Afflo is worn anywhere on the chest, attached using an adhesive disk.
Walkthrough of the Afflo mobile application.
Walkthrough of the Afflo mobile application.
Afflo is made up of three components: the wearable, the pod and the app.
All components of Afflo have been fully specified and validated through either testing or research.
Spectrograms used for signal analysis. The machine learning algorithm gave 82 % accuracy.
Initial sketches looking at the best way to assemble Afflo to minimise size.
Was es macht
Asthma is a condition making breathing difficult in response to external triggers. Matching triggers to individuals’ symptoms is difficult, currently done by trial and error. Afflo reduces this uncertainty by using AI to predict each user’s personal triggers.
As an asthma sufferer, I am acutely aware how diagnosis can be guesswork with costly and dangerous consequences. Asthma is a widespread disease, with 235 million people suffering from it globally. This huge demographic is currently served by outdated asthma management tools. Every year, asthma costs the NHS £1 Billion. After interviewing patients at Guys hospital and conducting an online survey, I found that the major gap in asthma management today is trigger identification. These problems could be solved through Afflo. By collecting data and sending it to the doctor, diagnosis could be made remotely with the help of AI, minimizing costs.
So funktioniert es
Afflo collects respiratory audio signals through a specialised microphone. This can be placed anywhere on the chest and is adhered daily using adhesive disks. The microphone is mounted within an acoustic chamber, blocking out external noise. Environmental information is collected through two methods; a sensor bundle for localised information, worn on a beltloop or backpack and online sources (using APIs) for macro level data. The wearable and pod pair with the patient’s phone via Bluetooth, periodically transferring data. To predict each patient’s unique trigger pattern, a machine learning algorithm analyses the two streams of data in the cloud. The results are presented back to the user through the Afflo Mobile Application in an easy to read format, allowing them to make lifestyle decisions, minimising future symptoms. Over an extended time period, this data can be reviewed by medical professionals remotely, to cost effectively refine treatment plans.
Ideation ran in parallel with research, investigating where potential solutions could fill market gaps and make use of existing, unapplied research. The design process ran in three streams allowing each part to influence one another; the physical design, the technology and the interface. Firstly, signal processing and acoustic testing specified the correct microphone, the most important component in terms of accuracy and functionality. A CAD model iteratively improved the design over time, increasing its resolution and manufacturability while developing a bill of materials. A look-alike model and exploded model were made to validate and demonstrate the assembly. A functional prototype of the UI/UX of Afflo was created using Sketch, Adobe XD and After Effects. Three iterations improved the design over time; hand drawn wireframes, unstyled screens and a working prototype. To test the opportunity for machine learning, a prototype was built using Python, forming the first step towards a diagnostic tool. This neural network was able to differentiate between a cough and speech, identifying when this key respiratory event had occurred with 82% accuracy. These four prototypes worked in combination to validate the viability, feasibility and desirability of Afflo today.
Wo ist der Unterschied?
Critically, Afflo is the only existing project looking to identify asthmatic triggers, rather than simply monitor or predict the onset of symptoms. After speaking with numerous patients and doctors, trigger identification is the major pain point in asthma management today. Patients are able to identify when they are experiencing symptoms but struggle to pinpoint what in their environment is causing it. Afflo differs from Respia and other smart inhalers by tackling the problem at its cause. These other concepts principally identify the onset of asthma attacks in children rather than looking for the root cause of asthma for each and every patient. Afflo has been designed for the patient. By involving all stakeholders throughout the design process, I have ensured that Afflo satisfies patients, nurses and parents. Afflo is not a purely research or design-based project. It creates a functional design with strong user experience.
I hope for Afflo to reach market and improve quality of life for asthma sufferers whilst making financial savings in health budgets. The next major step to reach this goal is to obtain funding. This capital would facilitate further development of the technology, continued acoustic testing and expansion of the neural network. The physical design of Afflo would be iterated, looking to reduce its size, making it less obtrusive. These steps would eventually allow Afflo to reach clinical trials, a critical stage before hitting the market.
Afflo is sponsored by the Institution of Engineering Designers, who provided funding to support the project. Afflo has been selected as one of four final year solo projects at the Dyson School of Design Engineering to progress to the next round of selection for the Global Graduate Show in Dubai.