The CHIRON architecture for a “continuum of care”

The CHIRON reference architecture seamlessly integrates devices and services creating interoperability between heterogeneous equipment and devices used throughout the overall care cycle from primary prevention (persons still healthy) to secondary prevention (initial symptoms or discomfort) and to tertiary prevention (disease diagnosis, treatment and rehabilitation) and in various domains (at home, in nomadic environments and in the clinical settings).

The CHIRON reference architecture is based on the continuous interaction and exchange of information between three layers:

  • The USER plane, principally involving interactions by and with the patients (monitoring and local feedback),
  • The MEDICAL plane, principally involving interactions by and with doctors (assessment of clinical data, diagnosis, treatment planning and execution, feedback to the patient),
  • The STATISTICAL plane, principally involving interactions by medical researchers (external knowledge management).

The USER PLANE mainly deals with data measured on the patient and his environment through a minimally invasive Wireless Sensor Network (WSN) and data provided by the patient himself (patient’s profile). It includes the mobile device of the patient running an optimised, light-weight service platform and providing first feedbacks to the patient i.e., information about his status, warnings and motivation towards healthier habits and lifestyle. In case of critical situations, the system will trigger a direct alarm to first aid services (emergency service).

The sensor nodes will be built with their own local data analysis and long-short term repositories. This distributed approach in data analysis and data storage will allow a reduction of the needed communication bandwidth (interchange limited to the already analysed data and not to the overall set of raw data) while an information distributed over many different repositories will enhance security and reliability.

An approach followed in CHIRON is to use various measurements and signals to measure the same underlying clinical phenomenon: the gradually worsening condition of a chronic patient. The aim is to improve the quality and robustness of the indicator, thus improving sensitivity without causing false alarms. The final goal is to come up with an embedded device or system that measures simultaneously a number of simple parameters from easily applicable low-cost sensors, but that taken together form a powerful diagnostic tool.

In the WSN of the USER PLANE, CHIRON will go beyond the state of the art in terms of management of systems made of resource-constrained embedded devices and in the seamless integration of heterogeneous devices.

All the data which are relevant for the clinical history and the risk profile of the patient are sent to the MEDICAL plane (in the form of raw data or elaborated info). In this plane, the medical staff will interact with the CHIRON system to validate data derived from each patient and correlate different parameters to his clinical history and the current care plan in order to make the appropriate decisions (planning of intervention and/or reviewing of the pharmaceutical treatments).
Finally, the (potentially large) volumes of manually and electronically collected data for each patient are stored in a medical repository inside the STATISTICAL plane. The medical staff uses that repository as a source of information about an entire community of patients that suffers of the same disease. The data in this plane can then be statistically analyzed to discover or validate macroscopic relationships between patient profiles, diagnosis and the efficacy of the medical treatments.
There is no central database but a distributed approach where patient’s data are stored in the local resources of the overall network; it will constitute a Virtual Patient Repository (VPR) defining a virtual structure for the data and providing a common interface. The system will search for data in the network and will correlate patient information.
The Virtual Patient Repository (VPR) represents the layer integrating the patient’s personal health system data – gathered through the Wireless Sensor Network of the non-clinical settings and enriched with personal information – with all the patient’s clinical data gathered in the clinical setting including data from the medical imaging system. The VPR will be fully integrated into the clinical workflow and will interface with the Hospital Information System. In this way the person-centric health care approach of CHIRON will be fully integrated and empowered with the specific hospital-related data that represent the focus of the HIS.
To achieve the maximum benefit, the patient’s profile resulting from all the available information will have “a structured content agreed at an international level, starting from a few generic “summaries” and gradually developing a series of “summaries” specific for each clinical context” (see “ European Commission Recommendation on cross-border interoperability of Electronic Health Record systems”, July 2008).

There are functional interdependence and coupling between the three planes: observations and insights obtained by the medical staff from the MEDICAL PLANE will be used to affect the process of data collection and other localized processes performed on the USER PLANE (e.g., the reliance on accelerometer data may be decreased if the accelerometer–derived activity inferences are determined to be consistently incorrect).

Likewise, insights gained in the STATISTICAL PLANE will be used by professionals in the MEDICAL PLANE to alter the specific medical rules or the nature of prescribed treatment for corresponding individuals. Using these three planes, the Project will provide extended, person-centric and multi-parametric monitoring and analysis functionalities that will help in providing early identification of worsening in the disease and prediction of degenerative events, enabling opportune intervention.

In the CHIRON system the analysis of the patient’s data will result in the activation of feedbacks at different levels:

  • A first feedback is activated automatically at the user plane (inner loop): actual data monitored through the sensors network and complemented with additional personal information are analyzed, compared with a personalized “expected health status behavior” (Alter Ego profile). From this, short and medium / long term feedbacks are generated in the form of reminder, warnings and motivations to healthier lifestyle or – in case of the occurrence of a critical deterioration of the health conditions – in the form of an alarm and request for direct medical intervention.
  • A second type of feedback is activated in a loop resulting from the integration of the user plane and the medical plane; it is generated by the medical professional that – alerted by the sensor’s data and supported, when necessary, by the results of clinical tests – defines or changes the therapeutic program of the patient.
  • The “medical feedback” will be supported by the integration of epidemiological information related to a community of patients with similar disease or symptoms; in this case an outer loop is activated involving also the statistical plane.

It will be a specific objective of CHIRON to reduce the “alert fatigue” i.e., to rank priority order and to reduce the number of computer-generated recommendations, alerts both for the clinician and for the patient.

The interaction of User, Medical and Statistical Planes in the CHIRON system

Important aspects of the CHIRON reference architecture are:

  1. The integration of heterogeneous subsystems that can operate autonomously and give partial input to the person centric health information system. These subsystems are built with their own local data analysis and long or short-term repositories;
  2. The reduction of the needed computing resources and of the needed communication bandwidth through the interchange of already analyzed data available in the local repositories;
  3. Information distributed – for security and reliability reasons – over many different repositories and distributed and conditioned data access to these heterogeneous repositories;
  4. Personalization (personalized monitoring, personalized risk assessment, personalized feedback and therapeutic program);
  5. Robustness and reliability of the captured data through a multiparametric monitoring approach.