Assistant professor Department of Computer Systems and Control Technical University of Sofia Bulgarian.
Vice-Head Department of Systems and Control, Head Advanced Control Systems Laboratory Technical University of Sofia Bulgarian.
Results of this investigation and some of implemented solutions are discussed hereafter. We focused on the following main problems:
Of course, this list cannot pretend to cover all possible aspect of hospital information systems, solutions, problems and similar. At the end of this material is presented briefly the structure of implemented intelligent hospital information system.
Integration of available medical apparatus and software systems in new system.
Starting with the investigation of how to integrate available equipment in one system we contacted directly vendors’ representatives for detailed technical data. Equipment includes from very modern to 10-15 years old machines. Interfaces to humans and to other computers vary much. Computer interfaces include RS232 / RS485 and Ethernet wired connections, floppy-disk data exchange and on-screen or printed output. Protocols are very different, too. All of this stands as big challenge to the integration team. We contacted technical groups supporting other hospitals and found that even products from one and the same vendor are not fully compatible on interface and protocol levels. As an example can be shown DICOM-based image systems. DICOM has variety of dialects and needs additional processing to make all sources fully compatible with all visualisation systems.
A hospital medical system collects a variety of patient information represented in many digital or hand-written types. All over the world, a huge number of standards of organisation and representation of the Electronic Health Record exist(HER). Unfortunately, many of them depend on local law regulations and even in European Community they are not synchronised. Hospital data (not originally included in HER) are harder to track they include information from many sources – diagnoses, many types of laboratory results, imaging results – X-ray / ultrasound / scanners / Doppler, medications, consultations. These data are very distributed. In a large hospital like the Medical University Sofia original data are held on the laboratories and clinics servers. They have to be exchanged and bound in patients’ records. Such a data base did not exist before the beginning of this project. One of the biggest problems was the fact that data are very changeable, the amount increased all the time and the structure depends on uncontrollable variety of external (mostly human) factors. Today, an information system based on unified generalised structure which covers all types of available is implemented in the MU Sofia medical information system.
One big objective of the presented project was to implement tracking patients when they finish the hospital phase of medication. Experimentation field was clinic of pulmonary diseases. Technological idea was to equip patients with mobile sensors controllable via smart phones and to collect and transmit data using SMS and/or data channels. Currently in the non-military areas of application the most common mobile/wearable biosensors available on the market are 1 or 3 lead ECG sensors, spirometers, pulse and blood pressure meters, thermometers, SPO2 meters and glucometers. All of these sensors were investigated. These are our conclusions about technical aspects:
One important remark is that because of financial limitations, number of obtained sensors is relatively low and the investigation was useful for experimental conclusions and future directions but is not enough representative for statistical purposes.
Developing and implementing hospital information systems including remote data acquisition requires strong data protection. But before this, some other problems important for the patients have to be resolved. First is data validity. Investigating paper-based process we found that in many cases information is wrong or interpreted inadequately or simply lost somewhere in document paths. This is a well-known administrative problem but here it is important because human health depends on it.
Where the biggest problems are:
Designing the solution for integrated hospital information system required us to think how to transfer current archives in digital form. Of course, the problem for archive digitalisation has existed for more than 20 years. The main problem we had to solve was digitalisation of video imaging library. Investigation about possibilities to digitise images of different types envisaged that the on-hand solution to use flat-bed scanners is not applicable. Available scanners for large X-ray pictures start from US$15000. Services for film scanning are from US$2 to 10 depending on picture and requirements. Financial limitations directed us to design low-cost scanners for all kinds of films and pictures. It is presented in. Its price is much lower and resolution is better than film grain. This system is equipped with software for images post-processing. Some results are presented in.
A very important additional problem is the size of digital archive. For medium level Bulgarian hospital (covering 30000-50000 people) only DICOM library is more than 5-6 Tbs/year. Today, disk devices have huge capacities but nevertheless increasing hospital data server with 10TB/year is problematic.
This work is started and partially funded by Bulgarian NSF under D002/113-2008 project.
General structure of hospital information system.
The Intelligent Hospital Information System structure and subsystems are shown on Figure 1. It offers the following features:
The future target of this work is to establish an environment for transformation of the treatment data to knowledge system which will improve the following elements:
This network has to provide doctors access to huge data sources (e.g. pictures, video films, etc.) in the hospital. It is useful for accessing data bigger than 50MB via fixed network resources. Now bandwidth of 1GB is possible and this guarantees on-line diagnostics and data exchange between sources, doctors’ terminals and storage servers.
It controls access over the local network, database consistency and tracks patients at home care.
Fixed clinical network provides connectivity for all machines and apparatus in the hospital from one side and servers and personnel’s terminals from the other side. Additionally it guarantees better redundancy and offers possibility to control internal clinical networks loading.
This network enables access for the medical personnel to data servers. Additionally, sensors and apparatus generating small amount of data can be mobile on hospital territory. This guarantees unbroken control when patients can carry their vital data acquisition sensors or simply to move patients over clinics without loss of connectivity.
It controls all administrative processes, hosts all records about manipulations (total and associated to every patient), personnel and patients archive, etc. Based on this, doctors can do different analyses and increase quality of medication.
This is SAN network offering: on-line information for all patients; fast access to disease history and hospital archive.
This is the core for future advances. Data mining services will be positioned there.
A hospital grid with specialised servers will be built on this basis.
Ivan Evgeniev Ivanov is also a member of IEEE and Bulgarian Union of Automatics and Informatics. He has led a number of projects oriented to (embedded) computer control systems, heterogeneous distributed systems.
Vesselin Gueorguiev is also a member of Bulgarian Union of Automatics and Informatics. He has led a number of projects oriented to real-time computer graphics, program code analysis and distributed systems.