Hackaton euvsvirus-gnuhealth-federation
0 likes283 views
The Gnu Health Federation has developed a COVID-19 surveillance module to assist governments in managing the pandemic by providing real-time epidemiological statistics and monitoring population behavior. This initiative aims to support health professionals in data collection, case management, and resource allocation at multiple levels, from local to global. The project emphasizes collaboration, the integration of diverse data sources, and the philosophy of open-source software to address health challenges effectively.
Hackaton euvsvirus-gnuhealth-federation
- 1. Tackling the Beast: Using the GNU Health Federation to help governments in the fight against the COVID-19 pandemic Pan-European Hackaton EU vs Virus: April 24-26, 2020 Challenge: Health & Life communication and prevention Permanent observatory to collect data on the behavior and habits of the population in order place preventive measures to avoid a new emergency
- 2. Background and justification Europe and the whole world are living through an unprecedented crisis due to the COVID-19 pandemic. Since its detection in December 2019 in the central Chinese city of Wuhan, the virus infected more than 2.8 million people (confirmed cases) and is involved in the death of more than 197.000 people in the world. The most affected countries are Italy, Iran, Spain, Germany, South Korea, France, the United Kingdom, and the United States. A global alliance has been established to fight this pandemic. As part of its effort, the GNU Health team has developed an additional module for COVID-19 surveillance. The initial tests have been successfully conducted. The module allows the processing of a full COVID-19 case tracking workflow from data collection to reporting, as well as epidemiological surveillance report on COVID-19 and other health conditions. Given this background, we are honoured to join the EUvsVirus challenge. Project objectives The objectives of the GNU Health Team in their contribution to the European and global effort to fight the COVID-19 pandemic are: adding additional functionalities to GNU Health to allow the real time surveillance of COVID-19 and other epidemics at four levels (local, regional, national, global). supporting health professionals and policy makers in their effort to contain the disease by providing a health information system for: ◦ providing real time COVID-19 epidemiological statistics at the GNU Health Federation portal. ◦ monitoring the habits and behaviours of the population in the context of COVID-19 ◦ Finding the causality and correlation in the susceptibility and disease progression based on the socioeconomic status, lifestyle, ethnicity, and other biological and environmental contexts. ◦ sharing ideas and resources with international teams to enhance
- 3. collaborative efforts ◦ promoting the philosophy of Free and Open Source software and social medicine to tackle the various health challenges that Europe and the world are facing today and will continue facing in the near future. Project scope Real time data collection and management of: ● population demographics and patient information ● population and patient demographics, domiciliary units, socio-economics, Lifestyle information ● in the case of COVID-19, extension of the lifestyle and socio-economics parameters to include parameters of physical distancing, isolation, contact tracing and quarantine Physical infrastructure management: ● procurement, inventory and stock management of medicaments, medical devices, protective masks and other personal protective equipment (PPE) ● management of the availability of resuscitation beds ● management of intensive care unit (ICU) bed capacity ● monitoring of bed availability for COVID-19 patients ● integration with hospital pharmacy, to include the supply, delivery, storage, inventory, billing and dispensing of medicines. ● batch management, including monitoring of expiry dates of medicines, and further details required to prevent improper dispensing of medications. ● tools for management of ambulatory care centre at low costs via mobile devices
- 4. Case management GNU Health is already compliant with most requirements of the WHO epidemiological case management, including: ● Support for COVID-19 case management workflow as defined by the WHO ● Recording for each COVID-19 patient the symptoms, medical history, medication history, contact history, travel history, associated laboratory and imaging testing, diagnostic activities ● Support of ICD-10, ICPM, ICD9-CM, Patient Functionality and Disability, WHO essential list of medicines, vaccination calendars, Millennium Development Goals (now SDG), Injury Surveillance System, among many other World Health Organization standards and guidelines. Reporting: ● patient evaluations at health facility, care at home, or via tele-consultation or at any other location ● real time identification of patients with severe symptoms, who are fragile or at high risk ● Person contact tracing Workforce management GNU Health includes operational human resource management functionality. The system allows for the registration of all employees and the management functions for health professionals working at a health institution. The appointment management module allows the optimal management of the hospital agenda planning by managing health professional calendars and appointments Remote medical assistance: ● Remote recording and access of patient medical records via various devices (smartphone, tablet, Laptop, PC, Raspberry Pi, etc.) ● Tele-consultation of patients
