARISES smartphone interface. |
| Form Of Engagement Activity | A formal working group, expert panel or dialogue |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Patients, carers and/or patient groups |
| Results and Impact | We have run two patient focus groups for design of the ARISES smarthphone interface. Our objective is to reduce the cognitive load of diabetic patients when they interact with the platform. Current focus groups to date: FOCUS GROUP 1 - 7/11/17 - 11 attendees (7 patients + 4 investigators), FOCUS GROUP 2 - 8/3/18 - 7 attendees (4 patients + 5 investigators) |
| Year(s) Of Engagement Activity | 2017,2018 |
| Description | Diabetes UK abstract identified inpatient seasonal variation in capillary glucose |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Professional Practitioners |
| Results and Impact | Data suggests that ambient temperature may impact glucose metabolism and that heat adaptation strategies may be impaired in diabetes. The seasonal impact on glycaemia remains unclear. We aimed to determine annual variation in glucose and glucose variability in inpatients. Significant variation in average capillary glucose was identified between the twelve calendar months and four seasons. This work was presented at the annual Diabetes UK conference attended by healthcare professionals from around the UK. |
| Year(s) Of Engagement Activity | 2018 |
| Description | Imperial College Student Poster Session |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | Local |
| Primary Audience | Postgraduate students |
| Results and Impact | MSc students participated in the development of the deep learning algorithms for the ARISES platform. They completed their work with two different algorithms to predict glucose levels for Type 1 diabetes. They demonstrated their work in the entrance of Imperial College. |
| Year(s) Of Engagement Activity | 2018 |
| Description | Invited talk on the Bio-inspired Artificial Pancreas at the New York University, Abu Dhabi |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Postgraduate students |
| Results and Impact | I was invited to give a talk on The Bio-inspired Artificial Pancreas at the New York University, Abu Dhabi. This talk was attended by undergraduate and post-graduate researchers of the university in addition to faculty. This soaked an increased interest in the field of diabetes technology. |
| Year(s) Of Engagement Activity | 2017 |
| Description | Presentation at the symposium organised by Centre for Continuing Professional Development, Imperial College |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | This was a talk about generating awareness to endocrinologists from China on the diabetes technologies being developed. |
| Year(s) Of Engagement Activity | 2017 |
| Description | Presentation of the at the symposium organised by Astra-Zeneca entitled 'Advances in Diabetes Technologies, Management and Prevention - An insight into Imperial College London's Innovative Research' |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | This was an event focused at educating 200 endocrinologists from Spain on current diabetes technology developed at Imperial College |
| Year(s) Of Engagement Activity | 2017 |
| Description | Talks to industry on The bio-inspired artifical pancreas for treatment of diabetes in the home |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Industry/Business |
| Results and Impact | I was invited to give a talk at ARM (https://www.arm.com) in Cambridge. They are one of the world leaders in embedded systems and my talk was focused on introducing them to diabetes technologies. The talk was attended by about 60 employees of the company. |
| Year(s) Of Engagement Activity | 2018 |
| URL | https://www.arm.com |
| Description | The Bio-inspired Artificial Pancreas Live Demonstration, Demonstration of the Bio-inspired Artificial Pancreas at the Royal Institution |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Public/other audiences |
| Results and Impact | We conducted a liver demonstration of the Bio-inspired Artificial Pancreas the the Royal Institution Family Fun Days. These are Weekend events to get the whole family excited about science. |
| Year(s) Of Engagement Activity | 2017 |
| URL | http://www.rigb.org/families/family-fun-days |
| Description | Third patient participation focus group focused on interface design and usability |
| Form Of Engagement Activity | A formal working group, expert panel or dialogue |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Patients, carers and/or patient groups |
| Results and Impact | FOCUS GROUP 3 - 6/6/18: User focus group consisting of people with diabetes to help get user feedback for the design of the ARISES smartphone interface. 11 attendees (3 patients, 9 investigators and students) |
| Year(s) Of Engagement Activity | 2018 |
| Description | Workshop on information Visualisation |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Postgraduate students |
| Results and Impact | A design oriented workshop of Information Visualization. Students learn how to design to support the formation of a mental model of data. Specifically, it is concerned with the representation, presentation and interaction with data, and therefore relevant to interaction design within the ARISES project. |
| Year(s) Of Engagement Activity | 2018 |