Automatic Machine Learning and User Modelling of Intramural Communication in hospitals
A novel system for communication between hospital doctors, CallMeSmart-doctor (CMS-Dr), has been developed and tested in a real setting. However, we have realized that this smart solution dedicated doctor-doctor communication could be even smarter. It could represent a totally new infrastructure for many actors and sectors, and be used in a much broader setting. Therefore, we want to renew and strengthening the ICT-research field through establishing a generic CMS solution: A fundamental new ICT infrastructure for all health care actors (not only doctors), and other actors with complex, temporarily and time critical communication patterns. Our experience so far is that role and responsibilities, and therefore the communication patterns change from time to time. Therefore, we want to renew the ICT-research filed by building in intelligence in the CMS generic solution, i.e., make the solution smarter through machine learning. The CMS-generic shall become Ubiquitous and Self-Learning (CMS USL). CMS USL will as such represent a bold and innovative research projects that will provide a renewal and revitalization of ICT and e-health research. The technological CMS-USL solution will represents the internationally forefront of ICT-research through new combination of advanced data based wireless communication that maintains context awareness, in addition to ubiquitous and self-learning machine mechanisms.
The CMS-Dr prototype focuses on context sensitive interfaces, middleware, and new interaction forms for mobile devices that support multi-modal communication in hospitals. These devices support media such as voice services, text-messaging and paging services, in an efficient and non-interruptive manner, as well as enable support for individual and role-based contact on a single device. That is, the user only need to carry one device for both personal and role based communication, which enables other users to, for example, contact someone assigned as “on-call” duties at a specific department, even if they do not know who that person is. At the same time it aims to balance between availability and interruptions, while it enables acute calls and alarms be forced through. Currently, by our knowledge, similar devices are not generally available for internal communication systems in hospitals.
The prototype senses the context automatically from different sensors, calendar information, work schedule, etc., to change the physicians’ availability and the phones profile, according to the collected context information. At the same time, the caller is given feedback about the physicians’ availability, and thereby it is possible for the caller to force through an emergency call, or forward the call to another physician at the same level, that is available. The system is based on ideas from existing research on interruptions, in combination with our ideas. A first version of the prototype is ready and has been tested in lab-settings with physicians as test users. The tests were performed as scenarios observed from real situations. The feedback was positive and has been used as input for improvement and further development of the prototype. CMS-Dr prototype is ready for testing in clinical settings, and a pilot has just been started at the Oncology Department at the University Hospital of North Norway (UNN). The solution has retrieved overwhelming enthusiasm and positive response from the test users so far.
The project focuses on fundamental aspects of intramural communication in hospitals. The goal is to develop novel models and techniques for machine learning and user modelling in order to make intramural communication more effective. The technological solution will represent the internationally forefront of clinical informatics research in automatic machine learning and user modelling through new combination of advanced wireless communication that maintains context awareness, in addition to ubiquitous and self-learning machine mechanisms.