Orchestrating Internet of Things and Machine Learning for Early Risk Detection to Ensure Inpatients Safety

Quality and patient safety are two areas of pivotal importance for clinical personnel and hospital management. and in worst cases to loss of lives, bring huge losses and long-lasting consequences for patients, family members, involved hospital personnel, as well as society. To address such a challenge and detect potential risks at an early phase, it is imperative to develop a means or system that is capable of improving inpatient safety, by monitoring and analyzing agitated and unusual behaviors and generating alerts to responsible personnel on an automatic, continuous, and 24/7 basis. Motivated by the recent advancement in IoT and ML technologies, this project aims to develop a viable solution for behavior surveillance and analysis of inpatients in mental and somatic healthcare, by integrating real-time data collection based on non-intrusive IoT devices and detection of unusual behavioral patterns based on ML algorithms for the purpose of behavior analysis and risk prevention. This innovation project is a joint effort among Sørlandet Sykehus (SSHF), University of Agder (UiA), and Egde Consulting AS in collaboration with our international partners from National Chiao Tung University (Taiwan). We target at developing a deployable solution with high-level decision support and early stage warning capabilities available for on-duty personnel and involved inpatients. We have been testing several different radar sensors in our lab at i4Helse in Grimstad since spring time. We have compared the results from the collected data with data from wearable sensors and a sensor placed under the matress. We are currently verifying the data to make sure that we can develop algorithms based on data from radar sensors. Testing and verifying have been more time consuming than predicted. The prodject has been presented to two ministers of the Norwegian governement and Barne-, ungdoms- og familiedirektoratet (BUFDIR), and at two sessions at Arendalsuka, Egde Consulting AS being responsible for one of them. We are preparing for data collection at DPS Østre Agder (Bjorbekk), where we are allowed to collect data from wearables and sensor in matress, in addition to radar sensors. We are applying REK to use radar sensor for monitoring pasients without consent. This is crucial for getting correct data to develop algorithms for the purpose of behavior analysis and risk prevention. We have not yet received an answer. We are in contact with the project TryggNatt at OUS, which is in the prosess of starting sleep monitoring.

Quality and patient safety are two areas of pivotal importance for clinical personnel and hospital management. Incidents that lead to patient harm, and in worst cases to loss of lives, bring huge losses and long-lasting consequences for patients, family members, involved hospital personnel, as well as society. To address such a challenge and detect potential risks at an early phase, it is imperative to develop a means or system that is capable of improving inpatient safety, by monitoring and analyzing agitated and unusual behaviors as well as generating alerts to responsible personnel on an automatic, continuous, and 24/7 basis. Motivated by the recent advancement in IoT and ML technologies, the StaySafe project makes a joint effort to develop a viable solution for behavior surveillance and analysis of inpatients in mental and somatic healthcare, by integrating data collection based on non-intrusive IoT devices and detection of unusual behavioral patterns based on ML algorithms for the purpose of behavior analysis and risk prevention. We believe that such a solution will be able to provide high-level decision support based on real-time data collection and analysis, thus helping prevent agitated patient behavior through early stage warning to on-duty personnel as well as involved inpatients This innovation project is a joint effort among SSHF, UiA, and Egde in collaborations with our international partner NCTU/TW and end users. The total budget for this project is ~13.3 MNOK including 50% NFR contribution and 50% partners' own contributions respectively. The scope of this project is in full conformity with this HELSEVEL call for innovation projects for the public sector, addressing two cross-cutting themes and research areas defined in the HELSEVEL program, i.e., Technology and digitalisation and Service innovation and implementation. The technology readiness level of this project is 4 or higher, targeting at a deployable solution within the next four years.