The successful integration in hospitals and primary care centers of systems based on Artificial Intelligence (AI) techniques has a high transformative impact on the efficiency of the health system, decision-making by its professionals and managers, and the improvement of health conditions and quality of life of patients and caregivers. This communication presents the Smarchronic platform developed through a project funded by the Valencian Innovation Agency- Generalitat Valencia (ref. INNEST/2020/47) in 2020 and 2021. This work has been carried out by an interdisciplinary team made up of researchers from Polibienestar Research Institute at the University of Valencia, technicians from two companies, Nunsys and Outcomes10, health professionals and researchers from La Fe hospital and researchers from Technological Institute of Informatics at the Polytechnic University of Valencia.
The Smartchronic proposes to apply AI techniques in combination with monitoring techniques based on Patient-Reported Outcomes (PROs) in patients aged >70 years, with chronic pathologies establishing a predictive model of frailty, based on the Frailty Index of Rockwood, a pathway prediction model, a 30-day readmission risk prediction model and non-invasive monitoring of patients through mobile technology (mHealth). The objective is to improve the integrated care received by these older patients who require long-term care, based on their frailty, considering their behavior and needs, from a proactive patient-centered perspective.
To develop and validate the platform, retrospective and prospective studies were carried out using data from the Information Systems (SI) of the Hospital la Fe (Valencia, Spain). The Project was approved in 2020 by the Ethical Clinical Research Committee the aforementioned hospital (Reference number 2020-138-1 and all necessary data exchange was compliant with current national and European data protection laws.
The platform consist of: a management web based platform; Machine Learning models of care pathways, classification of patients according to IF and hospital readmission; model training platform; analytical solution to optimize retrospective data for visualization within dashboards; and App for Smartphone to patient monitoring. All communications implemented SSL protocol, and for the determination of the calculations the ETL Pentaho Data Integration tool has been used.
The main result of the platform was the definition of a set of KPIs indicators, aimed at professionals to optimize the long-term care plans of chronic patients. The KPIs use data from the different modules (IF, care pathway, readmissions and PROs) to provide information on real adherence and deviations from the established care plans and those foreseen by SmartChronic, evolution of variables included in the system (fragility, critical conditions, risk of readmission), capacity and reaction of the health system.
In conclusion, piloting at scale in real healthcare environments such as the one proposed by SmartChronic is necessary to demonstrate the transformative impact of AI on the healthcare system. The project contributes, in the context of the integration of digital platforms into health operating environments, to demonstrate the existence of sufficient operational and economic benefits to justify a wider adoption of these technological solutions by health policy makers.