Periodic Reporting for period 1 - PHRASE (Personalised Health cognitive assistance for RehAbilitation SystEm)
Période du rapport: 2022-04-01 au 2023-03-31
Stroke is the second leading cause of death worldwide and the primary contributor to the burden of neurological disease. According to the World Stroke Organization (WSO), there are over 13.5 million stroke cases every year and 80 million stroke survivors globally.
In Europe, the annual costs associated with stroke are 60 billion euros, and prevalence is projected to increase from 9.35 to 12.45 million patients (12%) by 2045. Additionally, the WHO estimates the shortage of trained health workers will go from 7.2 million in 2013 to 12.9 million by 2035, and the United Nations (U.N.) estimates that 18 million additional health-workers will be needed by 2030 to meet the U.N. Sustainable Development Goals (SDGs) in low- and lower-middle-income countries. These growth trends constitute a true crisis for global healthcare systems.
About half of all stroke survivors present post-stroke dementia (PSD) or post-stroke cognitive impairment (PSCI), several specific and diffuse cognitive disabilities related to memory, attention, emotional states, orientation, language understanding and production. The sustained impairments and the lack of retention of the functional gains achieved at the hospital (up to 80% of function) after discharge lead to a 44% readmission rate within 6 months after discharge due to various health deterioration-related issues, including falls. Stroke readmissions cost approximately 19.400€ on average, with significant variations depending on health complications. Moreover, about 60% of patients experience chronic effects, including pain, mood disorders, and depression.
The objective of PHRASE (Personalised Health cognitive assistance for RehAbilitation SystEm) is to create a data pipeline, from data to intervention, that integrates the best available scientific knowledge and data infrastructures to deliver diagnosis, prognosis, and intervention tools for stroke and other brain-related diseases. PHRASE will deliver AI-based patient models to predict the patient's recovery path and the associated rehabilitation treatment (ie, frequency, intensity, volume, and type of training) required to obtain the best possible clinical outcomes. Additionally, PHRASE will deliver the intervention and monitoring solutions necessary to provide a continuum of training for the hospitals to patients’ homes.
PHRASE will enable health professionals to make more accurate assessments and prognoses, as well as tailor more effective treatment plans. For patients, this translates into faster recovery, reducing the treatment period at the hospital by up to 40%, and continued training after discharge to better retain motor and cognitive functions, resulting in a potential reduction in readmissions of up to 18%.
As a whole, the PHRASE pipeline will be a solution that can be trained and applied to diverse disease-specific datasets. Consequently, PHRASE has the potential to be valuable beyond stroke, as it can model patient recovery trajectories for various pathologies such as Alzheimer's, Parkinson's, Multiple Sclerosis, Chronic Pain, and numerous other chronic diseases.
In Europe, the annual costs associated with stroke are 60 billion euros, and prevalence is projected to increase from 9.35 to 12.45 million patients (12%) by 2045. Additionally, the WHO estimates the shortage of trained health workers will go from 7.2 million in 2013 to 12.9 million by 2035, and the United Nations (U.N.) estimates that 18 million additional health-workers will be needed by 2030 to meet the U.N. Sustainable Development Goals (SDGs) in low- and lower-middle-income countries. These growth trends constitute a true crisis for global healthcare systems.
About half of all stroke survivors present post-stroke dementia (PSD) or post-stroke cognitive impairment (PSCI), several specific and diffuse cognitive disabilities related to memory, attention, emotional states, orientation, language understanding and production. The sustained impairments and the lack of retention of the functional gains achieved at the hospital (up to 80% of function) after discharge lead to a 44% readmission rate within 6 months after discharge due to various health deterioration-related issues, including falls. Stroke readmissions cost approximately 19.400€ on average, with significant variations depending on health complications. Moreover, about 60% of patients experience chronic effects, including pain, mood disorders, and depression.
The objective of PHRASE (Personalised Health cognitive assistance for RehAbilitation SystEm) is to create a data pipeline, from data to intervention, that integrates the best available scientific knowledge and data infrastructures to deliver diagnosis, prognosis, and intervention tools for stroke and other brain-related diseases. PHRASE will deliver AI-based patient models to predict the patient's recovery path and the associated rehabilitation treatment (ie, frequency, intensity, volume, and type of training) required to obtain the best possible clinical outcomes. Additionally, PHRASE will deliver the intervention and monitoring solutions necessary to provide a continuum of training for the hospitals to patients’ homes.
