Objectives of the User Tests

The UCD process described in previous deliverables has come up with an application concept that overcomes the above problems. The aim of the testing has been to verify the benefit of this application and derive the requirements for the final demonstrator.

Components tested
The first prototype consisted of three main subsystems; the doctor system, the bedside system and the nurse system.

Doctor Subsystem
The doctor subsystem consists of an unobtrusive wearable computer, an interaction wristband, and a headset for audio in and output.

1. Wearable Computer: The wearable system is the doctor’s personal information gateway. It controls the input output devices, controls all sensors and performs most context recognition tasks and provides network connectivity (which ensures connectivity with the nurse, the bedside system and all hospital databases). We chose the QBIC wearable computer, which features a 400 MHz Xscale processor, Bluetooth and USB interfaces, and a Debian Linux OS, all integrated in a wearable belt. A USB WiFi adapter is used for communication with the other subsystems.
2. Interaction Wristband: The interaction wristband is used for patient identification using RFID reader and it provides a simple gesture interface. We used a small RFID-reader module attached to a USB-powered ’tmote sky’ sensor node. For gesture recognition we used the MT9 motion tracker sensor from XSense. It features accelerometers, gyroscopes, and magnetic field sensors, all on three axes and sampled with 100 Hz. The sensor was attached to a Bluetooth module including a battery pack.
3. Wireless Headset: For speech recording a wireless headset was used. Where a headset is not possible (because of the need for a stethoscope) a chest worn microphone will be used.

The doctor subsystem acts as the main gateway between sensors and the other sub-systems. It gathers and processes the information from sensors, and sends suitable commands to the nurse and bedside subsystem.
The gesture recognition and sensor processing software on the doctor subsystem was implemented using Context Recognition Network (CRN) Toolbox. The CRN Toolbox basically consists of a set of parametric algorithms that can be applied on data streams. An application is defined by selecting the algorithms to be applied, defining the parameters of each algorithm, and connecting the input and output streams to form the path of the data streams.

Nurse Subsystem
The nurse subsystem consists of an application server and a PDA as a thin-client frontend. It allows the nurse to enter and retrieve parts of the information relevant for her tasks operate those feature of the system, which are too attention consuming and require too much hands use for the doctor to perform. The nurse subsystem runs an application server which receives messages from the doctor subsystem and sends according forms to the PDA of the nurse which, in turn, can review the form, enter additional information and send it back to the application server. Also, if a dictation of the doctor is recorded, the application server can forward the audio file to the nurse’s PDA. The nurse subsystem is based on MS Windows IIS software.

Bedside Display
The bedside subsystem runs a Java application, which is simulating the document browser of the hospital information system. The browser is able to display different types of content. Examples of the manageable content so far are text documents like examination and laboratory reports, audio recordings of examination questions on the PDA, pictures of wounds or x-ray pictures. An icon on the top right corner displays the current activation state of the gesture recognition. Our document browser can be controlled with four basic commands (up, down, open, close). Further commands set the enabled/disabled state of gesture control and the display of the browser itself.