As a matter of fact, frailty is unavoidable. Each and every one of us will experience it at some point in later life, but its duration and the impact it will have can vary significantly. Even though, it is a condition that will affect a huge proportion of our older population, the spotlights of scientific research have been neglecting it as the main focus has always been to fight disease and not condition. In FrailSafe we strive towards a larger goal when it comes to investigating frailty. We are not only interested in basic research that will help us detect it, but we decided to walk the extra mile and using advanced data mining and analytics science, to be able to predict transition in frailty, and even to suggest interventions that could delay its progress and sometimes (why not?) reverse it.
Diagnosis of a condition rather than disease is definitely a medical challenge on its own, but in the context of FrailSafe, it is also a major challenge from a computer science point of view. Such a condition is not bound to change significantly over the rather limited duration of the project for most individuals. The progress is slow and the indicators are very hard to identify, especially in an age group of people that is quite turbulent in many aspects that are monitored by our sensors. Picking up the useful information out of tons of massively huge and noisy data that is the key to success. Our focus has been shifted in optimally managing the information that is being collected in a seamless, efficient and error-proof pipeline. Obviously, such techniques and methodologies that are being developed within the context of FrailSafe can be extremely useful and find numerous applications in other data overwhelmed domains, such as genetics, engineering, physics and many other fields where the accumulation of data is happening faster than it can be processed.
A few decades ago, information was so scarce that the challenge was to exploit it as much as possible and even artificially augment it, in an effort to improve statistical viability. Nowadays, the tables have shifted and soon we will realize that not only it is impossible to efficiently handle big data, but also it is soon going to be a challenge to store it for later or future use and analysis. Within FrailSafe, we have adopted our analytics approach to be scalable and ready to tackle the streams of data that our sensors are constantly collecting. We are collecting all information, in an unstructured, no-sql database that is optimized for the management and analysis of big data on the cloud. Then via continuous discussions and close interaction of the various teams within FrailSafe we have opted for various levels of analysis for each given dataset and need. For instance, some data are analyzed online, some other offline, some information is first preprocessed or aggregated and then prepared to be communicated.
All in all, we are now well into the big data era, where efficient management and analysis of information is going to be of uttermost importance and eventually it will shape our future society. Within FrailSafe we are confident, that through our novel and multidisciplinary approach we will be able to efficiently detect transition in frailty and thus provide the knowledge to prolong independence and joyful life in the later years of an ever-ageing population. After all, wasn’t George W. Curtis right when he was preaching that age is a matter of feeling and not of years?