PHRASE will enable health professionals to make more accurate assessments and prognoses, as well as tailor more effective treatment plans. For patients, this translates into faster recovery, reducing the treatment period at the hospital by up to 40%, and continued training after discharge to better retain motor and cognitive functions, resulting in a potential reduction in readmissions of up to 18%.
As a whole, the PHRASE pipeline will be a solution that can be trained and applied to diverse disease-specific datasets. Consequently, PHRASE has the potential to be valuable beyond stroke, as it can model patient recovery trajectories for various pathologies such as Alzheimer's, Parkinson's, Multiple Sclerosis, Chronic Pain, and numerous other chronic diseases.
In terms of data management, we have designed a process for data standardization that adheres to ethical standards and FAIR principles. We have also defined a workflow that includes guidelines for data collection, annotation, and metadata, including the decision on data and metadata structure standards (openMINDS). Furthermore, we have evaluated the relevant jurisprudence related to the project's scope. As part of these activities, we have collaborated with an external Ethics Advisor to ensure the implementation of privacy and confidentiality policies, minimize biases in data collection and analysis, ensure data quality and accuracy, and consider broader societal implications while distributing benefits and risks fairly.
In terms of technical progress, we have developed patient models for treatment personalization, diagnosis, and prognosis. These models have been integrated into a backend that connects to the Rehabilitation Gaming System (RGS) solutions, namely RGSclinic, RGSapp, and RGSwear, for intervention and monitoring. Moreover, new training protocols for cognitive functions such as memory, attention, executive function, and language, as well as a diagnosis task, have been developed in RGSapp. Additionally, improvements have been made to RGSclinic and RGSwear to facilitate future market updates. For example, the costs of required equipment have been reduced through the development of a proprietary body tracking algorithm. To integrate the pipeline, the current backend has been connected to the Virtual Research Environment (VRE), and an API has been developed to collect and register the RGS metadata in the EBRAINS Knowledge Graph (KG).
Lastly, as part of the clinical validation activities, we are working on a clinical validation protocol and corresponding documentation for the ethics committee. However, the limited budget for clinical activities as compared to project ambitions must be addressed and may impact the ultimate design of the clinical protocol.
In terms of technical progress, we have developed patient models for treatment personalization, diagnosis, and prognosis. These models have been integrated into a backend that connects to the Rehabilitation Gaming System (RGS) solutions, namely RGSclinic, RGSapp, and RGSwear, for intervention and monitoring. Moreover, new training protocols for cognitive functions such as memory, attention, executive function, and language, as well as a diagnosis task, have been developed in RGSapp. Additionally, improvements have been made to RGSclinic and RGSwear to facilitate future market updates. For example, the costs of required equipment have been reduced through the development of a proprietary body tracking algorithm. To integrate the pipeline, the current backend has been connected to the Virtual Research Environment (VRE), and an API has been developed to collect and register the RGS metadata in the EBRAINS Knowledge Graph (KG).
Lastly, as part of the clinical validation activities, we are working on a clinical validation protocol and corresponding documentation for the ethics committee. However, the limited budget for clinical activities as compared to project ambitions must be addressed and may impact the ultimate design of the clinical protocol.
PHRASE will deliver patient models to improve the diagnosis, prognosis, and intervention protocols for stroke and other brain-related diseases. This will enable better and more efficient treatment of stroke-caused impairments, leading to improved outcomes for patients and reduced costs for hospitals. We expect PHRASE to reduce outpatient treatment by 40% and readmission rates by 18%.
To ensure the uptake and success of the project, our first step is to gather the necessary data to validate the diagnosis and prognosis models and intervention approaches. After that, we need to protect the intellectual property (IP) and go through the certification process to commercialize the results in Europe. It will be crucial to have the necessary support to navigate the diverse regulatory landscape in Europe and to have expertise in internationalization and scaling of digital health products in the European markets.
To ensure the uptake and success of the project, our first step is to gather the necessary data to validate the diagnosis and prognosis models and intervention approaches. After that, we need to protect the intellectual property (IP) and go through the certification process to commercialize the results in Europe. It will be crucial to have the necessary support to navigate the diverse regulatory landscape in Europe and to have expertise in internationalization and scaling of digital health products in the European markets.