- 5. Geographical information system Capture of geographical information for domiciliary units and presentation of spatial and visualisation information via OpenStreetMap. Data security and privacy: Data anonymization, patient privacy, data security, digital signature and encryption can be enforced and adjusted to country laws and policies. GnuPG as established standard is used for encryption. Real time knowledge transfer and management: ● real time knowledge transfer to and from all the nodes based on user access and permissions (Federation model) ● All medical information and patient records can recorded, updated and transferred electronically to all nodes, as permitted ● Medical records can be searched (even in text form) and easily shared among health professionals. Information loss is reduced to a minimum Real time reporting of COVID-19 epidemiological statistics: ● Epidemic curve - Number of reported cases over time ● Demographics distribution of cases - Case distribution by age group, gender socioeconomic status, work ● Symptoms and severity - Number of reported cases by symptoms and severity ● Number of cases by geographical area - Number of deaths based on various criteria ● Number of persons tested
- 6. Epidemiological Surveillance We have chosen to provide support: locally, regionally, nationally and internationally. We therefore need 4 levels of overview. This document will describe the development task for the dashboards that will be provided by the GNU Health Federation in the task of the hackathon. We will be using OpenStreetMap for our mapping and matplotlib for construction of data dashboards. We will adopt the Cross- Industry Standard Protocol for Data-Mining (CRISP-DM) approach to specification of the requirements and the data that will be used in the visualisations. Where the GNU Health Federation approach can add something will be in allowing care centres and centres of respite to be recognised quickly, so that for example, if hotels are used for infected patients, their bedding capacity and support features could be entered quickly. The numbers of beds available could then be aggregated and displayed as well. GNU Health Federation Health Information System: Data collection The GNU Health Federation (GH Federation) is a distributed network composed of multiple heterogeneous nodes. A node is the basic unit of the GH Federation, and it can be a citizen Personal Health Record, a health professional, a health institution or a research center, to name some examples. In the case of the European Union, the EU GNU Health Federation would be composed of millions of citizens and thousands of institutions from all over the Union. The data generated in those nodes can be transferred to the GH Federation Health Information System (HIS) via GNU Health Thalamus, the message server and authentication authority. Visuals for Epidemiological Data for Different Levels of Locale If we are going to provide these screens we need to know what professionals need them and what they really need to see on them. In CRISP-DM, this is described as part of the
- 7. Business Understanding. And we also need to know what data needs to be accessed to build the needed screens. This is referred to as Data Understanding in CRISP-DM. Business Understanding Medical Objectives: to service the needs of key professionals in understanding the situation regarding the Virus. Impactful visualisations are needed by everybody who has to argue and convince patients and people who are asking for advice regarding the medical situation. Clear data availability allows for up-to-the minute evidence-based advice. Who needs those screens most: public health officers and epidemiologists. These professionals need: - - population data (by gender, age, sex, ethnicity, domiciliary units, socioeconomic status, work, families, lifestyle,...);; - - resources data: number of hospital beds and ICU beds, occupied and free, number of ventilators and respirators, ECMO. Healthy health personnel versus quarantined, sick and dead personnel;; - - patient data regarding: infected, symptomatic, died, recovered, inhospital, in ICU, with ventilator, with ECMO (stratified with risk factors, age, gender, comorbidity, living conditions eg retirement homes, slums, ...);; - - details of number of tests (PCR, antibodies) taken;; - - details of number of quarantined contact persons;; - - mapping of effectiveness of measures: do people stay at home, do they reduce close contacts with family members, friends and at the workplace, do they apply hygienic measures, etc. Our aim is to show that these professional needs can be serviced by GNU Health.
