Diarienummer 2013-03116
Koordinator Högskolan i Skövde - Institutionen för kommunikation och information
Bidrag från Vinnova 294 000 kronor
Projektets löptid juli 2013 - september 2016
Status Avslutat

Syfte och mål

The HELICOPTER proposal aim at exploiting ambient-assisted living techniques to provide older adults (end users) and their informal caregivers with support, motivation and guidance in pursuing a healthy and safe lifestyle. The goals are (i) to develop an object interface, (ii) to develop tools and an infrastructure, (iii) to develop an adaptive and evolvable analysis for risk for health problem and (iv) to test the whole system in a pilot study as well as (vi) to develop a business model for deployment.

Resultat och förväntade effekter

The HELICOPTER system prototype was developed, deployed and tested in a pilot study with reasonable, but perhaps an overly conservative configuration that may favor false negatives. That is, if there is evidence for a risk of a diagnosis, then the system require really strong anomalies. All parts planned were part of the prototype The pilot study provided a lot of learning experience, but no conclusive evidence of that it works. The simulation studies are inconclusive and require more time to see if it possible to achieve the expected ratio between true/false positives/negatives.

Upplägg och genomförande

The project followed a traditional waterfall model. Initial care was taken to ensure that we had consensus on what we wanted to achieve in sufficient detail by performing conceptual modeling, web-based enquiries concerning definition and prioritization as well as discussions. After this, the different packages worked independently, where there was two meetings per year to meet and discuss. Partial deliveries of technology was performed before reports were completed, e.g, the anomaly detection was delivered after a small simulation test, relying on earlier published results.

Externa länkar

Main project page of the HELICOPTER project. Java-lib for anomalies. Base: Pokrajac, D, et. al, `Incremental Local Outlier Detection for Data Streams´, CIDM 2007 Tool for Bayesian Belief Network Analysis API for continuous time simulation for qualitative features tests (e.g., disruptive behaviors).

Texten på denna sida har projektgruppen själv formulerat och innehållet är ej granskat av våra redaktörer.