Background More than a third of people over the age of 65 fall every year in the UK and those who fall once are two or three times more likely to fall again. Falls represent over half of hospital admissions for accidental injury, particularly hip fracture. Half of those with hip fracture never regain their former level of function and one in five die within three months The broad category of gait problems and weakness is the second commonest specific precipitating cause for falls. Our qualitative research indicates that people do not consider balance health to be an actionable component of their over all health. This is because we do not have the vocabularies or tools to objectively define it on an every day basis. From the clinical perspective, falls clinicians, physiotherapists and occupational therapists do not have a quick, objective tool to quantify their patient’s postural sway, instead, they are dependent on long form questionnaires like the Berg Balance Test. Aims This project aims to make balance an actionable component of an individuals health, allowing a long term intervention via a monitoring and training of balance health much before a fall happens. Towards this end, we designed an iOS application which measures static postural sway and then directs individuals to exercises for improvement. Postural sway is defined as the phenomenon of constant displacement and correction of the position of the center of gravity within the base of support. It is typically measured on Force Platforms and has been found to be a predictor of falls. From the outset, we envisioned our application with therapeutic cum diagnosis functionality which was suitable for usage by an audience with diverse abilities. Methods We tested the engineering prototype of our application 340 times with 26 individuals between the ages of 25 and 86. 6 individuals were asked to use the application 3 times a day over a period of 10 days, 20 individuals at two locations, London (10) and Andover (10) , were asked to use the application once a day. The purpose of the testing was to determine whether the inter-test variability of results for one individual was less than the intra test variability between individuals. On the initial testing day, we took 10 second long exposure pictures of every participant. The purpose of the testing was to check whether the application was sensitive to diverse abilities and to be able to identify parameters which define a ‘Balance Footprint.’ Results We found our application could identify good or bad balance using Root Mean Square of the oscillations. The long exposure photographs could substantiate this. Figure 1 shows graphs and pictures of a participant with high postural sway or “Bad Balance” alongside a participant who has low postural sway or “Good Balance.” Further analysis of 400,800 data points has lead us to identify 4 factors which comprise the unique ‘Balance Footprint’ of an individual. 1. SWAY SCORE How much do you sway? Calculated by the Root Mean Square of the readings. Sway Score or the amplitude of a persons balance. The higher the sway score, the worst the balance. 2. SYMMETRY How does your left side compare to your right hand side ? Calculated by comparison of strength between the left and the right sides of the body. Fallers tend to be asymmetrical. Figure 2 shows a participant with significant asymmetry. 3. SWAY ENVELOPE How is the movement distributed along your anterior posterior and media lateral axis? Visualised by plotting the distribution of movement along the media lateral and anterior posterior axis. Figure 3 shows the sway envelope of a participant ,the left side is equally balanced, the right leg has significant anterior posterior movement. 4. CONSISTENCY How much variation is there in your day to day scores? We found some participants were consistent with their balance scores, whereas others had scores which varied on a day to day basis. Hence we have identified 2 personas. High variance and low variance (Figure 4) Applying results to design rationale During our tests, we found a wide range of abilities between participants. Whilst there is a correlation between age and balance (Figure 5), there were outliers, certain participants had poor balance in spite of being in the younger cohort, and some older participants tested very well for their age. Our quantitative research indicates that this difference is due to the difference in activity histories over the life time of an individual. A 25 year old participant commented “Your app has shown me how awful my balance is! Not looking forward to being older and we have osteoporosis in my family! Lots of broken bones for me! O dear!!” This lead us to think that what seems like a straightforward act (standing on one leg for a period of 15 seconds ), can actually be demotivating for individuals. In order to keep the participants engaged, we needed to incorporate easier stances, which while challenging, were not a blockade to improvement. This would allow an individual to start at an ability appropriate level and build up to better balance in tiny increments. Hence we have created a training programme (Figure 6) which can computationally determine the ability of the individual during the on boarding process. Once the current postural sway id determined, the individual is assigned to one of 3 Stances ( Semi Tandem for Beginners, Tandem for Intermediate, Uni-pedal Standing for Advanced). Each stance has 4 levels of varying lengths. ( typically 15s, 30s, 45s, 60s ) An individual could be assigned to start training at Stance 1, Level 1 ( i.e. Semi tandem for 15 sec ) and gradually build up-to 60 sec over the course of 36 sessions, at the end of which the ‘wobble reduction’ would be used a measure to determine whether they were ready to pass on to the next level. We have added gamification elements in the form of giving meaningful tips, avoiding negative feedback, simplifying the interface by removing numbers and of-course, medals and celebration screens (Figure 7).