- 8. Data Understanding The data understanding stage of CRISP-DM process requires access to the data to be understood. These data comprise several tiers of data collected at a local, regional, national and worldwide level that must be safely accessed. For this we will use the GNU Health Federation model. The use of GNU Health solves a major issue that is the integration of data required when gathering data from multiple sources. A first examination of the data collected will provide information about the amount of data collected, the type (e.g., categorical: nominal or ordinal; numerical: continuous or discrete), descriptive statistics (e.g., range, average, standard deviation, median, skewness, kurtosis, mode) of each variable. In the subsequent step, the distribution of key variables, relationships between pairs or small numbers of variables, properties of sub-populations and simple statistical analysis will be performed to gain detailed insight about possible interesting attributes. Findings will be reported with summary tables and various types of graphs. A good model for an overview dashboard might be that as provided by Johns Hopkins: Figure 1: A Base Display Format If we aim to provide four visualisations: 1. one that would be suitable for a local understanding of the situation (L) 2. one for the regional situation (R) 3. one for the national situation (N) 4. one for the worldwide situation (W)
- 9. New GNU Health Epidemiological Visualisations In order to trace the Epidemic Curve, its Geographical Spread and the Capacity to deal with it, we could adapt the dashboard shown in figure 1, as shown in figure 2, with data provided live to the blue spaces. Figure 2: Epidemic Curve, Geographical Spread and Resource Capacity Showing: Epidemic curve - Number of reported cases over time, Geographic Spread, and Service Capacity. Our first attempt at a multi-level dashboard is shown in figure 3. Figure 3: Four Levels of Reportage
- 10. This misses some of the bottom views like news and total infected plot vs. time and daily cases. However with editing and layout change we can address these issues. Regarding an International Regional view, our first work suggests: For the National and Local views, it is here that country-specific monitoring of internal resources becomes important. Regarding the national and local screens of a real-time observatory based on our key proposals we may also use the panels on the right hand side of the screen to show possibly as an option: - - Demographics distribution of cases - Case distribution by age group, gender, socio-economic group, living conditions e.g. retirement homes, type of home, slums, etc. and other criteria;; - - Symptoms and severity - Number of reported cases by region by symptoms and severity, infected, symptomatic, died, recovered, inhospital, in ICU, with ventilator, with ECMO (stratified with risk factors, age, gender, comorbidity);); - - Number of persons tested, numbers with virus, means of testing, features of test, place of test, (PCR, antibodies), number quarantined, contact persons;; - R- Resources data: number of hospital beds and ICU beds, occupied and free, number of ventilators and respirators, ECMO. Healthy health personnel versus quarantanized, sick and died personnel;; - E- Effectiveness of measures: do people stay at home, do they reduce close contacts with family members, friends and at the workplace, do they apply hygienic measures, etc..
- 11. Supplementary Panels Regarding these visualisations, they may be shown graphically in a number of ways, using data visualisation techniques, such as: (1) (2) (3) Figure 3: (1) covariance matrix between variables A, B, C, D and E showing the joint variability of each pair of variables as a heatmap. (2) joint plot as a kernel density estimate (kde) displaying relation between variables X and Y with respective histograms/kde on top for variable X and on the right for variable Y. (3) pairwise bivariate distributions for sub-populations.
- 12. The final step of data understanding, will involve the assessment of the quality of the data by determining if data is complete, if there are variables that may contain errors or missing values, how frequent is the occurrence of such errors or missing values, how are they represented and how should be dealt with based on the information gathered on the business understanding. A report will be produced to summarize this information. Conclusions The GNU Health Federation architecture has shown itself to be a powerful tool for recording parameters that shows the impact of the socioeconomic determinants in the health of the population. This is a powerful issue that is at the heart of social medicine, where GNU Health has placed itself. It is precisely this aspect of GNU Health that makes it suitable as a tool for monitoring the treatment of a pandemic at local, regional, national and international levels. The GNU Health Federation would provide the solution to the challenge of creating a Permanent observatory to collect data on the behavior and habits of the population in order to place preventive measures to avoid a new emergency. Now, any citizen of the European Union, as well as primary care centers, hospitals, laboratories and research institution, can immediately be included as a participating node in the GNU Health Federation, contributing to knowledge about the sources and vectors of the pandemic, and receiving information from aggregated data elements. As health services adapt to cope with whole-population monitoring, the GNU Health Federation with its new real-time observatory functions can be a valuable partner system.
- 13. Many people have worked with us at this weekend hackaton. The European Commision, to the organizers, volunteers, the coordinator and mentors!. On befalf of our team, thank you! Name Country Role Armand Mpassy Germany Team member Joao Santinha Portugal Team member Edgar Hagenbichler Austria Team member Tom Kane Scotland Team member Leonardo D’acchille Argentina Team member Thomas Karopka Germany Team member Ingrid Spessotti Argentina Team member Axel Braun Germany Team member Fernando Sassetti Argentina Team member Abdrahman Elkafil Morocco Team member Luis Falcon Spain Team member Gus Hinestrosa Spain Team coordinator George Kakoulidis Greece Volunteer mentor